Comments on: Example Simulink model & scripts for continuous-time sigma-delta ADC

By: PoojanWagh

PoojanWagh — Tue, 16 Feb 2010 05:41:55 +0000

Hi, Daniyal. The ADC I have is a very specific type of ADC: a signma-delta ADC (and a continuous-time one at that). It is pretty complicated. If you’re building a controller, I’d suggest a Nyquist ADC. Sigma-Delta’s themselves include a feedback loop. Adding a feedback loop around the ADC amounts to having two interacting loops to stabilize. Not an easy task.

Anyway, I’ll attempt to answer your questions:
1. Mainly because it wasn’t germain to what I’m trying to model. You would need a decimation filter, but that’s done in the digital domain and is predictable and relatively easy to design.
2. Yes. This amounts to having a higher resolution and will only make things better. However, if you have a 12-bit quantizer to begin with, you might not need a sigma-delta (certainly not a continuous-time one).
3. This is difficult. The gains are sort of trial-and-error to keep the loop stabilized. There are some attempts to predict what they should be, but most people use them as starting points for their trial-and-error.
4. The output of the ADC shouldn’t have any additional delay. The integrators in the simulink model are there to form an error amplifier. The delays they introduce are taken into account when looking at loop stability. So, the integrators introduce some (ideal) delay that should be accounted for when designing your feedback loop.

I get the strong impression that what you want isn’t really a continuous-time sigma-delta, but rather some sort of PWM controller that involves an ADC. If you’re still interested, email poojanwagh at this domain and I’ll take a look at it.

By: amitv singh tajput

amitv singh tajput — Thu, 07 Jan 2010 06:26:58 +0000

i m also working on that we can share our view

By: Daniyal

Daniyal — Tue, 22 Dec 2009 16:59:26 +0000

I have some question if u’d be kind enough. I must say I am new to ADC’s, we have an ADC with our machine tools controller but we don’t know anything about it except for its sampling frequency and resolution (12-13) bits. So since we were modelling the whole controller so we needed to simulate an approximate model of the ADC as well. Anyways.
1- Why the your ADC does not have any filtering and decimation at the output ?
2- Can i replace your ’sign’ block with a 12 bit quantizer ?.
3- Wha about the Gain in the integrators block ? I mean can we actually compute them or just arbirary.
4- Our ADC does not have any delays at its output (since it will actually destroy the whole controller). So these simulink integrators, do they have substancial delays ?

By: PoojanWagh

PoojanWagh — Sun, 20 Dec 2009 06:58:18 +0000

What specifically are you confused about? Do you have the simulink output (sampled digital words) in a matlab variable (vector)? If so, it should be pretty easy.

By: krishna

krishna — Thu, 17 Dec 2009 12:50:38 +0000

hii there,
I am doing project on delta sigma ADC for gsm receiver …. i have developed a simulink model and now i should develop a code that runs the FFT for the model that i have develped i am a bit confused in developing the related code for the model that i have developed ..could u please help me in developing the FFT code for the model that i had build……..

By: Poojan Wagh

Poojan Wagh — Mon, 16 Nov 2009 12:26:30 +0000

Take a look here: http://www.circuitdesign.info/blog/2009/01/non-radix-2-fft-in-cadenceoceanskillspectre-using-cadenc-ipc-to-talk-to-matlab-or-anything-else/
It has more than what you need, but all the pieces are described in the post. You should be able to use that to compute SNR. The dynamic range is merely the range of inputs (or outputs) between 0 dB SNR and maximum SNR. If your ADC behaves well, this is equal to the maximum SNR.

By: Bukhari

Bukhari — Mon, 16 Nov 2009 11:59:23 +0000

hi. i am doing project on Sigma delta analogue to digital converter and i need the matlab code for that how to calculate the Dynamic range and how to design the whole code?
thanks

By: wagh

wagh — Thu, 29 Oct 2009 20:43:14 +0000

Sorry, John. I can’t offer much help. In fact, I don’t know why you’d want to simulate an ADC using Matlab unless it was some sort of algorithmic ADC (SAR, Sigma-Delta) that involved some signal processing. Basically, your plain-old flash ADC just does one thing: y = round(x * L)/L — assuming x is +/- 1 and L is the # of levels you want to quantize to.

If you want to look at things like SNR and sampling, it’s pretty easy to hand-compute it, and for sampling, pure Matlab isn’t really suited (Simulink is better). Maybe you knew that already.

By: John well

John well — Thu, 01 Oct 2009 13:11:29 +0000

Hi,

I am learning matlab(Self teaching using books) but one question is really giving some headache. I hope you can help me with that:

write a computer program to simulate an analog to digital converter using matlab. tha a/d must have the following specifications:

full-range input of +- x volt
y bit resolution
z sampling frequency

Please , I really need your help or if you can recommend me some books that deals with such things; it will really help me.

By: Poojan Wagh

Poojan Wagh — Tue, 12 May 2009 17:35:24 +0000

Gilbert: A simulink model for a 3rd order should be pretty simple. Just replace the 2nd-order noise-shaping filter with a 3rd-order.