Simply brilliant Shahriar, thanks to your explanation of 1-PAM and the logical progression thru 8-PAM, I finally really understand the concept of complex pulse amplitude modulation. I thank you for your excellent teaching style and visual aids both graphic and by building the actual circuit and conducting experiments, helps build the idea and concept into understanding. Two thumbs up young man, thank you.
I'm an electrical engineering student. I absolutely love these videos. So much practical knowledge in these. Of all the junk on youtube these videos are spectacular. Keep them coming. Pretty soon you'll be making money doing this kind of thing.
Shahryar.....now that i can see you my iranian fellows found success and glory im really happy for all of you and should inform u guys that you are the pride for all persians . you raised our national flag. So thank you
Your an awesome teacher. I'm going to use this in my AP physics class this year. I'm an Air Force veteran telelcom professional. I now teach in Thailand! Gonna have to convince the school to get a spectrum analyer and some kits from Sparkfun. I'll be teaching the senior class this year! Teacher Tylor on Facebook.
You're so good at teaching. Please consider doing tutorial like videos too. I get that many people like tear down videos. But you also have another attentive audience
The BlinkM uses a 120Hz PWM square wave. I used an RC filter with a 3dB bandwidth of only a few Hz to clean that up. In theory, an active filter can clean up the signal much better (higher roll of factor) which would allow for faster data rate. Great question by the way. Thank you.
Hi Shahriar. I was pleasantly surprised to find your new video today. I just recently discovered your channel and I am really enjoying your great videos. As always, I found that the balance between theory and practical experimentation is just right. Thank you for an other great video and I am looking forward to the next one.
You are right, it is possible to stay in the PWM domain. But measuring exact duty cycle on three different pins at the same time was not so straight forward. So I decided to go to analog.
I think this is a really cool video. Concepts like constellation is not teached at our University of Applied Sciences, not even on the telecommunication courses. Yet they are very interesting. Well done explaining it!
Interesting idea. Some ideas after looking at this a few minutes (I'm an EE ). a)The emission spectrum (data value on x axis, intensity on y axis) of the emitter and the detection spectrum of the receiver should be mapped and used to remap the values sent to the emitter. b)Something similar needs to be done for the fiber to evaluate it's effects on R, G and B intensity. c)The RC filter is introducing very large phase shifts in the data. This fundamentally introduces a time domain error in the detector. I suggest a direct PWM demodulation approach to reduce this error. d)I would investigate using larger spectrum separation to improve crosstalk between R, G and B channels i e. use of Infrared and Ultraviolet LEDS. e)Consider abandoning using multiple amplitude levels as part of the modulation scheme. Multiple different color LEDS, each transmitting NRZI data with full on, or full off, could be more easily detected and demodulated. f)A tight cluster of standard Red, Green and Blue, LEDS can be used as light detectors. Lastly, you're probably already aware of this but multi-wavelength transmission over optical fiber. GREAT CONTENT. JUST SUBSCRIBED.
this is a great video. the QAM showed at the start with the 2 dimensional constellation is for Phase. a single sine wave sent down a pipe varying by amplitude and phase. its cool to see the recovered clock signal. decoding is so much harder than encoding.
Very very nice! I can see how this kind of ingenuity would spark the interest of students! You would make a great Teacher (If you aren't already one). At the end I expected that you would increase the bitrate and comment on the growth of the regions or increase bits per symbol and show if the doubling of dots affects data integrity. Keep it up!!!
The setup is cheap, but you can use ot to show really advanced concepts. You could do probabilistic shaping of the constellation, coded modulation, you could use it to demonstrate shaping gain due to increased number of dimensions. This is really cool!
Another great video. Thank you. What was the major part(s) that was limiting the speed of data transmission? Was it just the color sensor or a combination of things?
thank you for the great video! I really enjoyed the balance between theory and practice which is often lacking in my university classes. I hope you can do a similar RF example in the future.
This is very cool, have you tried using Manchester style encoding scheme to create a self-clocking signal? You could try increasing the bit rate and the bits/symbol until you came across the limit. Further, you could try using an algorithm to scan through the constellation repeatedly and determine which transitions are clearly distinguished and which ones are less reliable. Maybe you can get more resolution in some regions of the constellation. Would this be effective in a single mode fibre?
Great presentation -- as usual! Question: Why do the three digital color signals need to go through an analog step (RC network, ADC)? Why not just feed the WDM pulse streams directly into to the microcontroller?
The LED brightness is set by pulse width modulation. Normally you'd filter the PWM signal so that the LED intensity is constant for the duration of a symbol. That way, you're using less optical bandwidth. That's not possible here without modifying the RGB module, so Shahriar does the filtering after the photodiode. One benefit of doing this is that he can saturate the receiver and get a digital signal.
@@ignaciomartinez6245 ahh right so the LEDs are super fast PWMing - and then he gets the fast PWM flashes on the other end & low passes it to get the analogue level back. the low pass is needed to remove the high frequency noise but also to smooth out the high frequency PWM into a coherent stable 1/4 level, i figure.
The synchronization symbol is not the same as a "0". So if I were to send 0-0-0, the transmitter would send S-0-S-0-S-0. That is a 50% overhead just to get a CDR! ;)
I was about to ask why filter the receiver rather than the transmitter, but then I remembered you wanted to use the off the shelf PWM gadget. Did you test the frequency response of the sensor at all? Great work on the video, would love to see a follow up on RF modulation ;)
That was a fascinating video! You may want to experiment with data rates using compression to increase the entropy per symbol. That could make your test jig a bit faster (depending upon the type of data transmitted, of course). The compression being accomplished on the PC.
I know that its an old post, but the rgb sensor is out of stock, gone, not being produed anymore.. Anyone knows of a replacement without a digital section ? The speed is just below of what i need it to be...
Interesting video, thanks a lot. I have one question, why is the data rate so slow? If you are using multi-wavelength, multi-level signals shouldn't you get high speeds.
Was it said that the RC filters were why data had to be sent so slow? Why not just send on or off on each color band and go faster. Might be able to up the transmission rate without more expensive parts.
using PWM with a low pass filter is something ive seen done in a lot of places to make analog wave forms. it'd be cool to use a soundcard with a light on the end of it and then encode data in the single channel but with phase as well. i never quite figured out how to lock onto the phase without some kind of extra encoding to put it there like manchester or something.
on that scope thing the lines that are at different heights are go up at different angles, if the computer can detect this you can use this as a value to detect this angle you could use this to get a colour as soon as the value is not 0, however the noise would overlap more at this point so i dont know if that would have more of an effect at this point.
I am trying to send a analog signal of atleast 1.5-2Mhz, 2 digital signals of 5v and 15v signals over a single fiber optic cable without using a computer. Meaning the input would be coaxial then sending it over fiber.
great why not going reverse, making the rgb led as receiver, get rid of the fiber, point the sensor to a color paper, and then make th rgb led display it
Simply brilliant Shahriar, thanks to your explanation of 1-PAM and the logical progression thru 8-PAM, I finally really understand the concept of complex pulse amplitude modulation. I thank you for your excellent teaching style and visual aids both graphic and by building the actual circuit and conducting experiments, helps build the idea and concept into understanding. Two thumbs up young man, thank you.
I'm an electrical engineering student. I absolutely love these videos. So much practical knowledge in these. Of all the junk on youtube these videos are spectacular. Keep them coming. Pretty soon you'll be making money doing this kind of thing.
Shahryar.....now that i can see you my iranian fellows found success and glory im really happy for all of you and should inform u guys that you are the pride for all persians . you raised our national flag. So thank you
Your level of videos are way above the level on you tube i have seen so far. Fantastic.
Your an awesome teacher. I'm going to use this in my AP physics class this year. I'm an Air Force veteran telelcom professional. I now teach in Thailand! Gonna have to convince the school to get a spectrum analyer and some kits from Sparkfun. I'll be teaching the senior class this year! Teacher Tylor on Facebook.
You're so good at teaching. Please consider doing tutorial like videos too. I get that many people like tear down videos. But you also have another attentive audience
You are better then most of my teachers!
Can u tell abstract of this project
@@akashom9031 could you elaborate your question a litle bit please? I don't understand it.
I love this channel. Beautiful, absolutely beautiful
Can u tell abstract of this project
The BlinkM uses a 120Hz PWM square wave. I used an RC filter with a 3dB bandwidth of only a few Hz to clean that up. In theory, an active filter can clean up the signal much better (higher roll of factor) which would allow for faster data rate. Great question by the way. Thank you.
Thank you for this experiment. I learn something new in every one of your videos!
Hi Shahriar. I was pleasantly surprised to find your new video today. I just recently discovered your channel and I am really enjoying your great videos. As always, I found that the balance between theory and practical experimentation is just right. Thank you for an other great video and I am looking forward to the next one.
You are right, it is possible to stay in the PWM domain. But measuring exact duty cycle on three different pins at the same time was not so straight forward. So I decided to go to analog.
I think this is a really cool video. Concepts like constellation is not teached at our University of Applied Sciences, not even on the telecommunication courses. Yet they are very interesting. Well done explaining it!
Interesting idea. Some ideas after looking at this a few minutes (I'm an EE ). a)The emission spectrum (data value on x axis, intensity on y axis) of the emitter and the detection spectrum of the receiver should be mapped and used to remap the values sent to the emitter. b)Something similar needs to be done for the fiber to evaluate it's effects on R, G and B intensity. c)The RC filter is introducing very large phase shifts in the data. This fundamentally introduces a time domain error in the detector. I suggest a direct PWM demodulation approach to reduce this error. d)I would investigate using larger spectrum separation to improve crosstalk between R, G and B channels i e. use of Infrared and Ultraviolet LEDS. e)Consider abandoning using multiple amplitude levels as part of the modulation scheme. Multiple different color LEDS, each transmitting NRZI data with full on, or full off, could be more easily detected and demodulated. f)A tight cluster of standard Red, Green and Blue, LEDS can be used as light detectors. Lastly, you're probably already aware of this but multi-wavelength transmission over optical fiber. GREAT CONTENT. JUST SUBSCRIBED.
this is a great video. the QAM showed at the start with the 2 dimensional constellation is for Phase. a single sine wave sent down a pipe varying by amplitude and phase. its cool to see the recovered clock signal. decoding is so much harder than encoding.
Very cool experiment. All of my life I have always tried to use things in way they were NOT intended to be used. Kudos for doing just that!
Can u tell abstract of this project
Very very nice! I can see how this kind of ingenuity would spark the interest of students! You would make a great Teacher (If you aren't already one). At the end I expected that you would increase the bitrate and comment on the growth of the regions or increase bits per symbol and show if the doubling of dots affects data integrity. Keep it up!!!
The setup is cheap, but you can use ot to show really advanced concepts. You could do probabilistic shaping of the constellation, coded modulation, you could use it to demonstrate shaping gain due to increased number of dimensions. This is really cool!
This was awesome as usual. I would love to see a whole course on some of these topics where you dive into theory.
Great video as always... I realy like that you give the thoretical background about the topic you're going to speak.
Thanks :)
I really like how clean and neat is your lab. Nice videos! 1 + subscriber here from Brazil.
Impressed as always. Your effort really is appreciated thank you
Another great video. Thank you.
What was the major part(s) that was limiting the speed of data transmission? Was it just the color sensor or a combination of things?
Can u tell abstract of this project
Interesting concept, and well executed. Well done.
The bandwidth of the PD was greater than of the PWM frequency, so I didn't actually measure it beyond that.
thank you for the great video! I really enjoyed the balance between theory and practice which is often lacking in my university classes.
I hope you can do a similar RF example in the future.
This is very cool, have you tried using Manchester style encoding scheme to create a self-clocking signal? You could try increasing the bit rate and the bits/symbol until you came across the limit. Further, you could try using an algorithm to scan through the constellation repeatedly and determine which transitions are clearly distinguished and which ones are less reliable. Maybe you can get more resolution in some regions of the constellation. Would this be effective in a single mode fibre?
Great presentation -- as usual! Question: Why do the three digital color signals need to go through an analog step (RC network, ADC)? Why not just feed the WDM pulse streams directly into to the microcontroller?
The LED brightness is set by pulse width modulation. Normally you'd filter the PWM signal so that the LED intensity is constant for the duration of a symbol. That way, you're using less optical bandwidth.
That's not possible here without modifying the RGB module, so Shahriar does the filtering after the photodiode. One benefit of doing this is that he can saturate the receiver and get a digital signal.
@@ignaciomartinez6245 ahh right so the LEDs are super fast PWMing - and then he gets the fast PWM flashes on the other end & low passes it to get the analogue level back.
the low pass is needed to remove the high frequency noise but also to smooth out the high frequency PWM into a coherent stable 1/4 level, i figure.
The synchronization symbol is not the same as a "0". So if I were to send 0-0-0, the transmitter would send S-0-S-0-S-0. That is a 50% overhead just to get a CDR! ;)
I was about to ask why filter the receiver rather than the transmitter, but then I remembered you wanted to use the off the shelf PWM gadget. Did you test the frequency response of the sensor at all? Great work on the video, would love to see a follow up on RF modulation ;)
I understood more of this than a thought....excellent, more please ?
That was a fascinating video! You may want to experiment with data rates using compression to increase the entropy per symbol. That could make your test jig a bit faster (depending upon the type of data transmitted, of course). The compression being accomplished on the PC.
I know that its an old post, but the rgb sensor is out of stock, gone, not being produed anymore.. Anyone knows of a replacement without a digital section ? The speed is just below of what i need it to be...
Doesnt even need to have a build in gain..
SparkFun Spectral Sensor Breakout - AS7262 Visible has a similar spectral response
I really enjoy your videos,and learn loads about new (new to me) principles .Thank you
Great tutorial vide. File with code is not on site. Is that piece of wood holding optic cable palm tree wood?
Very nice explanation. I wish I have watched this during my communication class ✨👌
Can u tell abstract of this project
Interesting video, thanks a lot. I have one question, why is the data rate so slow? If you are using multi-wavelength, multi-level signals shouldn't you get high speeds.
I think it is fixed now.
Was it said that the RC filters were why data had to be sent so slow? Why not just send on or off on each color band and go faster. Might be able to up the transmission rate without more expensive parts.
These are golden videos.
using PWM with a low pass filter is something ive seen done in a lot of places to make analog wave forms. it'd be cool to use a soundcard with a light on the end of it and then encode data in the single channel but with phase as well. i never quite figured out how to lock onto the phase without some kind of extra encoding to put it there like manchester or something.
on that scope thing the lines that are at different heights are go up at different angles, if the computer can detect this you can use this as a value to detect this angle you could use this to get a colour as soon as the value is not 0, however the noise would overlap more at this point so i dont know if that would have more of an effect at this point.
the link on your site for the program code is not working !
You're making me choose a telecommunication option in my degree
I am trying to send a analog signal of atleast 1.5-2Mhz, 2 digital signals of 5v and 15v signals over a single fiber optic cable without using a computer. Meaning the input would be coaxial then sending it over fiber.
what happens to the clock recovery if the symbol 0-0-0 occurs in the tx data?
Can you please, do a video simar to this for wireless? It would be so useful for biginers.
Can anyone tell abstract of this project
it is a crime not to smash the like button under this video
This was enthralling, thank you very much for the great videos! My opinion, it was over before I wanted it to be :-(
Can u tell abstract of this project
fantastic. Almost like looking at a unit cell in xtalography
Thanks you alot was really eye opening
Can u tell abstract of this project
EEVblog is nothing compared to your channel.
great why not going reverse, making the rgb led as receiver, get rid of the fiber, point the sensor to a color paper, and then make th rgb led display it
I think LUCENT TECHNOLOGIES have done an Color PIPE-SPLITTER Chip for
PIPE Muxing. Color-Head-DataPackets. ?? let me know!
I like your videos, well done my man.
27:00 hopelessly slow... that's what happens when i try to do hw and stumble onto this channel
Learned a lot , thank you
excellent demonestration.
Believe it or not, it is a salt shaker made from bamboo! ;)
Beautiful
Thank you
Well done!
Superb!
tanks for this great video.
Wow ... awesome ....
Awesome!
Genius.
👍👍
wow!
You rock. I
Hey that is not bamboo! That is coconut wood!
passwords in 3d