10MHz Rubidium Frequency Standard and Signal Distribution Amp Follow-Up
HTML-код
- Опубликовано: 9 авг 2013
- gerrysweeney.com/10mhz-rubidiu...
Followup to article:
gerrysweeney.com/build-a-10mhz...
When testing the 1 PPS output I found some strange output which needed further investigation. I also decided to make further use of the micro controller and status LED to indicate the 1 PPS signal to provide a degree of visual confidence that the frequency standard was working. 
Наука
Excellent video, thanks Gerry for your time and efforts.
HI Paul, glad you like the vids and very pleased to know you are learning something. Thanks for the feedback. Gerry
The ringing on the 1pps signal could be caused by it not being terminated by 50 Ohm.
Thanks for the feedback, I used some matt black tape and took the measurement from that which seems to work well. Gerry
Hi Amandel, I did look at that, but n mine the ripple is definitely there and I have since found other references to the same problem. The problem I did have though was some "ringing" on the wiring and I fixed that by slugging it with a bit of capacitance - a little bodgy but does the trick and is simple. glad you liked the video, thanks for the comments. Gerry
Fair enough, a microcontroller might cost a bit more but let's not forget the value of your time.
It's often tempting to try and squeeze something into the fewest lines of code, or into the fewest and cheapest components, but there's no point spending half a day trying to save 50c on parts!
Good video as always, thanks for sharing.
I like this video much better than the first one you did on the subject. That one was more about fabrication, and this one was about the actual circuit and how it works. I have tried a couple times to watch the entire first one, and never make it far, so maybe I'm wrong, but this video held my interest and I really enjoyed it. Good work!
Your welcome, thanks for the comments and support. Gerry
Hi @BersekViking, I just modified the firmware on my prototype, changed the code around and implemented and GPIO shadow register and that solved the problem, I can now use interrupts reliably. Thanks for the pointer. I have spent so long using the PIC18Fxxx series chips I had long since forgotten this issue so thanks for bringing it back to my attention - most helpful. This could justify a video in its own right. Gerry
The data sheet for my Accubeat rubidium standard does state a max temp for the unit. So yes, it does need to dissipate some of the heat so it does not overheat.
You don't want the unit getting too hot. You might want to keep it below 80° F.
so so well done gary,,,very good work ,,,mark
Great video, as I remember i had some problems with IOC with PIC12F627 microcontroller 10 years ago (in time that bug now is feature was not documented).
According to errata "If a change on the I/O pin should occur when the read operation is being executed (start of the Q2 cycle), then the GPIF interrupt flag may not get set". Interesting in that time I scratched had for days because program passed testing in simulator and when you put it on chip i don't work, you get lost about 25% of interrupt.
The trick I learned at my last place of work was to use Tipp-ex (correction fluid). You paint a little onto the heatsink while it's cold, leave it to dry, then point the IR thermometer at that. Works really nicely if you're doing thermal testing with a thermal camera (e.g. a FLIR).
Thanks Gerry...ordered 10 from them just now. There's another seller here in the states that I'll get some from as well. Can never have too many "spares" of anything electronic...always nice on a Sunday afternoon to know you probably have the part you need in your own supply!
Dino
Hey Gerry,
Another great video, don't take people's feedback as criticism, people just have questions and suggestions.
It's always seems crazy to me that people complain when using a cheap micro controller on place of a number of other components.
Thanks again, keep it up!
Tom
nice follow up gerry, don't let people get you down man. your videos are full of insightful and educational info. is it possible to get a rough schematic of the circuit diagram of the buffer? i would like to build something like this. tnx
Thanks for the follow up. Finding fault and tweeking is 90% of a build.
I'm with you on the heat sync, once you have a lock you have a lock, any extra heat just comes from below par insulation and if the heaters have to come on now and again to compensate, so what, everything is happy. If it loses sync you can program the pic to beep if it does.
Re MC's, they where made to make life easier, some aren't up to certain jobs but on the whole even if you buy a CONT....
That would make sense, it did seem like there was some issue, it consumed so much time I just gave up in the end. Next time I will use an PIC18Fxxx that I know works! Thanks for the insight. Gerry
Hi Dino, thanks for the feedback. I am using an ICD3 but if you are just wanting to program the PICKit3 is a lot more cost effective. To be honest, I rarely use the debugging capabilities of the ICD3, I could probably work just as well with a PICKit3 myself. Gerry
That could well be it - the whole chip only has one port so that would make sense, I think the 12F series are pretty close to the 16F series so it probably applies. I will have to try it again with that in mind and see if I can get it to work. Thanks for posting the info. Gerry
Hi John,
thanks for the feedback. Indeed you can't please em all, next time I think I will put a 100 pin micro in to work the LED :) that should get some people going - or even better, I have a whole bunch of Z80's, EPROMS and SRAMS....whoooohaaaaa....now that would be fun...
Gerry
I've purchased one FE-5680A from ebay and i am in worry because i think that the unit is at the end of life. First, is it normal for 1pps signal to led blink faster than 1 second? I tried that just by connecting an LED directly on pin6 and to the ground. Also the frequency i've measured on my PM6680B is 9.99992186 Hz. The last 2 digits drifts. Maybe is fluke PM6680B, i know that is not calibrated. But when i press CHECK button on the frequency counter i see stable 10MHz reference clock.
I know this is an old video but I'd like to see more videos that use PIC's. I just installed MPLAB X but I can't get the LED blinker that you wrote to compile. Obviously I don't have something set up right or have a typo.
It looks like you stopped making videos based on the age of the last one. I hope all is well with you and you can find some time to start making them again.
If you follow the traces on my FE-5680A you find the lock signal originates from the output of a quad current amp (via a 1k resistor) and the absolute max rated value of that Op Amp output is 30mA... so having either a transistor or PIC connecting is far more sensible that running an LED direct. +1 for using a simple PIC.
Is the noise on the 1PPS output not by any chance 10Mhz?
What i did to clean it up was using a simple schmitt trigger gate at the 1PPS output.
Also to encase this rubidium i used a simple red/green led that is red when heating up and green when ready, and using a 555 as a pulse stretcher to flash a separate yellow led, it also reliable triggers on the 1microsecond pulse so it is always in sync with the 1PPS output.
My guess to that jitter on the 1pps signal while warmup is defintately crosstalk from the micro. On the video it looks like, that the pulse you generate on the led is visible on the scope. No suprise, interrupts wont work.
Video distribution amp is on its way. Now just need to find an affordable Rubidium Std and power supply. Thanks for sharing this with us.
Your welcome ArtR0001 good luck with the build...Gerry
On the subject of oscilloscope triggering - your unstable trigger on the 1PPS was due to the scope measuring too low a frequency on the trigger input (i.e. the output of its threshold detector).
If you go into the Trigger Mode menu and set Mode to "Normal" (not Auto) then you should be able to get a stable trigger on the 1PPS without having to keep hitting Single Sweep.
Thank you. Gerry
Does the white not also reflect IR?
The PIC has a hardware bug that causes problems if you write the to the port often.
So if you are in a tight loop writing the output, it will glitch on the other pins..
I used many hours bugging before I found out about it.
I'm not able to find the page that informed me about it right now.
Thanks Connor
Gerry - any recommendation(s) for a supplier of the buck regulators you used? Thought it would be a good idea to have a supply of them around for projects as you have done. See lots of them on eBay of course but always nice to have a reliable supplier.
Dino
Gerry - your videos are one of my favorites...you tell the story well. Will finally bite the bullet and get a Rb standard to replace my Z3801A GPS unit. There is a seller on eBay that has the FE version with 10 MHz output via a SMA connector which I think I'll try. What kind of hardware PIC programmer are you using with your Mac via MPLab?
Dino KL0S
It's not a question of it reflecting IR, it's a question of how much IR it emits - that's what your thermometer is looking for.
An ideal blackbody has an emissivity of 1.0. In practice, the closest you can get to this is carbon black (lampblack), which is e=0.98. Brushed aluminium is around e=0.2.
Your IR thermometer is probably assuming e=0.98, meaning it'll read low if you try to take a reading from a warm metal surface.
Your goal is to get the emissivity as close as possible to 0.98.
Have a look at the supporting blog page for the first video (in the link in the video description), I drew a rough sketch of the video channel. I would find something more current though, the CLC409 is obsolete. Not worried about the complainers, they are almost entertaining in their own right :) Thanks for your comments. Gerry
Yes its a consideration, Microcontrollers could be out of reach for some because of the points you mention. Gerry
Hi Gerry,
Enjoy your channel and 53131A OCXO add on board.
One thing I've found with EBAY OCXO's. While it is true that crystals are better after aging the ones I've bought that failed had other issues. One the oven would cut out after 10 or 12 hours, another one's internal reference voltage had bad drift and another one could not change the freq through the tuning pin.
These are still a great bargain and most times sellers will send you another one if the one they sent was bad. My favorites are the MV89A from Morion and the Milliren260. both are double ovens and the Milliren is claimed to be close to rubidium standards which seems to bear out compared to my rubidium standard.
Rob
+Hornbill
The worst ones I've found are the Milliren 260's. That figures, they go for ~$1000.00 each so I imagine companies aren't too eager to replace them unless they have to. If you look on eBay the dates are around 2000. It's the "LDC" line; 990319-1 would be 1999, March, 19. The number after the dash is the order it was built on the specific day. I checked with Milliren, the MBTF of those OCXO's is about 17 years. I had two out of 4 bad in the batch I bought over a couple year period, most were just out of MTBF. I do have one running for about 3 or 4 years now with no problems however and when they work they are spectacular, right up there with my rubidium.
I guess it comes down to why they were removed from service.
Rob
I agree with Gerry. Hobby electronics is about enjoying making things not about doing everything by industry standards. If everyone followed best practice then there would be a lot less fun in experimentation. If Gerry wants to use a pic then I see no issue, he could have used a 555 or even just a simple RGB LED. You could probably even use an asic if you wanted to. Gerry is right, you don't have to follow industry standard if you don't want to. Do what you want to get the job done I say.
Hi Gery,
I totaly agree with you, useing a pic controller is the most elegant way to make the puls for the LED, don't mind what others say.
Hermann Kovert thanks for the comment - yes I do ignore negative comments, if I did not I would give up :) Most people complain because its beyond reach for them (knowhow, tools, experience etc) and I understand that. I like using micro controllers, even for trivial tasks :)
+Gerry Sweeney I'm late to the party, but here goes... I'm among that group who lack the know-how and tools to use a micro-controller for this purpose, but I would still argue that it's unnecessary. I suppose the "heartbeat" is entertaining, but what does it tell you about operating status that the solid LED didn't? I would stick with a 2-state indicator, and propose the following 3-component solution. Swap the green LED for a red/green bi-color LED and put a 330-ohm resistor on its cathode to ground (~10mA LED current). Use a jelly-bean TO-92 P-channel MOSFET to drive the green anode (5V on source, LED anode on drain). Connect the red LED anode and the P-channel's gate to RBS_RDY. When power is on and the RBS hasn't yet locked (RBS_RDY = 5V), the LED will be red, and when the RBS locks (RBS_RDY = 0V), the LED switches to green. No micro-controller or software is required, the colors are intuitive as to operating status, no hours are spent fighting with PIC edge interrupts, and the BoM cost is lower. // Schematic via Dropbox: db.tt/4FJRaR6z // I don't expect to change your mind; I'm just offering an alternative for viewers who, like me, don't code. Great project, nicely executed -- cheers!
+FlyingShotsman the heartbeat tells me that the RBS is putting out a 1 second pulse that at a glance I can see, its just another level of confidence, if the LED is pulsing one per second you can be pretty sure that the 10Mhz output is also present, if suddenly the LED started to pulse more slowly, or faster than you might expect then that could indicate a problem that a solid LED would not show you.
I understand that some people don't code, what I was aiming to show in (part of) this video is actually there is not really that much to it. Of course its only suggestion, for me personally its easy because I have put the time in to learn how to use the micro-controllers.
+Gerry Sweeney
My take.
There's a great book out that anyone and everyone interested in electronics should get. It's called "The Art of Electronics". You see that word "Art", well that's exactly what electronics is. Several engineers can come up with a solution to accomplish a specific task and each one can do it differently. I can use uC's or I can light an LED with a 555 timer. Personally I use a uC. It's quicker (at least for me!) and it uses less parts. I don't need a tri-color LED and I don't need to stock a 555 timer. In fact I have a pile of 555 timers from the 90's that I never use. I can program an ATTINY11 that costs me 37 cents (100 pieces) and use less parts in the circuit to light that LED and make it dance any way I want right down to PWM driven by an encoder. It's hard to beat, but someone might want to use a different part or circuit. Remember it's an "Art" and because someone does it differently than I do doesn't mean my way is wrong, it's just what works for me or you.
I have a lot of magazine articles behind me and there's always someone who says you did it wrong, they never seem to do YT videos or write articles though. So Gerry, keep doing your art the way it works for you, it works fine for me too!
Rob
+mosfet500 cheers for your comments Rob I will :)
That was my line of thinking. So I built a 555 monostable to generate a 300ms pulse from the 1uS pulse, and still I saw the same behaviour. I also ran the micro at 20Mhz to try that - but with all of that - still no cigar....Gerry
Thanks Mark.
Aluminum sheet emissivity is about 0.1 so IR thermometer with standard e=0.95 will not work. You can stick tape and measure it or change emissivity.
Hi Dino, the first lot I ordered a couple of years ago and I can't remember who the seller was. The other day I ordered 20 more from a seller called "ynaan" in Hong Kong, not used them before but feedback on 250,000 sales is 99.2% so cannot be too bad. Mostly they are the cheapest I found at £8.05 for 10pcs which is less per unit than I can buy the LM2596 for! Gerry
Ah yes, so it does, fantastic, thanks for the pointer, thats helpful....Gerry
(Continued --)
The better IR thermometers allow you to adjust the emissivity to correct for this without needing to paint or Tipp-ex the heatsink. Sadly, finding out the correct emissivity constant for a given material involves either looking it up in a table (and hoping it's right) or measuring with a contact temperature probe and calculating it based on the measurement error.
I would contact the manufactor of the product to ask what is best if the datasheet do not explain it.
I guess your ringing is caused by impedance miss match. Use coax and provide a local ground return and that'll almost certainly fix it.
IR thermometers don't work properly on reflective surfaces.
Basically, since the emissivity and reflectivity are inversely correlated, and most IR thermometers are calibrated for a fairly high emissivity. As such, when you're measuring a reflective surface, the thermometer will read falsely low.
The easy way to get an accurate reading of a reflective surface is to stick something non-reflective on it. Black electrical tape works well. Let it thermally equalize, and then measure the tape.
Yes I agree with you. i like micro controllers, they are easy and cheap, why would I want to mess about with building a 555 circuit and then having to be stuck with the single function. The cost of the 555 + the electrolytic cap, cap for pin 5 will probably cost more than a PIC12F675...and even if it does not, as you say, time is important. I guess the only problem is, PIC's are inaccessible to you if you are not inclined to use micros or don't have the tools or programming capability.
Glad you find the video's useful. Yes an FPGA would be funny to control an LED, that would be people going :) Gerry
Hi Tom, no worries, I am not bothered at all, I just find it entertaining if I am honest :) Gerry
Read this about I/O on the PIC 16FXXX series.
cont.. buy a £5 atmega or pic, it still works out cheaper and easier than piss balling about with 555's , adc's, dac's and all the support components.
I would recommend all hobbiers and beginners to start with a big 28 pin mc for there prototypes and one offs so you can expand on them. If you find you need to mass produce, that's the time you hone down the Bom and boards.
You are also completely right about getting a good case for a permy project.
Thanks for the down to earth videos
Yeah I would have probably considered a 6-pin device but a) I did not have any and b) I was not making a PCB. Its amazing though that a 6-pin micro-controller can be purchased for just 40 cents in one off qty :) Great little devices indeed.
Yup!
Read this about I/O problems on the pic 16FXXX series
RUclips will not allow me to post a URL, so search for Read Modify Write Problem and microe
Yes so true...
Yes I am a bog fan of using them too - I guess its probably hard for people that are less comfortable with using micro controllers. Gerry
Oh yes, I see what you mean. Stick some masking tape on and it reads about 35 degrees C, that seems more like it. Thanks for the tip. Gerry
Sorry english not my language :). You can try google "emissivity table" to look how different materials emit IR.
Your "noise/ringing" looks to all be related to poor layout and/or probe technique... We could help if we saw the details of your setup.
Yeah, me either...:)
I'm not expert in PICs, but I suspect that external interrupt is not working because it's synchronized with processor clock. And your PIC is probably running on 0,5-1Mhz internal oscillator, so with 1us pulse you just get race condition and random behavior :)
Please check your Facebook also
LoL RUclips is comments are laggy as fudge xD
I can _guarantee_ you that if you had used a 555 timer there would have been several people that complained because you didn't use a microcontroller. In other words, you can't win. :)
The RUclips comments system absolute shite!
Yes I am a bog fan of using them too - I guess its probably hard for people that are less comfortable with using micro controllers. Gerry