Can Integration Help With Our Noise Gamble?

Noise is the ever-present source of headaches on many engineering designs, and often is one of the biggest underlying challenges a designer team faces. (See: Noise annoys, but do we secretly love it?). That's why I cringe when I hear an engineer casually remark, “Assuming that the noise is random and Gaussian (white)…” in any analysis or discussion.

There are two reasons for my concern. First, just because noise is random doesn't mean it is Gaussian. Second, assuming that it is Gaussian is actually a pretty big assumption to make. Lots of noise in the real world just isn't nice like that.

To cite a few examples, there's pink noise, impulse noise (of many types), Rician noise, cosmic noise, and one of my personal favorites, cyclostationary noise (where the mean and other statistical moments are periodic, as from a rotating radar antenna or a machine bearing). Note that you have to distinguish between the noise source , and the resultant noise type , as there are many possible sources for each type of noise.

Why do we make that Gaussian-noise assumption? It's an easy one to make, especially if you don't have the ability or opportunity to look at the noise of your situation in detail.

Even better, it simplifies the mathematical analysis, plus the hardware filtering and signal-analysis algorithms you may be planning to use. So we take a gamble, assume it is Gaussian, and hope it all works out in the end.

That's why I think that increased use of application-specific standard products (ASSPs) or application-specific ICs (ASICs) may help us overcome the result of this noise simplification. In principle, these ICs are designed with a specific application in mind, by their nature. Hopefully, the design and application teams have fully studied the noise of this application, and embedded filtering and signal-processing algorithms that are optimal for the situation at hand — though they may be quite wrong for others.

The design may need a low-pass, band-pass, time-varying, non-linear, or matched filter, to cite just a few options. The signal processing may also need to have algorithms that are tailored to the realities of the noise, such as multiple time constants and time bases, or special types of averaging.

These can truly make the ASSP (or ASIC) work where simplistic approaches may not be the right choice. After all, the noise seen by an electrocardiogram's analog front end (AFE) which is dealing with microvolt signals is quite different than the noise in a motor-control loop with MOSFETs switching nearby.

Or look at one of Jim Williams's earliest articles for EDN (“This 30-ppm scale proves that analog designs aren’t dead yet,” EDN , Oct 5, 1976), where he built a scale with dual time constants in the strain-gage front end: a fast one to accommodate mechanical noise from vibration, then switching to a slower one to deal when a sample (a baby!) was placed on the scale. It's unlikely that either “noise” source was Gaussian.

Noise is always there; how you handle it depends on what you know about it. Hopefully the developers of the latest ASSPs/ASICs understand it and can put that knowledge into their chip (or chipset), but only after verifying their analysis and approach to dealing with it.

What sort of noise annoys you? Have you seen circuits and algorithms which were well-tailored to non-Gaussian noise?

Related posts:

16 comments on “Can Integration Help With Our Noise Gamble?”

1. DEREK.KOONCE
June 12, 2013

Will there be a customizable noise supression IC in the future? What will be the difference between that an a good filter circuit? I do not see much difference.

It seems this may be an opportunity for fuzzy logic versus the basic PID control circuit that Pease was up in arms about. Also I remember the audio-phile challenge he had out there for the Monster Cable connection versus the connection he put in a box.

Scary thought that it could turn into a fad and suck in gullible new engineers.

2. Bill_Jaffa
June 12, 2013

I know that engineers complain about noise and have to deal with it all the time, but look at it this way: if it wasn't for noise, our jobs would be so much easier, that lesser-skilled folks do do them, in many cases. Dealing with noise separates real, experienced engineers from textbook-mostly ones! 😉

3. Bill_Jaffa
June 12, 2013

There are also “tricks” you can do by adding noise to your signal, then sampling at a higher rate, process the results, and get a few more bits out of your ADC. But you have to be sure that the added noise is white, or else you'll get more bits but they will be wrong–but, hey, you and your boss will be happy, at least for a while, anyway.

4. Davidled
June 12, 2013

Noise is like dust. Noise is everywhere.  Engineer designs all kind circuit to suppress or minimize noise or redesign the size of module in order to eliminate noise.  Other job opening for Noise engineer is required in the job market that specializes and understands noise source and noise cancellation.   For example, inside cabin of vehicle has more noise while driving compared with that in engine ideal.  Noise comes from engine and vibration and friction between metal and metal. This kind noise should be considered in the Electronic module.

5. Steve Taranovich
June 13, 2013

I have had some good results using noise “dithering” on high speed ADCs in the past. Every slight improvement in SFDR helps in some designs.

There is always some correlation between quantization noise and input signal. Usually, if the input frequency is an exact sub-multiple of the sampling frequency then the SFDR may be reduced somewhat.

To help minimize this effect, you can add 1/2 lsb rms broadband noise or less (Caution—adding too much noise can hurt the dynamic range) with the input, then the quantization noise will be randomized. Walt Kester, ADI, had a good article on this “ADC input noise: The good, the bad, and the ugly. Is no noise good enough?”

June 14, 2013

@DaeJ >> For example, inside cabin of vehicle has more noise while driving – and there's another good blog topic – audible noise canceling – touched on previously here:

http://www.planetanalog.com/author.asp?section_id=385&doc_id=558728

But I'm sure there's more to be said.

June 14, 2013

Bonnie Baker has been discussing also – came up in this msg-thread:

8. jbike
June 15, 2013

I could not agree more. MOST of the noise that I deal with is not perfect thermal noise or even gaussian for that matter. Rather, it is related to our love/hate relationship that we all have with the ubiquitous switching power supply. Number two in my book is noise related to the mixed signal aspect of nearly every circuit we  design these days. This worries me from an integration perspective, since this unwanted coupling both magnetic and electrostatic can be minimized by adding distance and/or guards… a precious commodity of which there is not much of on a silicon chip.  How many projects do you work on that are pure analog? I'd say few these days. Third on my hit list are external noise sources including power line related problems and noise from external equipment. Next we can discuss all the noise source that are in theory supposed to be statistical but turn out not to be so for a variety of reasons.

9. goafrit2
June 17, 2013

>> Will there be a customizable noise supression IC in the future?

I think better process and good circuit will always triumph any attempt to get a customizable suppression IC as a product. We need to understand that most of the issues with ICs can be traced to immature process that is not reliable. And once the IC is stable, we upgrade to a new one and the whole thing continues.

10. goafrit2
June 17, 2013

>>  if it wasn't for noise, our jobs would be so much easier, that lesser-skilled folks do do them, in many cases

But noice in a process is good. People utilize that ability to model behaviors of neurons in most neuromorphic circuuits. You have this stoachastic process you need in biocircuits and you get that out via the inherent noice in circuits and systems

11. goafrit2
June 17, 2013

>> There are also “tricks” you can do by adding noise to your signal, then sampling at a higher rate, process the results, and get a few more bits out of your ADC.

It is turning the lime into lemonade design. My boss likes it a lot. However, most of those extra bits are marketing bits. You claim it is 14 bits when indeed ENOB of 12 bits. Only the novice will be confused!

12. goafrit2
June 17, 2013

>> For example, inside cabin of vehicle has more noise while driving compared with that in engine ideal.

The pros that work on cabin design especially in cars, they simply jam the noise especially when it is within the audio frequency range. You do not want that to be heared. There are tricks but total elimination of noise is not possible. All one has to do is manage it and design circuits and systems that can function reliably within them.

June 17, 2013

@jbike – You should probably write more about this in a blog – especially the issue of noise due to proximity of the digital and analog cktry within the integrated analog/mixed signal IC.

June 17, 2013

@DaeJ >> Noise is like dust . And related to that, noise is like dirt – with a similar flexible definition: stuff that you don't want mixed in with stuff that you do want. So the same stuff can be OK in some places and completely unwanted elsewhere. Kind of like hair in your eggs; or eggs in your hair.