Good reminder! I was just thinking about adding an ESP32 to my car but was worried that the range wouldn’t be long enough. I will try an external antenna.
For most projects the size and extra expense in any mass manufacture puts ufi in the hard basket, I think the DA offers those with manufacture runs a cheaper alternative in a smaller self contained package.
On ESPEasy, I have the option to have multiple runs while scanning for a network. This often gives a lot more access points as not all access points react immediately on an active scan. So you may want to perform several runs when scanning to get a good idea of the performance difference. Also you might want to turn the WiFi off and on again (and wait a few-100 msec) to start scanning on the other antenna. When turning on the WiFi, it performs a RF calibration, thus maybe this can also improve the antenna performance if the antenna is selected before enabling WiFi. I will also try to get one of those "T-shirt" modules, to test with. Have not had one of those yet. On my own board designs, I almost always use the ones with an ufl connector, so the antenna can be positioned for best performance. Even those "sticker" antennas perform way better than the PCB ones, at least that's what I've seen.
I agree with your statements. I averaged a few scans because I also saw that they differed from scan to scan. And I would expect that Espressif does all the needed stuff that the switching gets the best result (like calibration, etc.) Maybe they already did and I do not know because they kept it as a "secret". I am glad if you do your own tests. And I agree that external antennas are a good choice. But they are clumsy and add to the BOM...
Dual antennas are very helpful where there is multi path (destructive) interference. Perhaps a more enlightening test would be to orientate a metal plate quarter of a wavelength away from one antenna such that it is in a null (ie a sharp drop in RSSI due to the destructive interference). Then switch to the other antenna, where the signal should be outside the null, due to being spatially separated. Adding the RF switch will increase insertion loss and so degrade the noise figure. Therefore the benefits of multi path tolerance have to balanced against slightly degraded sensitivity.
I agree with your analysis. I for sure would do those tests when designing such an antenna. For a user like me, it seems to be more important to just see how big the effect is in reality. In addition, the effects will be different in every situation (13cm wavelength!). So it is more a general feeling that it could help. But only if the switching works properly ;-) The insertion loss of the RTC6603SP chip is 0.4dB @ 2.5GHz, BTW.
@@AndreasSpiess Interestingly, there exists "space time block coding" aka STBC. Here, the access point uses two antennas at the same time in a way that a single receiver antenna can make use of. If the interference exists for one of the transmitting antennas, the other antenna may come in loud and clear. The ESP6266 supports STBC, it seems. (The other way around it is trivial, the AP can use whichever antenna works best to receive the single signal from the ESP).
I am very happy to see someone test and talk about this product. I'm actually working on a street light project for my neighbourhood where our streets are private, we have ~40 streets poles each distant by
That pointer with the little, white gloved, hand on it always brings a smile. Mr. Spiess, your content is always exceptional, and your integrity is beyond reproach. You are a true treasure of RUclips.
The shown WIFI routers with multiple antennas are completely different from this ESP32 module. All of the antennas are active at the same time. There is no switch that selects just one antenna. Multiple active antennas are used for beam forming like a phased array or to receive multiple signals simultaneous (MIMO). If the ESP32 could use both antennas as a phased array to do beam forming, it would get a better signal than any one of the antennas can receive by itself.
Thank you for your info. I know about beamforming but thought that it needs an array of precisely spaced antennas. The antennas of the routers I showed have moveable antennas. That is why I thought they could not be used for this purpose.
@@AndreasSpiess I am no expert, but I believe wifi does not actually know what the world around the router is like. It just tries different combinations of phase shifting between the available antennas until it gets the best rx signal. Then tx can use the same parameters. Another interesting fact is that the antennas on many wifi routers are somewhat fake. They are made to look impressive. But actually they are just empty plastic shells with the actual antenna inside being much smaller. The wavelength of a 5 GHz antenna would only be 6 cm and the antenna probably a half wave in length. Yet the router has huge antennas.
I think that multiple antennas in routers implement MIMO operation - multiple inputs and multiple outputs, they work together at the same time. Here the antennas work one at a time. I assume dual antennas were included to improve directivity. Auto mode would allow connecting to the known AP using an antenna with the better RSSI. This can easily be tested by logging the RSSI for a nearby AP in ant0, ant1 and auto mode. The latter should return the maximum value out of the two.
Yes, I wanted to remark this as well! The access points that have multiple antennas actually have multiple receivers (and transmitters), not an antenna selection switch.
In routers it's more complicated. They indeed use multiple antennas for MIMO but also in this same time do what this ESP module aka try to find best antena for every connected device. In most advanced ones also beamforming signal.
@@AndreasSpiess I learn a lot from your videos and, on rare occasions when I can add something (I think) worthwhile, I do with all my respect to the creator
For a test I would propose, try to connect to a WiFi network, and then compare the two-antenna to the one-antenna version. I imagine the auto-mode to be effective when the module knows what network you want, and then switches between the antennas to get the best signal for the desired network. In such case, a test might include a setup you can carry around, seeing how it switches between antennas when the relative direction of the signal changes.
Yes; it seems we don't know what AUTO is optimizing for. If a network is selected, it may choose the best antenna for that. but if not, like in this test, it could as well try to get as many as possible as try to optimize the strongest (or pick up one arbitrarily, a.k.a. simply don't know what to do... :-)
Maybe this would lead to different results (as I mentioned in my video, I do not know). However, I do not understand why auto should not work for scanning. A module first has to scan the frequencies to find an access point. If auto does not work it probably would not find the proper access point. At least this was my reasoning.
This is very interesting. Now we just need to wait for other companies to make pre-soldered development boards so we can just buy one from many sellers online with everything ready to go upon opening the package. Thank you for showing us this new strange device variant.
Ivan Grokhotkov is the chap at Espressif who knows most about the software. You can get his attention by leaving a clear description of the problem on their forum, but they like you to check to see if anyone else has mentioned it before, and if not Ivan will usually find the trouble out and fix it. He's got a sidekick who comes under the name of Me-no-Dev and he can usually deal with most things.
again a very interesting video. And the results are quite a surprise, I wonder if Espressif will elaborate on it someday. Looks promising though! thanks for the update!
Dual antennae usually used for diversity purposes. Dual rx are useful for calculating multipath signals and diversity gain. All cellular communications use atleast two rx paths for receive diversity. Two TX is mostly used in MIMO feature.the antennas kept at 90 degree, the purpose may be +45 & -45 cross polarization.
Thank you for the additional info. I did not know that cellular phones use space diversity. I only knew it from HF systems below 30MHz. And I agree with MIMO. But I think this is another story...
If you turn the complete circuit upside down the position of the antenna's 1 and 2 are swapped. If assumed the characteristics above and below the antenna (the z axis) are the same this would give interesting additional information.
Thank you so much, it was a great note to mount the esp 32 antenna outside our PCBs. I wish you can do a video that comapres between bare ESP32 modules out side thier boards
One option that you may want to test is to orient the board so that one of the antennas is oriented horizontally, and one vertically, and see if that helps with lower power signals. I know that RF that hits certain typologies can end up switching from horizontal to vertical polarization, and having a platform that allows you to switch between the two for some environments may give you far better coverage. That said, WiFi wasn't really designed for 'weak signal' work, so there are likely to be other issues that prevent a stable connection if polarization gives you a significantly different signal level. And that may also be part of what the results you see from the Auto test are showing, that suggests that of the 56 signals observe on one antenna, and the 30 some odd on the other, only 20 overall were within whatever level the system accepts across both antennas. But I don't know, I don't work for Espresif.
In multi-path conditions, sometimes an irreducible bit error rate (BER) is present. But diversity combining wouldn’t switch between antennas but use them both and some kind of quality measure to combine them.
On HF we sometimes use "space diversity" where we do a similar thing with the two antennas. But we also do not connect them. We select the stronger signal after the receiver. This can create very good results.
Hi, could you tell me how to make ESPHome send three numbers (channels) to the TV, please? I've been looking for days on this and haven't found anything. Thanks.
Nice, it's quite frustrating when the documentation doesn't help, I spent days trying to resolve issues with a Gyro chips interrupt parameters to find that the manual had an error.
Ah, wrong datasheets! I ran into some of those during my time in Corporateland. One especially egregious case: I was designing a module that needed some sort of little MCU for monitoring, communicating via I2C. I was leaning toward an AVR device, but Management decreed that I should use a particular 8051-based MCU because an engineer in another division had already used it successfully in the same role, and had firmware already developed and ready to adapt. After reviewing the datasheet, I went along with this. Well, my module having been designed and built, I made the pilgrimage to the other division in search of firmware to adapt, and found that the other module hadn't been built nor the firmware tested, so my module became the test platform. And, surprise! The documentation for the MCU's I2C interface was wrong; in reality, a key feature needed to support... I guess we're not supposed to use the old terms?... anyway, it could operate fine in "mad scientist" mode, but in "hunchback" mode there was something fundamentally broken. And, of course, there already being one "mad scientist" on the bus, these modules were all supposed to operate as bus "hunchbacks". I don't recall what the resolution was - whether we came up with a firmware hack to make it work well enough for the purpose, switched to a different MCU, settled for a dumb module, or just abandoned the whole project before getting module communication sorted out. (This wasn't even a brand-new MCU, so the re really ought to have been at least an erratum for the I2C issue, but I guess I2C on MCUs is almost entirely used for talking to EEPROMs, temperature sensors, and the like, so it's entirely possible that I was the first one to try to use it that way.)
I wonder, on balance, if a single external 1/4 wave vertical antenna is going to be better or worse than a (well optimised) switching between two pcb antennas? I have got into the habit of using an external antenna for esp32's I put outside, and this would be easier and smaller
An external antenna is for sure a good possibility and usually has a better performance. Often, people do not want one because of the added size or cost.
Me too, I often are confronted with the limited coverage of the standard antenna. I started adding AP's in my network, but still, an external antenna performs much better. I can live with the additional cost and form factor.
I use "3d antenna" like the one on TTGO module. I made the same test as Andreas Spiess and saw that the 3d antenna perform really better than the PCB one. So for me it is a good quality / price / space ratio
You can solder sma connectors to antenna traces and connect your qo100 amplifer to one of them and set it to tx only and use other antenna for rx. It should increase range a lot.
Initially, I thought perhaps the dual antenna setup might be for diversity signal RX. Is there something in the docs that points in that direction? Can diversity RX be done in software? I used to use some cellular modems that had that feature if you connected two antennas to them that could improve reception in noisy environments or improve weak signals if the device was far from the nearest cell tower.
I would call it "slow space diversity" because it has 2 antennas but only one receiver. But it switches to the better antenna from time to time (at least this is what I assume, but not know). This should improve reception, particularly for fixed setups.
This is exactly the sort of thing that completely frustrates me about products from China, guessing about how they work due to the lack of pertinent information. It costs essentially nothing to post on the Web ALL of a product's useful info. There are numerous such products that I simply don't purchase due to their lack of information, and I am certain there are many, many other prospective purchasers just like me who simply purchase something else or nothing at all for that very reason. The people who purchase these products are NOT typical 'plug and play' consumers who want products that they don't need to think about. We design things with these products, for ourselves, for our friends, and for businesses. We NEED technical information on these products, ALL of it. But we are kept in the dark. It's as if the vendors don't really want to sell their products. The primary reason, (other than your dry sense of humor,) that I watch your channel is because you are much better and more dedicated at gaining that sort of info than I am, ( as I said, I just give up and buy something else, or buy nothing at all,) making this my 'go to' source for specific technical information on this sort of product. Even if your humor is dry, you are a fountain of information. So I always sit in the first row. I sincerely hope your channel gets 10 million subscribers so that you will have the leverage necessary to pry from the suppliers or manufacturers themselves ALL of the essential technical information about their products. We NEED it. Thank you for all of the info that you DO drag out with your detailed research and your experiments. Oh, and thanks again for your dry sense of humor.
Thank you for your kind words. I know that many Espressif engineers watch my videos. We will see if something will happen. But I have to confess that I am also not good at documenting things. I know of many other engineers who are the same. We want to build, not write ;-)
@@AndreasSpiess Oh yea, do you now have/carry a spare ECM power relay for your automobile? Maybe some tools and a contact file would be enough. Hoping for the best...
Shape reminds me of a B2 bomber. Seems like a lot of effort that can be solved by a UFL antenna. That 20 seems to be a software limit at face value. Think it is picking the best 20 access points based on RSSI. I know there is a soft limit to the amount of access points the ESP can store in memory on a single scan.
it seems you could make 2 wifi devices with this.. make a nice repeater too...input on one, output on another... the auto switching maybe fast (or slow!)..so you miss alot of ssid's..when its set to one its constantly on and doesnt miss any frames... ...this with external antennas would be interesting... (you also missed the chance to compare against an external antenna in this vid!) good stuff Andreas.. i didnt even know this existed till now! ..thanks!
Unfortunately, the ESP32 only has one radio. And repeaters in such a small space are nearly impossible because of cross-talk from the transmitter to the receiver when they work at the same time.
Thanks for a great video of this new module. I have a general question, How do you distinguish from a WiFi and LoRa antenna? I bought several antennas from second hand but I don’t know how to identify them, nothing written on them. Thanks for any help!
This has some potential but it needs more tests in the field. Next a ESP32 with a build in Yagi antenna. ... wait I see some potential in that fitted in a nice weatherproof enclosure. Or maybe we could use one with a uFL connector soldered on a pcb with a microstrip line collinear antenna.
Andreas, it looks like your standard ESP32 had the antenna over the breadboard, while with the WROOM-DA, it was sticking out over the edge of the breadboard. I wonder if this could adversely influence the coverage of the ESP32? After all, you did suggest that any PCB design should not have the antenna over the board... it should stick out to avoid RF shielding. Am I right?
Thinking aloud: Would it be possible to de-lid the module and have a look at the RF switch? Watching W2AEW's channel, it seems like it shouldn't be a complicated circuit. Won't tell us much about the switching algorithm, but it's a start.
So, when will there be a 3 antenna version with one for the Z axis? Would that even be useful? Maybe for very mobile devices (drones?) since it could help with kinda directional antennas since the right antenna would only send in roughly the right directions. Personally I like external antennas more than PCB antennas when I can't use POE for a Project, at least I think it's easier to place them outside a case.
There would be no additional benefit in using 3 axis antennas. A rod-like or omnidirectional antenna is weakest in performance when it points directly towards the communication peer. So when you have a second one at 90° angle (transverse to the peer direction), that one is guaranteed to have the good orientation. Having a third antenna would have the same 90° to the weak antenna, but there is no additional gain in sensitivity / range / receive strength.
Adding an antenna with a different polarization would make sense because the effect of polarization is quite high. But then the device would become much bigger and probably harder to manufacture. Maybe they will create one with two connectors. Then everybody could create the setup he needs (also with external antennas).
Nice video andreas. But is it possible that your test algo is not correct for the test of the auto function? Wlan scanning is a good to have feature. I think you must connect to a wlan and the algol of the chip checks on wlan idle which ant have the best receiving quality (or sending?) and switch between the two ant? Can you connect to the pin how controls the rf switch?
AS I mentioned, I am not sure if I chose the right test procedure. However I do not know why they would not use a proper antenna selection for the scanning. If you do not find the access point it is probably hard to connect to it... I also do not see how we could use the TX signal because our chip does not know how good it is received at the opposite station. I hope we will get some insights from the Espressif engineeers...
@@AndreasSpiess Its possible that the engineeers think you will "hard code" the ssid and pw. I does not matter how you send the signal, because the receiving system also have multiple antennas most of the time. Its only how good the receiving signal of you choice is. But look in the capsman of you wlan system. You will see also both ignal qualitiy in the list.
I solder one pin with my ordinary soldering iron and then a second one at the opposite side. When everything fits I solder the rest and check with the ohm meter if I created any bridges.
When several years ago i did some experiments with Wifi signals around my house i noticed a lot of signals from cars and other moving things. (Mobile phones) Also it seems some phones constantly change there mac address to give privacy e.g. in shopping centers where they do/did tracking. So the timeframe is also very important when sampeling.
You are right with Mobile phones. They change the MAC address often. In my case, I do not see it as a problem as not a lot of moving Smartphones were present (names usually are different, and I know a few AP names from past tests).
Usually I agree and I am pleased with your torough reviews. I may be mistaken, but I think you are missing the point. Two antennas are not meant to discover more networks, but to split incoming and outgoing traffic on separate channels to increase speed and throughput. This is the main advantage ! Second my friend, W is not = to V in English ! LOL ! It's not V-ROOM it's WROOM like the roar of a car ! Have a great day ! Auf Wiedersehen !
If the antenna-select GPIOs are accessible as described, then they could be monitored during operation to see if the auto mode is even switching between antennas
A good idea. I should have thought about it before I soldered it to the PCB :-( But I hope we will get the info and do not have to do reverse engineering.
Interesting to find out that the two antennas cannot be used simultaneously. I was looking for a module where I can use WiFi and BT simultaneously. So this module won't work for my purpose. Thanks for testing as it helped me.
Afaik there's only one amplifier and associated hardware for generating the signal. This is just two antennas, not two separate radios (which I agree could be good but rip power consumption)
From my own experience the esp 32 has not enough space for both libraries. Maybe try to compile both libraries in the the Compiler just to test if it works bevor you buy one for that purpose
@@Tom-ku8bu is that for Arduino programming I assume? I've not tried that but I use the IDF at home and for work, no problems at all with WiFi and Bluetooth libraries concurrently. Still tonnes of space for other code too. I even made a wi-fi to Bluetooth speaker bridge using an esp and it handled that no problem (audio from VoIP to the esp, which pumped it out over Bluetooth a2dp to a Bluetooth speaker)
@Andreas there is a use for having a send and a receiving: replace the receiving Antenna with a high gain directed one... new challenge: long range WLAN with ESP Modules Edit: The challenge is meant to be w/o breaking the emision laws!
@@AndreasSpiess is it possible to solder a miniRCA on the module? (e. g. like on Raspi 3/4 the antenna trace runs between the groundplane, on raspi it needs steady fingers, a high thermal capacity iron and a bit luck)
The antennas are rotated 90° in relationship to each other, perhaps having dual antennas allow us to select the antenna that better match the electric field direction of the router, perhaps some further tests rotating the antenna of the router in vertical or horizontal position and then measuring the signal intensity at each antenna of the ESP32 would be interesting. Does this make any sense? Great video as always !!!
That's the case. An omnidirectional antenna is good only when the other network node is transverse to the rod. In the direction of the rod the field is weakest. So having another antenna at 90° will always cover this dead zone of the other antenna. A link is best when antennas are parallel. It's weakest when antennas point towards each other (omnidirectional rod antennas only).
We also have to consider that we often do not exactly know the antenna direction of the opposite antenna (without opening the box). Plus, in buildings, reflections change the signals considerably. That is why I think having two antennas and selecting the better one should get better results in most cases.
Hi Andreas, at a quick glance it appears like in your spreadsheet the auto-mode only lists networks that were found by both single antennas and not either of them. Could it be a simple software bug?
Is there maybe something where it is disregarding signals with too low a gain to be normally useful? Is it trying to filter? It would make sense that the 20 signals it finds are the strongest. Maybe do a test where you connect to your own wifi from a far distance. See if it has some advantage there.
In my opinion, it should switch for weak signals because there it matters. For strong signals, any antenna will do. I hope they will give us the needed info when they watch the video...
The auto switch mode may need an established link to perform correctly. Since the test only is scanning for networks, there is no link and no link quality rating. Having only the Rx dB values of the own node may be perhaps not sufficient.
I agree (and mentioned it) that I do not know if this is an appropriate test. However, the decision probably has to be based on receiving and not on transmitting because our chip does not know the signal strength on the other side of the link.
@@AndreasSpiess I don't know if a normal WIFI link does, but if firmware supports it, the receiver can talk back to the transmitter and tell the last frame's Rx strength. So the original transmitter knows which antenna was best to send with. This method is also commonly used for dynamic Tx power control.
Could you test upside down with a 180° rotation where antenna 1 is exactly where antenna 2 was. This would permit to compare and see if the two antenna behave the same.
I would also compare the received stations in relation to the A0 and A1 orientation difference versus the West, South and East orientation of the module. So if A1 (right side antenna?) points to West, I would try to correlate reception through A0 with South direction reception. And may be flip it upside down too and correlate in the reverse direction.
(For some reason my comment disappeared.) Hi Andreas, I'm totally new to this field and I would appreciate your help please. I have Wemo D1 Mini with the battery shield but for some reason it draws 120mA with an empty sketch and, more importantly, 4,1mA when in deep sleep, which based on your videos and other guides and reviews is WAY too much. From other sources I've seen it draw 0,17mA and lower in deep sleep. Do you know what might be the culprit? What should I look into? Thank you very much
@@AndreasSpiess Yes, there are. But there is not 'best of deep sleep' video that I could find. I understand this is a complex question, but just in short - the only thing I need is deep sleep, battery and a reed switch (just a single GPIO pin). The device will be connecting to the wifi once is a month or so. What would you recommend?
Regarding the comment from 2:45 "If you do your own design..." I don't think this is a good advise to have the module antenna hanging out from the PCB, it just makes the design fragile and awkward. And FR4 is "invisible" to radio waves. What you need to do is respect the keep-away area indicated in the Espressif datasheet: No track or ground plane below the antenna section, only empty FR4 on both sides of the PCB.
I guess it was more meant to be like "clear the copper of that area". People forget to remove the copper from such areas, which is why "move the antenna outside the PCB" is kind of working. The copper is visible, and almoste always grounded. But yes, fragility is an issue
The board I showed with the overhanging antenna is from Espressif, BTW. My experience is that plastic around antennas influences their impedance considerably. I did no tests with FR4 but with PLA and the black covers of of ordinary antennas. And you have to consider the „stackup“ of PCBs in the impedance calculations of striplines, too.
@@AndreasSpiess The experience you have on PLA close to antennas is interesting, I think many makers among your audience (me included!) package ESP8266 or ESP32 devices inside PLA enclosures, considering the enclosure to be fully transparent to RF. It would be very interesting if you could elaborate on this. Maybe a good subject for a future video in one of your channels?
Maybe could use it to find the WiFi router or your lost phone. If it could compare the WiFi strength from the left with the right and then calculate from which direction the signal comes from
I would guess it's a combination of unfinished software and wrong use. No idea why it scans less SSIDs but In my opinion it's likely that it can switch to the correct antenna when associated with a station. (Since it may be utilising sounding frames typically used for mumimo to choose the antenna.)
It only looks "ugly" until you realize you could mount it at 45 degrees at a corner of your main PCB, at which point it would blend in perfectly. Just make sure the main PCB is actually missing the "L"-shaped corner bit that would be under the antennas themselves.
Hi Andreas, I would recommend you blur out the wifi networks and their mac addresses in the future as it is possible for malicious actors to determine your location from wifi networks.
According to its datasheet, the RTC6603SP RF switch chip cannot connect both antennas at the same time. And I did not test a second chip because I thought it is more a software than a hardware problem...
more antenna's and less io ! brilliant ! espressif are great engineers for shooting themselves in the foot. why cant we have an original esp32 with more io (like a useable serial port3) and more flash. lots more flash.. (and more io did i mention more io....)
Shame that manual doesn't say anything about it. Most likely diversity works only when connected to specific station. It checks RSSI of station from both antennas and decides which one has better signal.
@@AndreasSpiess I conducted a test to compare the signal strength of a single PCB antenna connected to ESP32 and a dual antenna connected ESP32 WROOM DA module. After testing both modules, there wasn't a significant difference in the RSSI value between the two (a difference of 10 to 12 RSSI points). Is there a way to connect with you to show that I can share the images of the serial monitor where the RSSI values are getting printed? To conduct this test, we configured the two modules to scan nearby wifi using an ardArduinouino example code. We configured the antennas in the DA module in AUTO mode as per the Arduino ide example code and placed the two modules next to each other. After flashing the code onto the DA module, we observed that the DA module was able to sense a few extra numbers of WIFI networks in comparison to the normal ESP32 with a single PCB-connected antenna. But not so much in the RSSI values. We expected a better response from the DA module. I would be grateful if we could just verify if these are the expected results from the DA module, as I have to make a selection between the modules for my next product
@Andreas Spiess is it possible for us to connect in any way so that I can share my project with you? Your guidance on this matter would be very valuable to me
For a mesh you do not need two antennas (as the Meshtastic project shows). These projects listen for a time and transmit for another time. You cannot transmit and receive at the same time on such close frequencies. You would only hear yourself.
I am buying only ESP32u with uFL..... and the range extension is huge. Ideal for outdoors projets.
Good decision!
Good reminder! I was just thinking about adding an ESP32 to my car but was worried that the range wouldn’t be long enough. I will try an external antenna.
For most projects the size and extra expense in any mass manufacture puts ufi in the hard basket, I think the DA offers those with manufacture runs a cheaper alternative in a smaller self contained package.
Which ones do you like?
On ESPEasy, I have the option to have multiple runs while scanning for a network. This often gives a lot more access points as not all access points react immediately on an active scan.
So you may want to perform several runs when scanning to get a good idea of the performance difference.
Also you might want to turn the WiFi off and on again (and wait a few-100 msec) to start scanning on the other antenna. When turning on the WiFi, it performs a RF calibration, thus maybe this can also improve the antenna performance if the antenna is selected before enabling WiFi.
I will also try to get one of those "T-shirt" modules, to test with. Have not had one of those yet.
On my own board designs, I almost always use the ones with an ufl connector, so the antenna can be positioned for best performance. Even those "sticker" antennas perform way better than the PCB ones, at least that's what I've seen.
I agree with your statements. I averaged a few scans because I also saw that they differed from scan to scan.
And I would expect that Espressif does all the needed stuff that the switching gets the best result (like calibration, etc.) Maybe they already did and I do not know because they kept it as a "secret".
I am glad if you do your own tests. And I agree that external antennas are a good choice. But they are clumsy and add to the BOM...
Dual antennas are very helpful where there is multi path (destructive) interference.
Perhaps a more enlightening test would be to orientate a metal plate quarter of a wavelength away from one antenna such that it is in a null (ie a sharp drop in RSSI due to the destructive interference). Then switch to the other antenna, where the signal should be outside the null, due to being spatially separated.
Adding the RF switch will increase insertion loss and so degrade the noise figure. Therefore the benefits of multi path tolerance have to balanced against slightly degraded sensitivity.
I agree with your analysis. I for sure would do those tests when designing such an antenna. For a user like me, it seems to be more important to just see how big the effect is in reality.
In addition, the effects will be different in every situation (13cm wavelength!). So it is more a general feeling that it could help. But only if the switching works properly ;-)
The insertion loss of the RTC6603SP chip is 0.4dB @ 2.5GHz, BTW.
@@AndreasSpiess Interestingly, there exists "space time block coding" aka STBC.
Here, the access point uses two antennas at the same time in a way that a single receiver antenna can make use of. If the interference exists for one of the transmitting antennas, the other antenna may come in loud and clear.
The ESP6266 supports STBC, it seems.
(The other way around it is trivial, the AP can use whichever antenna works best to receive the single signal from the ESP).
I am very happy to see someone test and talk about this product. I'm actually working on a street light project for my neighbourhood where our streets are private, we have ~40 streets poles each distant by
Thank you for your kind words. An I hope you will be successful with your project!
That pointer with the little, white gloved, hand on it always brings a smile.
Mr. Spiess, your content is always exceptional, and your integrity is beyond reproach. You are a true treasure of RUclips.
Thank you for your kind words!
The shown WIFI routers with multiple antennas are completely different from this ESP32 module. All of the antennas are active at the same time. There is no switch that selects just one antenna. Multiple active antennas are used for beam forming like a phased array or to receive multiple signals simultaneous (MIMO). If the ESP32 could use both antennas as a phased array to do beam forming, it would get a better signal than any one of the antennas can receive by itself.
Thank you for your info. I know about beamforming but thought that it needs an array of precisely spaced antennas. The antennas of the routers I showed have moveable antennas. That is why I thought they could not be used for this purpose.
@@AndreasSpiess I am no expert, but I believe wifi does not actually know what the world around the router is like. It just tries different combinations of phase shifting between the available antennas until it gets the best rx signal. Then tx can use the same parameters.
Another interesting fact is that the antennas on many wifi routers are somewhat fake. They are made to look impressive. But actually they are just empty plastic shells with the actual antenna inside being much smaller. The wavelength of a 5 GHz antenna would only be 6 cm and the antenna probably a half wave in length. Yet the router has huge antennas.
@@BaldurNorddahl wow! marketing in design.
Most routers nowadays still only have 1 radio per band and the antenna is switched but they do have their own amplifier
@@lifeai1889 modern wifi features like MIMO and beam forming only works if you have multiple antennas active concurrently.
I think that multiple antennas in routers implement MIMO operation - multiple inputs and multiple outputs, they work together at the same time. Here the antennas work one at a time. I assume dual antennas were included to improve directivity. Auto mode would allow connecting to the known AP using an antenna with the better RSSI. This can easily be tested by logging the RSSI for a nearby AP in ant0, ant1 and auto mode. The latter should return the maximum value out of the two.
Yes, I wanted to remark this as well!
The access points that have multiple antennas actually have multiple receivers (and transmitters), not an antenna selection switch.
In routers it's more complicated. They indeed use multiple antennas for MIMO but also in this same time do what this ESP module aka try to find best antena for every connected device. In most advanced ones also beamforming signal.
Thank you for the additional info. I never looked into the router antennas.
@@AndreasSpiess I learn a lot from your videos and, on rare occasions when I can add something (I think) worthwhile, I do with all my respect to the creator
For a test I would propose, try to connect to a WiFi network, and then compare the two-antenna to the one-antenna version. I imagine the auto-mode to be effective when the module knows what network you want, and then switches between the antennas to get the best signal for the desired network. In such case, a test might include a setup you can carry around, seeing how it switches between antennas when the relative direction of the signal changes.
Yes; it seems we don't know what AUTO is optimizing for. If a network is selected, it may choose the best antenna for that. but if not, like in this test, it could as well try to get as many as possible as try to optimize the strongest (or pick up one arbitrarily, a.k.a. simply don't know what to do... :-)
Maybe this would lead to different results (as I mentioned in my video, I do not know). However, I do not understand why auto should not work for scanning. A module first has to scan the frequencies to find an access point. If auto does not work it probably would not find the proper access point. At least this was my reasoning.
Great information, as always. Thank you 🙏
That breadboard power supply looks interesting too.
I showed it in the last mailbag videos.
This is very interesting. Now we just need to wait for other companies to make pre-soldered development boards so we can just buy one from many sellers online with everything ready to go upon opening the package. Thank you for showing us this new strange device variant.
I hope that this will happen. The module is quite new.
I believe some newer bluetooth modules have a dual antenna configuration to measure the Angle of Arrival (AoA). That could be interesting to test
I never saw AoA for BLE so far, but I saw it with Ultra Wideband on the new Iphones. There it works very well.
AoA is on BLE 5.1. No ESP chip supports BLE5.1 yet
Ivan Grokhotkov is the chap at Espressif who knows most about the software. You can get his attention by leaving a clear description of the problem on their forum, but they like you to check to see if anyone else has mentioned it before, and if not Ivan will usually find the trouble out and fix it. He's got a sidekick who comes under the name of Me-no-Dev and he can usually deal with most things.
Thank you. It is already on their internal Jira list...
on the road to antenna diversity, but not quite at the destination. :-)
True!
again a very interesting video. And the results are quite a surprise, I wonder if Espressif will elaborate on it someday. Looks promising though! thanks for the update!
I am sure that the Espressif engineers get the message ;-) Maybe they do something...
Dual antennae usually used for diversity purposes. Dual rx are useful for calculating multipath signals and diversity gain. All cellular communications use atleast two rx paths for receive diversity. Two TX is mostly used in MIMO feature.the antennas kept at 90 degree, the purpose may be +45 & -45 cross polarization.
Thank you for the additional info. I did not know that cellular phones use space diversity. I only knew it from HF systems below 30MHz. And I agree with MIMO. But I think this is another story...
If you turn the complete circuit upside down the position of the antenna's 1 and 2 are swapped. If assumed the characteristics above and below the antenna (the z axis) are the same this would give interesting additional information.
That would have been an excellent test to see if ant1 is still better most of the time (without asking Pedro to do the test).
Could you hook up a oscilloscope to the diag pins to see how/when the antenna select switches in auto mode?
The two testpins are no more accessible:-(
Thank you so much, it was a great note to mount the esp 32 antenna outside our PCBs.
I wish you can do a video that comapres between bare ESP32 modules out side thier boards
This is probably too specific and would not interest a lot of people :-(
One option that you may want to test is to orient the board so that one of the antennas is oriented horizontally, and one vertically, and see if that helps with lower power signals. I know that RF that hits certain typologies can end up switching from horizontal to vertical polarization, and having a platform that allows you to switch between the two for some environments may give you far better coverage. That said, WiFi wasn't really designed for 'weak signal' work, so there are likely to be other issues that prevent a stable connection if polarization gives you a significantly different signal level. And that may also be part of what the results you see from the Auto test are showing, that suggests that of the 56 signals observe on one antenna, and the 30 some odd on the other, only 20 overall were within whatever level the system accepts across both antennas. But I don't know, I don't work for Espresif.
Two antennas with different polarization would help a lot. But it is much more complicated to manufacture, I think.
@@AndreasSpiess not if you orient the board on edge and elevated @45 degrees
In multi-path conditions, sometimes an irreducible bit error rate (BER) is present. But diversity combining wouldn’t switch between antennas but use them both and some kind of quality measure to combine them.
On HF we sometimes use "space diversity" where we do a similar thing with the two antennas. But we also do not connect them. We select the stronger signal after the receiver. This can create very good results.
Is the performance difference the same (between the normal module and the T-shirt) if you orient them vertically?
This depends on the polarization of the other module.
Hi, could you tell me how to make ESPHome send three numbers (channels) to the TV, please?
I've been looking for days on this and haven't found anything.
Thanks.
I have no idea on how to do this. Sounds rather complicated. But maybe somebody already has solved it.
Nice, it's quite frustrating when the documentation doesn't help, I spent days trying to resolve issues with a Gyro chips interrupt parameters to find that the manual had an error.
That is even worse because usually we tend to trust the datasheets!
@@AndreasSpiess As Dave Jones would say, "that's a trap for young players!" ;)
Ah, wrong datasheets! I ran into some of those during my time in Corporateland.
One especially egregious case: I was designing a module that needed some sort of little MCU for monitoring, communicating via I2C. I was leaning toward an AVR device, but Management decreed that I should use a particular 8051-based MCU because an engineer in another division had already used it successfully in the same role, and had firmware already developed and ready to adapt. After reviewing the datasheet, I went along with this.
Well, my module having been designed and built, I made the pilgrimage to the other division in search of firmware to adapt, and found that the other module hadn't been built nor the firmware tested, so my module became the test platform.
And, surprise! The documentation for the MCU's I2C interface was wrong; in reality, a key feature needed to support... I guess we're not supposed to use the old terms?... anyway, it could operate fine in "mad scientist" mode, but in "hunchback" mode there was something fundamentally broken. And, of course, there already being one "mad scientist" on the bus, these modules were all supposed to operate as bus "hunchbacks".
I don't recall what the resolution was - whether we came up with a firmware hack to make it work well enough for the purpose, switched to a different MCU, settled for a dumb module, or just abandoned the whole project before getting module communication sorted out.
(This wasn't even a brand-new MCU, so the re really ought to have been at least an erratum for the I2C issue, but I guess I2C on MCUs is almost entirely used for talking to EEPROMs, temperature sensors, and the like, so it's entirely possible that I was the first one to try to use it that way.)
I wonder, on balance, if a single external 1/4 wave vertical antenna is going to be better or worse than a (well optimised) switching between two pcb antennas?
I have got into the habit of using an external antenna for esp32's I put outside, and this would be easier and smaller
An external antenna is for sure a good possibility and usually has a better performance. Often, people do not want one because of the added size or cost.
@@AndreasSpiess On the other hand, with an external antenna one has more freedom in positioning the ESP.
Me too, I often are confronted with the limited coverage of the standard antenna. I started adding AP's in my network, but still, an external antenna performs much better. I can live with the additional cost and form factor.
There are wroom modules with ipex connector if you need an external antenna
I use "3d antenna" like the one on TTGO module. I made the same test as Andreas Spiess and saw that the 3d antenna perform really better than the PCB one.
So for me it is a good quality / price / space ratio
Ugly? No. It looks cool! I really want it soldered on a PCB and the entire board looks like spaceship :)
RUclips has its own rules for thumbnails ;-) Glad you like it!
You can solder sma connectors to antenna traces and connect your qo100 amplifer to one of them and set it to tx only and use other antenna for rx. It should increase range a lot.
Maybe they will come up with a design with two antenna connectors. That would create some new possibilities, I agree.
Initially, I thought perhaps the dual antenna setup might be for diversity signal RX. Is there something in the docs that points in that direction? Can diversity RX be done in software? I used to use some cellular modems that had that feature if you connected two antennas to them that could improve reception in noisy environments or improve weak signals if the device was far from the nearest cell tower.
I would call it "slow space diversity" because it has 2 antennas but only one receiver. But it switches to the better antenna from time to time (at least this is what I assume, but not know). This should improve reception, particularly for fixed setups.
This is exactly the sort of thing that completely frustrates me about products from China, guessing about how they work due to the lack of pertinent information. It costs essentially nothing to post on the Web ALL of a product's useful info. There are numerous such products that I simply don't purchase due to their lack of information, and I am certain there are many, many other prospective purchasers just like me who simply purchase something else or nothing at all for that very reason.
The people who purchase these products are NOT typical 'plug and play' consumers who want products that they don't need to think about. We design things with these products, for ourselves, for our friends, and for businesses. We NEED technical information on these products, ALL of it. But we are kept in the dark. It's as if the vendors don't really want to sell their products.
The primary reason, (other than your dry sense of humor,) that I watch your channel is because you are much better and more dedicated at gaining that sort of info than I am, ( as I said, I just give up and buy something else, or buy nothing at all,) making this my 'go to' source for specific technical information on this sort of product. Even if your humor is dry, you are a fountain of information. So I always sit in the first row. I sincerely hope your channel gets 10 million subscribers so that you will have the leverage necessary to pry from the suppliers or manufacturers themselves ALL of the essential technical information about their products. We NEED it. Thank you for all of the info that you DO drag out with your detailed research and your experiments. Oh, and thanks again for your dry sense of humor.
Thank you for your kind words. I know that many Espressif engineers watch my videos. We will see if something will happen.
But I have to confess that I am also not good at documenting things. I know of many other engineers who are the same. We want to build, not write ;-)
WROOM-DA sounds like a sound car getting up at high RPM and going down.
True!
I think that’s the intention behind the Wroom name.
Pretty sure it’s not w-room ;)
@@AndreasSpiess Oh yea, do you now have/carry a spare ECM power relay for your automobile? Maybe some tools and a contact file would be enough. Hoping for the best...
Awesome info, thanks for much Andreas.
My pleasure!
Shape reminds me of a B2 bomber. Seems like a lot of effort that can be solved by a UFL antenna. That 20 seems to be a software limit at face value. Think it is picking the best 20 access points based on RSSI. I know there is a soft limit to the amount of access points the ESP can store in memory on a single scan.
Maybe 20 is the reason. But this seems not to be the case if I select only one antenna. Strange...
Lord Voldemodem is disrupting your signals for sure!
:-))
Good info @Andreas Spiess, Useful ! how about BLE dual antenna config? is the same radio switching concept work there?
I have no information about BLE.
@@AndreasSpiess thanks
it seems you could make 2 wifi devices with this.. make a nice repeater too...input on one, output on another...
the auto switching maybe fast (or slow!)..so you miss alot of ssid's..when its set to one its constantly on and doesnt miss any frames...
...this with external antennas would be interesting... (you also missed the chance to compare against an external antenna in this vid!)
good stuff Andreas.. i didnt even know this existed till now! ..thanks!
Unfortunately, the ESP32 only has one radio. And repeaters in such a small space are nearly impossible because of cross-talk from the transmitter to the receiver when they work at the same time.
Indeed usefull ! The 2 antenna's seems to me a good idea.
I agree!
0:11
T Shirt... so modest of you
:-)
Thanks for a great video of this new module. I have a general question,
How do you distinguish from a WiFi and LoRa antenna? I bought several antennas from second hand but I don’t know how to identify them, nothing written on them. Thanks for any help!
There are videos on this channel on this topic. You need a nanoVNA
Perhaps the auto mode does some calculation on whether the rssi is feasible for a real connection versus 'can see it at all'.
Maybe. I hope they will tell us...
Thanks for making another ESP32 video!
You are welcome!
This has some potential but it needs more tests in the field. Next a ESP32 with a build in Yagi antenna. ... wait I see some potential in that fitted in a nice weatherproof enclosure. Or maybe we could use one with a uFL connector soldered on a pcb with a microstrip line collinear antenna.
Hey why not dual ufl?? 😎
Two connectors would be cool, I agree.
And I have a directional WiFi antenna in my lab waiting for a test. Maybe when the weather is warmer...
Andreas, it looks like your standard ESP32 had the antenna over the breadboard, while with the WROOM-DA, it was sticking out over the edge of the breadboard. I wonder if this could adversely influence the coverage of the ESP32? After all, you did suggest that any PCB design should not have the antenna over the board... it should stick out to avoid RF shielding. Am I right?
You are right. But I usually compare things which are available and used by my viewers.
Thinking aloud: Would it be possible to de-lid the module and have a look at the RF switch? Watching W2AEW's channel, it seems like it shouldn't be a complicated circuit. Won't tell us much about the switching algorithm, but it's a start.
The chip they use is in the datasheet: An RTC6603SP (0.1 GHz - 3.0 GHz SPDT Switch)
So, when will there be a 3 antenna version with one for the Z axis? Would that even be useful? Maybe for very mobile devices (drones?) since it could help with kinda directional antennas since the right antenna would only send in roughly the right directions.
Personally I like external antennas more than PCB antennas when I can't use POE for a Project, at least I think it's easier to place them outside a case.
There would be no additional benefit in using 3 axis antennas. A rod-like or omnidirectional antenna is weakest in performance when it points directly towards the communication peer. So when you have a second one at 90° angle (transverse to the peer direction), that one is guaranteed to have the good orientation. Having a third antenna would have the same 90° to the weak antenna, but there is no additional gain in sensitivity / range / receive strength.
Adding an antenna with a different polarization would make sense because the effect of polarization is quite high. But then the device would become much bigger and probably harder to manufacture. Maybe they will create one with two connectors. Then everybody could create the setup he needs (also with external antennas).
Very cool, great assessment, thanks
You are welcome!
Nice video andreas. But is it possible that your test algo is not correct for the test of the auto function? Wlan scanning is a good to have feature. I think you must connect to a wlan and the algol of the chip checks on wlan idle which ant have the best receiving quality (or sending?) and switch between the two ant? Can you connect to the pin how controls the rf switch?
AS I mentioned, I am not sure if I chose the right test procedure. However I do not know why they would not use a proper antenna selection for the scanning. If you do not find the access point it is probably hard to connect to it...
I also do not see how we could use the TX signal because our chip does not know how good it is received at the opposite station.
I hope we will get some insights from the Espressif engineeers...
@@AndreasSpiess Its possible that the engineeers think you will "hard code" the ssid and pw. I does not matter how you send the signal, because the receiving system also have multiple antennas most of the time. Its only how good the receiving signal of you choice is. But look in the capsman of you wlan system. You will see also both ignal qualitiy in the list.
Yes, Lord Voldemodem and Pimmelmatz are definitively the names you should use for your WiFi :-)
Good to know!
Can you show how to solder those chips? Do you use hot air or do you really solder them with an iron?
I solder one pin with my ordinary soldering iron and then a second one at the opposite side. When everything fits I solder the rest and check with the ohm meter if I created any bridges.
@@AndreasSpiess Thx. Then I will not try the heat gun.
@@michaelkaercher I use the heat gun for desoldering of small parts only.
When several years ago i did some experiments with Wifi signals around my house i noticed a lot of signals from cars and other moving things. (Mobile phones) Also it seems some phones constantly change there mac address to give privacy e.g. in shopping centers where they do/did tracking. So the timeframe is also very important when sampeling.
You are right with Mobile phones. They change the MAC address often. In my case, I do not see it as a problem as not a lot of moving Smartphones were present (names usually are different, and I know a few AP names from past tests).
Usually I agree and I am pleased with your torough reviews. I may be mistaken, but I think you are missing the point. Two antennas are not meant to discover more networks, but to split incoming and outgoing traffic on separate channels to increase speed and throughput. This is the main advantage !
Second my friend, W is not = to V in English ! LOL ! It's not V-ROOM it's WROOM like the roar of a car ! Have a great day ! Auf Wiedersehen !
The ESP32 only has one radio. So no improvement of the speed :-( Would be nice.
Sirquit? Sounds like me when trying to figure out microcontrollers 🙂
:-)
If the antenna-select GPIOs are accessible as described, then they could be monitored during operation to see if the auto mode is even switching between antennas
Good idea - and how frequently they are switching...
A good idea. I should have thought about it before I soldered it to the PCB :-( But I hope we will get the info and do not have to do reverse engineering.
If access point have multiple antenna, it is more for mimo / beam forming.
Are you sure? These antennas are moveable and I thought beam forming needs some precision in distance. But I never looked into it.
Great review
Thanks for sharing your results 🙂
Thank you!
Interesting to find out that the two antennas cannot be used simultaneously. I was looking for a module where I can use WiFi and BT simultaneously. So this module won't work for my purpose. Thanks for testing as it helped me.
Afaik there's only one amplifier and associated hardware for generating the signal. This is just two antennas, not two separate radios (which I agree could be good but rip power consumption)
From my own experience the esp 32 has not enough space for both libraries. Maybe try to compile both libraries in the the Compiler just to test if it works bevor you buy one for that purpose
The ESP32 uses the same radio for Bluetooth and Wi-Fi (similar frequencies).
@@Tom-ku8bu is that for Arduino programming I assume? I've not tried that but I use the IDF at home and for work, no problems at all with WiFi and Bluetooth libraries concurrently. Still tonnes of space for other code too. I even made a wi-fi to Bluetooth speaker bridge using an esp and it handled that no problem (audio from VoIP to the esp, which pumped it out over Bluetooth a2dp to a Bluetooth speaker)
@@DrGreenGiant Interesting. I also wanted to make a wi-fi to Bluetooth speaker bridge with an ESP32. By chance is your code in github?
Given the alignment of the antennas could this be a diversity setup for for a mobile based system?
I would call it "space diversity". It can be used for any application, not only mobile, I think.
@Andreas there is a use for having a send and a receiving: replace the receiving Antenna with a high gain directed one... new challenge: long range WLAN with ESP Modules
Edit: The challenge is meant to be w/o breaking the emision laws!
If they would add connectors instead of antennas, we could build interesting things. Like that it would not be easy to properly mount a connector.
@@AndreasSpiess is it possible to solder a miniRCA on the module? (e. g. like on Raspi 3/4 the antenna trace runs between the groundplane, on raspi it needs steady fingers, a high thermal capacity iron and a bit luck)
@@emmoemminghaus6455 I did not try. But it should be possible.
@@AndreasSpiess I like to try... but the only source for this module is Mouser, and i have not enougth volume to avoid the shipment cost.
Have you seen the ultra wideband modules that you can use for locating each other like apple tags?
Yes. But they are very different. I plan a video about them but need more info.
The antennas are rotated 90° in relationship to each other, perhaps having dual antennas allow us to select the antenna that better match the electric field direction of the router, perhaps some further tests rotating the antenna of the router in vertical or horizontal position and then measuring the signal intensity at each antenna of the ESP32 would be interesting. Does this make any sense?
Great video as always !!!
That's the case. An omnidirectional antenna is good only when the other network node is transverse to the rod. In the direction of the rod the field is weakest. So having another antenna at 90° will always cover this dead zone of the other antenna.
A link is best when antennas are parallel. It's weakest when antennas point towards each other (omnidirectional rod antennas only).
We also have to consider that we often do not exactly know the antenna direction of the opposite antenna (without opening the box). Plus, in buildings, reflections change the signals considerably. That is why I think having two antennas and selecting the better one should get better results in most cases.
Hello if I buy this module can I use the BLE and WiFi at the same time.
This has nothing to do with the antenna. All "standard" ESP32s can use BLE and WiFi in parallel, but with restrictions.
Can the code deployed on the ESP32 be encrypted/locked (for a commercial product) to prevent, of course, a potential competitor from stealing it?
I do not know because I only work with open source SW.
@@AndreasSpiess Thank you.
So perhaps the auto mode uses a different metric for choosing the antenna. RSSI maybe? Of course that doesn't really explain the low unit count.
We will see what they will say…. Or maybe we will get a fix ;-)
Hi Andreas, at a quick glance it appears like in your spreadsheet the auto-mode only lists networks that were found by both single antennas and not either of them. Could it be a simple software bug?
Well possible. I hope we will hear from Espressif
NAND logic? Looking for an OR result... lol Maybe that's it!
Is there maybe something where it is disregarding signals with too low a gain to be normally useful? Is it trying to filter? It would make sense that the 20 signals it finds are the strongest. Maybe do a test where you connect to your own wifi from a far distance. See if it has some advantage there.
In my opinion, it should switch for weak signals because there it matters. For strong signals, any antenna will do. I hope they will give us the needed info when they watch the video...
That's the first time I've heard it as "Read The *Fine* Manual".
I learned if from a viewer ;-)
Great video as always Andreas! 😁
Glad you enjoyed it!
The auto switch mode may need an established link to perform correctly. Since the test only is scanning for networks, there is no link and no link quality rating. Having only the Rx dB values of the own node may be perhaps not sufficient.
I agree (and mentioned it) that I do not know if this is an appropriate test. However, the decision probably has to be based on receiving and not on transmitting because our chip does not know the signal strength on the other side of the link.
@@AndreasSpiess I don't know if a normal WIFI link does, but if firmware supports it, the receiver can talk back to the transmitter and tell the last frame's Rx strength. So the original transmitter knows which antenna was best to send with. This method is also commonly used for dynamic Tx power control.
Could you test upside down with a 180° rotation where antenna 1 is exactly where antenna 2 was.
This would permit to compare and see if the two antenna behave the same.
I would also compare the received stations in relation to the A0 and A1 orientation difference versus the West, South and East orientation of the module. So if A1 (right side antenna?) points to West, I would try to correlate reception through A0 with South direction reception. And may be flip it upside down too and correlate in the reverse direction.
This would have been a very good idea! Maybe I will do it later. When I have better info from Espressif.
the Fine manual! i like that
I learned it from one of my viewers ;-)
(For some reason my comment disappeared.) Hi Andreas, I'm totally new to this field and I would appreciate your help please. I have Wemo D1 Mini with the battery shield but for some reason it draws 120mA with an empty sketch and, more importantly, 4,1mA when in deep sleep, which based on your videos and other guides and reviews is WAY too much. From other sources I've seen it draw 0,17mA and lower in deep sleep. Do you know what might be the culprit? What should I look into? Thank you very much
Comments with links often are deleted on RUclips. I made a video about the D1 Mini and deep sleep. These boards are not made for that purpose.
@@AndreasSpiess what would you recommend for deep sleep, suitable for a beginner? In the batch that i bought were also ESP-1 and ESP-1s.
@@guyfromostrava There are many videos about deep-sleep on this channel.
@@AndreasSpiess Yes, there are. But there is not 'best of deep sleep' video that I could find. I understand this is a complex question, but just in short - the only thing I need is deep sleep, battery and a reed switch (just a single GPIO pin). The device will be connecting to the wifi once is a month or so. What would you recommend?
Regarding the comment from 2:45 "If you do your own design..."
I don't think this is a good advise to have the module antenna hanging out from the PCB, it just makes the design fragile and awkward. And FR4 is "invisible" to radio waves. What you need to do is respect the keep-away area indicated in the Espressif datasheet: No track or ground plane below the antenna section, only empty FR4 on both sides of the PCB.
I guess it was more meant to be like "clear the copper of that area". People forget to remove the copper from such areas, which is why "move the antenna outside the PCB" is kind of working. The copper is visible, and almoste always grounded. But yes, fragility is an issue
The board I showed with the overhanging antenna is from Espressif, BTW. My experience is that plastic around antennas influences their impedance considerably. I did no tests with FR4 but with PLA and the black covers of of ordinary antennas. And you have to consider the „stackup“ of PCBs in the impedance calculations of striplines, too.
@@AndreasSpiess The experience you have on PLA close to antennas is interesting, I think many makers among your audience (me included!) package ESP8266 or ESP32 devices inside PLA enclosures, considering the enclosure to be fully transparent to RF. It would be very interesting if you could elaborate on this. Maybe a good subject for a future video in one of your channels?
@@opsahle2 I can confirm Andreas' exprerience. I started to use separate ULP/SMA separate antennas in enclosure to compensate for the signal loss.
@@opsahle2 just perform some rssi tests with and without the box ;-)
Maybe could use it to find the WiFi router or your lost phone. If it could compare the WiFi strength from the left with the right and then calculate from which direction the signal comes from
I do not think this would be very precise... Apple has something in their new iPhones which work remarkably well. But only on short distance.
+1 because the tshirt comparison! haha
:-)
I would guess it's a combination of unfinished software and wrong use. No idea why it scans less SSIDs but In my opinion it's likely that it can switch to the correct antenna when associated with a station. (Since it may be utilising sounding frames typically used for mumimo to choose the antenna.)
I hope we will get the needed info from the developers. I am curious.
It only looks "ugly" until you realize you could mount it at 45 degrees at a corner of your main PCB, at which point it would blend in perfectly. Just make sure the main PCB is actually missing the "L"-shaped corner bit that would be under the antennas themselves.
A corner mounting indeed is a good idea. I did not think about it.
Hi Andreas, I would recommend you blur out the wifi networks and their mac addresses in the future as it is possible for malicious actors to determine your location from wifi networks.
There are many other possibilities to find my address. We do not care here in Switzerland. Maybe we are naive ;-)
did you try another module and confirm this behaviour? It almost looks as if in Auto both antennas get connected 😎
According to its datasheet, the RTC6603SP RF switch chip cannot connect both antennas at the same time. And I did not test a second chip because I thought it is more a software than a hardware problem...
Very nice video
Thanks!
Thanks for subtitles in Portuguese
:-)
You're welcome. I do my best to the help the community.
I believe the design is meant for stealth applications.
I think so. If you have plenty of space you probably are better off with an external antenna.
i thought it was super cool looking
:-)
You can use one for wifi and the other for Bluetooth?
No. Bluetooth and WiFi use the same radio.
Read the fine manual 😂
:-)
more antenna's and less io ! brilliant ! espressif are great engineers for shooting themselves in the foot.
why cant we have an original esp32 with more io (like a useable serial port3) and more flash. lots more flash.. (and more io did i mention more io....)
The -S3 seems to have more IO. But I did not check how much.
I hope it does not replace the one antenna version otherwise it doesn't fit in a model car (1:87)...
For sure they will keep the standard antennas. I can imagine you do not have lots of space in a 1:87 model ;-)
The angle between antennas suggest that maybe this is intended for Bluetooth positioning
So far I did not find a Bluetooth example and nobody from Espressif mentioned this protocol.
I have an idea-er: You should make us a phased-a-radar cantenna.
Unfortunately, I am not so good at antennas and machining :-(
Of course a software guy starts at zero - all vector indexes start at zero.
We don't use pascal any more.
And the hardware guy at 1. Like R1 or C1
@@AndreasSpiess My compromise of starting vector indexes at 0.5 was dismissed without, I thought, proper consideration.
Interesting video.
Thank oyu!
The better labeling would be: ROTFL32 😄
:-))
Shame that manual doesn't say anything about it. Most likely diversity works only when connected to specific station. It checks RSSI of station from both antennas and decides which one has better signal.
I wonder why this could not be done while scanning :-( RSSI is available at least ( as shown in my tests)
@@AndreasSpiess It is all in software, I hope they will implement this in future.
Possible use every core for a different software? Select the antenna can be done with the new OS dev version. Oh he tells it right now, haha..
The antennas are connected to the radio, not to the cores :-(
@@AndreasSpiess Yes but i can be switched so far i have heared.
@Andreas Spiess Can you please share your code related to this project ??
As said in the video: I used the example file provided in the Arduino IDE
Heyy @@AndreasSpiess What are the sources of the json files while installing of esp32 broads in Arduino ide
@@vaibhavsingh6775 Just search for "using the ESP32 in the Arduino IDE". The -DA is part of the standard ESP32 family.
@@AndreasSpiess I conducted a test to compare the signal strength of a single PCB antenna connected to ESP32 and a dual antenna connected ESP32 WROOM DA module. After testing both modules, there wasn't a significant difference in the RSSI value between the two (a difference of 10 to 12 RSSI points). Is there a way to connect with you to show that I can share the images of the serial monitor where the RSSI values are getting printed?
To conduct this test, we configured the two modules to scan nearby wifi using an ardArduinouino example code. We configured the antennas in the DA module in AUTO mode as per the Arduino ide example code and placed the two modules next to each other.
After flashing the code onto the DA module, we observed that the DA module was able to sense a few extra numbers of WIFI networks in comparison to the normal ESP32 with a single PCB-connected antenna. But not so much in the RSSI values. We expected a better response from the DA module.
I would be grateful if we could just verify if these are the expected results from the DA module, as I have to make a selection between the modules for my next product
@Andreas Spiess is it possible for us to connect in any way so that I can share my project with you? Your guidance on this matter would be very valuable to me
I actually think it looks cool.
I agree because I love antennas!
Why didn't they built a module with two IPEX connectors?
I do not know. Would be cool for experimenting...
2 antennas could be useful as for a repeater function . for a mesh maybe . iot mesh .
For a mesh you do not need two antennas (as the Meshtastic project shows). These projects listen for a time and transmit for another time. You cannot transmit and receive at the same time on such close frequencies. You would only hear yourself.
Instead you could have flipped the ESP32 over so A0 and A1 were pointing in each others former direction and then check if you get the same results
I agree. This would have been a very good idea!
👍
:-)
i tested the range , it is giving 700m on line of sight . 104m inside building.
Very good!
@@AndreasSpiess sir you so kind, your video really helped me a lot.
**creates issue in internal Jira**
Thank you! I will be on one of your Friday's meetings (Pedro will organize it).
BLE not work with wifi. May be BLE and wifi radio work same time with 2 seperated antenna. 🤔
Bluetooth and WiFi use the same radio in the ESP32. So they never will work exactly at the same time.