[sdiy] Better testing techiques for audio circuits

[Harry Bissell]

I find that spice can be a good indicator of a general direction (architecture)for circuit design. Some versions
such as LTSpice can even output a .wav file and you can listen to the result. But be prepared for...

LONG LONG simulation times... some of my recent works have taken several days to generate 45 seconds of audio...
Models that crash, don't work, simulate poorly...
Simulations that LIE...  (says it works when it won't, says it does not work when it does...)

examples of the last class... transistor models are perfect. Use a dozen transistors and they are perfectly matched. It might
be that in the real world a 1% difference would render the circuit useless...

I once did a clamp circuit that worked VERY well... but when I realized the upper clamp could source current which the lower clamp
could sink... I looked for what the standing current was. It was 100 AMPS... naturally my mA signals was clamped very effectively.
The 741 opamp models happily went along with the joke... :^)

Trying to simulate mixed signals is mostly a nightmare. DO NOT expect a digital model to play well with an analog one. In fact to model
what 4000 CMOS would realy do, I resort to crude transistor-level models for an all-analog simulation. These are slow and of questionable
accuracy, but at least I don't have weeks of trying to figure out why a circuit that ~should work~... doesn't

The best use for spice, IMHO... is to make a model of a circuit you actually build, and compare the results. Add parasitics to the spice model to make it converge with what the real circuit is doing... and use the model to gain insight into why the real world does what it does...

all in all I like Spice a lot... it can tell you (most times)if you have made an abjectly stupid mistake...

H^) harry

----- Original Message -----
From: Justin Owen <juzowen at gmail.com>
To: SDIY List <synth-diy at dropmix.xs4all.nl>
Sent: Wed, 04 Apr 2012 04:21:15 -0400 (EDT)
Subject: Re: [sdiy] Better testing techiques for audio circuits

One follow up question to this...

How useful are simulations like SPICE in showing noise and distortion level in a circuit?

...or more specifically - if I was to simulate a 'bare bones' version of a simple, discrete circuit and a 'precision' version of the same circuit - how useful would simulations be in showing any differences between the two with regards to improvements (or drops) in noise and distortion?

Thanks again.

Justin

_______________________________________________
Synth-diy mailing list
Synth-diy at dropmix.xs4all.nl
http://dropmix.xs4all.nl/mailman/listinfo/synth-diy

-- 
Harry Bissell & Nora Abdullah 4eva