jason9

modified 7 years ago

4 Bit Delta Sigma ADC

311

03:38:17

Is it correct that this is Delta Sigma? The four wires in the middle are the digital output. As you can see, when it is sent to an DAC and filtered it gives a very clean output even though it is only 4 bits. In fact, it produces no more than 1mV of noise even though it has an output of up to 7.5V.

published 7 years ago

nikisalli

7 years ago

Finally someone using my counter oscillator! Lol nice work

eekee

7 years ago

@nikisalli: I've used it! only in private circuits though, I think.

jason9

7 years ago

It’s very useful for high speed.

hurz

7 years ago

Whats the advantage over my 555 hack? http://everycircuit.com/circuit/5715732271988736

jason9

7 years ago

The 555 timer makes the circuit go at a crawl, while @nikisalli’s version preserves the speed of the circuit.

hurz

7 years ago

Right, for some reasons nikis version does not load the Simulator to much, even do the same maximum oscillation.

eekee

7 years ago

I somehow missed your 555 circuit, hurz.

eekee

7 years ago

The clock is 15MHz (on my computer), and the signal of course 20kHz. That's a 9.5 octave difference; huge! The filter is second order, about 12dB/octave. If it covers 8 octaves, then the clock frequency would be attenuated by 96dB; 63096 times. 9 octaves would make it 108dB; 251190x. If I understand right, that means you would have very little noise even if you only used a simple 4-bit ADC. Sorry, jason9!

eekee

7 years ago

Incidentally, niki's clock can be slowed down with a single (large) resistor between A and everything else.

jason9

7 years ago

What are you sorry for, @eekee? Also, how can his clock be slowed down with a large resistor between A and everything else? In this circuit, it runs the same as ever at 300ohms, but stops altogether at 330ohms.

eekee

7 years ago

Good question. :) I'm doing a bit of work with a non-sigma-delta ADC to see if i can get close to your 1mV noise figure. -- Odd, it works witth 20MΩ for me.

eekee

7 years ago

Ehh, I'm getting 204mVp-p (102mV amplitude) distortion. how do you measure noise?

eekee

7 years ago

http://everycircuit.com/circuit/4882597749719040

jason9

7 years ago

I measure noise with a high pass filter and if some of the signal still gets through I just estimate the amplitude of the noise, else I just look at the values on the scope.

jason9

7 years ago

I think it’s not working for me because of the way the circuit works or something is putting a load on the simulator and then it’s simulation lowers in quality. If it was just the oscillator alone then an RC network (possibly with the MOSFETs parasitic capacitance as the C) could slow it down, as I saw in your recently published circuit about fast clocks.

eekee

7 years ago

I see. I also made a version with a very tight notch filter, imitating THD measurement gear, but I used a simple LC notch filter. Those take time to settle, this one looked like it would take about an hour!

eekee

7 years ago

The resistor makes the oscilator output smaller, I didn't think about that when I told you. You need to take output from the A pin which is always a square wave and has huge current capacity. I hope that fixes it, anyway. :)

eekee

7 years ago

I improved the notch filter, it only wants 5-10 minutes now. I think it indicates about the same as the difference circuit; between 6V & 8V, min and max voltages go up & down in about the same pattern. http://everycircuit.com/circuit/5063866307575808

eekee

7 years ago

Okay, yours is definitely better than mine :) http://everycircuit.com/circuit/5313350744670208

jason9

7 years ago

Your Q is WAAAAY too high! You don’t need the 1uOhm resistor and it works just fine if the 40mOhm resistor is replaced with a 1ohm resistor and it lowers the response rate to just a few seconds instead of 10 minutes. Because of your handy little filter, I found that it works better if 7.5V were added to the op-amps output and it’s limits were changed to +-7.5V and the input was lowered by half a volt.

jason9

7 years ago

What I mean by adding 7.5V to the op amps output is adding a 7.5V volt source.

eekee

7 years ago

AC simulation gives up without the 1uR resistor, which is unrealistically low for inductor winding resistance as it is. :) Lowering the Q certainly improves response time, but will miss some distortion. the point is to filter out only the correct output.

jason9

7 years ago

Yes, but who’s willing to wait TEN MINUTES just to measure THD?

hurz

7 years ago

@jason9 and @eekee, to filter out the 20kHz signal and see whats left over as harmonics, you do not need a high Q for the notch. L and C are perfect, just make it for 1kOhm impedance and put a 1kOhm resistor like here http://everycircuit.com/circuit/5018442196582400 and you can see with 7Vpp input signal there is a problem short before one of the peaks. Anyway, tune input down to 6.5Vpp and the noise is within 2-3mVpp, not to bad.

add comment in editor