Huh... simplest possible realization of delta-sigma ADC basically, neat. I once tried doing a delta-sigma ADC on a breadboard, but I overdid it with a separate comparator and an active integrator circuit with an op-amp. It never struck me to just use a passive RC circuit as the integrator and flip-flop input threshold as the comparator. I like it :)
@Jeri Ellsworth Instead of directly running the Q output to the audio, what if you ran it through an array of 128 bit shift registers, say the CD4517BE. At the end of the array of CD4517s then take the audio and run it through the low pass, but mix the original audio back with the delayed signal of the 4517s. You would clock the 4517s with the same clock, but probably around 44.1khz Would this not make a crude digital delay? Once my other project is done I will experiment with this, as I have 10 CD4517s, it would happen!
I think this is delta sigma ADC. I have worked on delta sigma ADCs. You can say that the D flip-flop is used as “quantiser”. The 10k resistor is essentially summing the negated quantiser output (q-bar) and the input voltage. I.e. it’s doing (vin- quantized_vout). Then the combination of 10k and 0.1 capacitor is making an integrator.(low pass filter) So, is first order delta sigma modulator. If you show the FFT of q output, you should see the noise shaping and it should show a slope of 20db/decade
Excellent. Interesting how, as the sample rate drops, aliasing effects take hold, and make the output less accurate but rather aesthetically interesting.
thanks soooo much Jeri: I've seen specs (in audio equipment) about a 1 bit a/d and always wondered how it worked. I teach and experiment with microcontrollers (and other electronics) so this video is very valuable to me. Thank-you, Mark.
I haven't sweep it too far, because the distortion depends on the sample frequency and the low pass filters. The designer would need to tweak the values for their application.
This video popped up as new on my TH-cam subscription. Now, I see that it is several years old. In any case, I enjoyed the short circuit ( your pun, not mine :-) ) demonstration and explanation. Good job. FWIW, study Nyquist to advance to a higher level of understanding. A quick Google search will get you started. I always enjoy your videos.
I wonder how close this is to the 1-bit ADC processor (Direct Stream Digital) that Sony uses for their Super Audio CD. I'll have to find some more information about Sony's processor but I'm sure it won't be as clearly explained as this.
Jeri, Very interesting video - I don't know if I have anything to offer to you, but I am an Engineer with a large German test equipment company. Your videos are of interest to myself and my colleagues at work. Please keep up the good work :)
This is used in all audio codec. Iphone for example. The actual bit is function of your oversampling rate. Iphone has 24bit deltasigma codec supplied by cirrus logic.
They do use flash converters, but typically only 1 or 2 bits or resolution. The SNR is so low, that using extra bits only samples the noise. I've seen specs like 1-2 bits and 8 Gsps. They they correlate the bit streams to get phase shifts between the antennas.
Great to have you back, you frequently seem to post something on what I'm learning in school right now, and please don't Dave Jones up your vids, one is good enough ;)
I really like the partial intrusion of your face. (You *do* have a very pretty face, Jeri.) Whether you planned the vid that way or not, it's still a catchy way of including the announcer.
I can't find did anyone ask this before me, but I just find my old portable CD player and it has sticker on it that says "1 bit DAC"?!?! How audio CD can sound so good with only 1 bit DAC? Audio CD is 16bit 44.1KHz as far as I know. Please explain this to us if you know.
It's 1 bit because the PWM (pulse width modulated) signal (the green one in the middle) is a very fast (hundreds of kHz) train of either on or off states. In other words it is digital, which makes it resistant to noise and have the ability to be loss-lessly modified by digital filters or computers. Also, as opposed to the PCM encoded CD signal (which also is digital), the 1 bit signal is very simply and cheaply converted to analog or from analog (both of which is shown in the video). So you can think of the 1 bit signal as a different, but equivalent (e.g. to a CD) digital encoding of the sound. In your CD player, the 16 bit 44.1kHz is digitally (algorithmically) converted to a one-bit train, which then is able to be very simply and efficiently converted into analog as above.
misium actulay the explanation to that is when u use 1 bit dac u sample way above nyquist rate and then u use decimators after decimators to lower the sample rate and then eantualy u use an lpf that cause the quantization noise to be reduced alot! for exmaple 1 bit noise therotcilay can have 16bit effective enob and 16 bid adc can have about max 14bit estimated enob who would u choose? consider 1 bad dac is extreamly easy to design.
I read the Wiki on Delta-Sigma Modulation. So its realy NOT PWM as mentioned. Its more like "Pulse Density modulation" or "Pulse Frequency modulation" which makes more sense. still a realy interesting method for sampling and synthesizing signals.
The only problem is is that you have to have some very fast switching power mosfets at a speed of 3GHz for CD quality sound (44100 x 65536). There are noise-shaping and dithering algorithms that can be applied to lower that quite a bit(
How do you do a sine wave in binary? It would have to be hexadecimal. Wish I paid attention when I was being taught the use of these measurement devices.
I've rewachted the video and I think I get it. There will be a requirement that the signal has an amplitude > 0.7 to guarantee that the toggling would skip. Fluctuations of a lower amplitude could be missed as the toggling might not skip resulting in loss of data?
Thanks for the video! But i can't understand, why flip-flop's nQ signal changes sometimes on falling edge and sometimes on rising edge of CLK as shown on your diagram(1:19)?
Quite brilliant. Is it possible to digitize a certain frequency band (let us say 500 Hz up to 10 Khz) with this circuit and/or how critical is the filter at the output when a whole frequency band has to be restored from a digital signal to a sine wave. Does the restore filter have to be adapted for all the parts of the audio range?
Jeri thanks for taking a few minutes to talk with me at NYC Maker Fair. Have you tried using a photo cell other than a pot for the input? It could come up with interesting feedback with music as the input and a stepper circuit on the output with some party lights.
Good to see you back and cheers for the interesting video. I know my next question is going to be a long shot, but is there a chance us to find out what kind of projects you are working on at Valve?
Huh... simplest possible realization of delta-sigma ADC basically, neat. I once tried doing a delta-sigma ADC on a breadboard, but I overdid it with a separate comparator and an active integrator circuit with an op-amp. It never struck me to just use a passive RC circuit as the integrator and flip-flop input threshold as the comparator. I like it :)
I've never gotten bored while listening to you. Talk as much as you like!
Your short circuits videos are always interesting. Thanks!
It will sample amplitude also. You can see that it's reconstructing the low points in the sine wave.
@Jeri Ellsworth
Instead of directly running the Q output to the audio, what if you ran it through an array of 128 bit shift registers, say the CD4517BE.
At the end of the array of CD4517s then take the audio and run it through the low pass, but mix the original audio back with the delayed signal of the 4517s.
You would clock the 4517s with the same clock, but probably around 44.1khz
Would this not make a crude digital delay?
Once my other project is done I will experiment with this, as I have 10 CD4517s, it would happen!
I think this is delta sigma ADC.
I have worked on delta sigma ADCs.
You can say that the D flip-flop is used as “quantiser”.
The 10k resistor is essentially summing the negated quantiser output (q-bar) and the input voltage. I.e. it’s doing (vin- quantized_vout). Then the combination of 10k and 0.1 capacitor is making an integrator.(low pass filter)
So, is first order delta sigma modulator.
If you show the FFT of q output, you should see the noise shaping and it should show a slope of 20db/decade
The new place is great. Finally enough room to stretch out a bit.
Excellent. Interesting how, as the sample rate drops, aliasing effects take hold, and make the output less accurate but rather aesthetically interesting.
Congrats on being hired for Valve Software! =D Can't wait to see what you guys come up with and awesome video like always!
There is a frequency shift, but it's not linear like a FM transmitter.
Thanks Jeri. I like the way your explain how components and circuits work.
thanks soooo much Jeri: I've seen specs (in audio equipment) about a 1 bit a/d and always wondered how it worked. I teach and experiment with microcontrollers (and other electronics) so this video is very valuable to me. Thank-you, Mark.
A FANTASTIC EXAMPLE OF HIGH AND LOW DEFINITION DIGITAL SOUND!
Interesting, it looks like the simplest model of Delta Sigma ADC modulator.
I haven't sweep it too far, because the distortion depends on the sample frequency and the low pass filters. The designer would need to tweak the values for their application.
I think you should have shown an fft for the final output in place of or in addition to the waveform output.
It's always nice to see how much fun one can have with parts worth only a couple of cents.
It's a MDO4104-6 and it's the companies. I have an old B&W Tek at home.
Brilliant advances you have made Jeri Ellsworth . Great vid !
Well I just learned a ton. Thanks! I hope you like your new digs. Sounds like you have lots of room.
This video popped up as new on my TH-cam subscription. Now, I see that it is several years old. In any case, I enjoyed the short circuit ( your pun, not mine :-) ) demonstration and explanation. Good job. FWIW, study Nyquist to advance to a higher level of understanding. A quick Google search will get you started. I always enjoy your videos.
I like the lower sample rate. Reminds me of the 8bit days.
This I showed the reconstruction of the signal in the video for the demonstration, but you's feed the q output to a MCU or FPGA.
This one wasn't so short. I originally tried to make them 3 minutes long.
I wonder how close this is to the 1-bit ADC processor (Direct Stream Digital) that Sony uses for their Super Audio CD. I'll have to find some more information about Sony's processor but I'm sure it won't be as clearly explained as this.
Great seeing a new video! Don't let the haters hate on the uploading of old fmcg vids, I've enjoyed going back through them.
I've seen one bit ADCs before, but always with an explicit voltage comparator. That's slick.
@Zappyguy111 it really stands out when the sample rate is lowered.
It's a nice unit.
If you zoom in on the Q output, do you see any occasional metastability events?
This is a good demonstration for the elimination of "key clicks" using morse code (CW) on home-brew ham radio equipment.
Jeri, Very interesting video - I don't know if I have anything to offer to you, but I am an Engineer with a large German test equipment company. Your videos are of interest to myself and my colleagues at work. Please keep up the good work :)
Interesting circuit and very simple to use.
Hmm, on second thought, no one should ignore the benefits of editing video.
Yep. It's pumping down right now.
I think a guitar effect box would be the killer app for this. :)
This is used in all audio codec. Iphone for example. The actual bit is function of your oversampling rate. Iphone has 24bit deltasigma codec supplied by cirrus logic.
That casual "electron microscope in the garage" line never fails to make me laugh.
Thanks. We have cool things in the works.
Reminds me of a class-D amp. Nice!
Cool. Wonder why they choose that over flash converters?
I'm sure using a comparator set 50% the rails would yield better results. Might be worth a test.
We have a new hardware R&D dept.
Likewise.
I haven't tried anything, but music, which worked well.
They do use flash converters, but typically only 1 or 2 bits or resolution. The SNR is so low, that using extra bits only samples the noise. I've seen specs like 1-2 bits and 8 Gsps. They they correlate the bit streams to get phase shifts between the antennas.
sounds like many good videos to come...
thanks a lot
Nice demo of a delta sigma ADC
Great to have you back, you frequently seem to post something on what I'm learning in school right now, and please don't Dave Jones up your vids, one is good enough ;)
Ow man. That was one of your cooler projects. Well counting the ones i seen on your channel anyway :)
Nice! Very high sample rate, 1-bit ADCs are commonly used in astronomical radio interferometry. This gives me some ideas.
I moved most of them up. I still have a few in PDX that need to be moved.
Nice video. Looks like this would have some music synthesis applications!
Yep. It drops of fast when the sample rate gets low.
I really like the partial intrusion of your face. (You *do* have a very pretty face, Jeri.) Whether you planned the vid that way or not, it's still a catchy way of including the announcer.
I can't find did anyone ask this before me, but I just find my old portable CD player and it has sticker on it that says "1 bit DAC"?!?!
How audio CD can sound so good with only 1 bit DAC? Audio CD is 16bit 44.1KHz as far as I know.
Please explain this to us if you know.
It's 1 bit because the PWM (pulse width modulated) signal (the green one in the middle) is a very fast (hundreds of kHz) train of either on or off states. In other words it is digital, which makes it resistant to noise and have the ability to be loss-lessly modified by digital filters or computers. Also, as opposed to the PCM encoded CD signal (which also is digital), the 1 bit signal is very simply and cheaply converted to analog or from analog (both of which is shown in the video).
So you can think of the 1 bit signal as a different, but equivalent (e.g. to a CD) digital encoding of the sound.
In your CD player, the 16 bit 44.1kHz is digitally (algorithmically) converted to a one-bit train, which then is able to be very simply and efficiently converted into analog as above.
misium actulay the explanation to that is when u use 1 bit dac u sample way above nyquist rate and then u use decimators after decimators to lower the sample rate and then eantualy u use an lpf that cause the quantization noise to be reduced alot!
for exmaple 1 bit noise therotcilay can have 16bit effective enob
and 16 bid adc can have about max 14bit estimated enob
who would u choose?
consider 1 bad dac is extreamly easy to design.
The 6 rubbed off
It's not really delta-sigma either, so it's somewhere between.
Please explain about Delta-Zigma conversion briefly....!!!
So you made a class D unity gain amplifier? A class D buffer? :D
Wow. You are like a celebrity to me. I hope I meet you someday. Hope you brought your umbrella!
wow, almost sounded like an Oboe there...circuitgirl is becoming synthgirl. Love your videos...your amazing.
Oh wow! You've just invented DSD audio!
Cool, I just watched that FMSG episode and was wondering what he was talking about.
Thanks.
jerri ellsworthe: has an electron microscope in her garage, is more interested in teaching us about circuits.
you're awesome!
Sounds like it also has the makings of a nice guitar effects box (though not nearly as grandiose a use case as astronomical radio interferometry).
This is really neat- I'm gonna have to go try it later
No problem.
congrats on your job with valve!
I read the Wiki on Delta-Sigma Modulation. So its realy NOT PWM as mentioned. Its more like "Pulse Density modulation" or "Pulse Frequency modulation" which makes more sense. still a realy interesting method for sampling and synthesizing signals.
The only problem is is that you have to have some very fast switching power mosfets at a speed of 3GHz for CD quality sound (44100 x 65536). There are noise-shaping and dithering algorithms that can be applied to lower that quite a bit(
This is a master/slave flop, so it's unlikely it will make it past the first internal stage.
What is the amplitude of the input sine wave?
Awesome as always Jeri, keep it up !
Hi Jeri, Nicely explained. Thank you.....!
Please explain about Delta-Zigma conversion briefly....!!!
How do you do a sine wave in binary? It would have to be hexadecimal. Wish I paid attention when I was being taught the use of these measurement devices.
Watch out you will be adding opamps to your next design.
I was thinking today that a window comparator would make a pretty simple one bit a to d converter.
I've rewachted the video and I think I get it. There will be a requirement that the signal has an amplitude > 0.7 to guarantee that the toggling would skip. Fluctuations of a lower amplitude could be missed as the toggling might not skip resulting in loss of data?
This is simplistically cool!
Seems almost like a delta-sigma but incredibly simple
I finished the cage and then the person I built it for ran out of money and sold it.
Best of luck on your new job!
That in itself is awesomeness. None the less I simply admire her. She's intelligent, beautiful AND (thanks to you) knows what she wants. ;)
Make a video when you get the guitar pedal going.
Awww!😍 If you were my teacher then I would never missed a class😍😘
It could be use to control the speed of a DC motor but the response would very un-linear. Try breadboarding it and see what happens?
That is a possible application.
Thanks for the video! But i can't understand, why flip-flop's nQ signal changes sometimes on falling edge and sometimes on rising edge of CLK as shown on your diagram(1:19)?
Quite brilliant. Is it possible to digitize a certain frequency band (let us say 500 Hz up to 10 Khz) with this circuit and/or how critical is the filter at the output when a whole frequency band has to be restored from a digital signal to a sine wave. Does the restore filter have to be adapted for all the parts of the audio range?
it wasn't a serious question, merely trying to seed the concept of electric cars in Jeri's mind : )
Jeri thanks for taking a few minutes to talk with me at NYC Maker Fair. Have you tried using a photo cell other than a pot for the input? It could come up with interesting feedback with music as the input and a stepper circuit on the output with some party lights.
Simple ADC (mcu) controlled capacitor Touch Sensor circuit ?
Thanks! Very easy to follow you.
I'm late to the game, but this was very helpful. Thank you.
thanks for the CLEAR explanation of this underdog circuit. also congrats with the new job! did you say you had an electron microscope?!
Missed you ♡
Good to see you back and cheers for the interesting video. I know my next question is going to be a long shot, but is there a chance us to find out what kind of projects you are working on at Valve?
Whats abaout frequency Guzheng instrumen music up down?because Device do not 4(12356 .....),Thanks info video
Nifty minimalist circuit! Tell us about the electron microscope, please. Playing with one of those beasties was a childhood dream.
I'm still doing hardware for them.
Check out, "The Signal Path" #32
Simple and subtle sigma-delta, I like it! Did you sweep it? What is it's frequency response like?