AW: [sdiy] The future of synth DIY

Haible Juergen Juergen.Haible at nbgm.siemens.de
Fri Mar 2 16:04:49 CET 2001


	>>Discrete diy soldering and integration of monolithic circuits have
	>>nothing in common. Design engineers for monolithic chips spend
	>>*a lot* of time to think about "possible" errors in advance.
	>>You wouldn't do that in discrete circuits, because it's so easy
	>>to simply solder it and measure afterwards.
	>
	>There is an analogy here with computer programming.
	>When I did it in the 60s, you had a deck of cards and got one (or
	>if you were really lucky and had inside contacts) two runs a day.
	>So you thought damn hard about whether the syntax and logic were
right.
	>Perhaps if we could only solder one component per hour, we would 
	>become better designers? just a thought..

My ratio of (analysis + simulation) / soldering is somewhere about 100 : 1. 
Contrary to common belief, I'm conviced that Spice is an excellent tool
to design discrete analog circuits almost to the very last detail.
(Making a bad conceptual error does happen, but then it's often
better for me to dump that whole veroboard than optimizing there.)

Nevertheless, I might return to a more conventional design process
in the future: Soldering hurts the back, but simulation hurts the eyes. 

As for that Elektor PWM phaser, don't expect a large sweep range
to be achieved easily. Martin has pointed out the main problems
already. My additional 2 cents: consider the PW range and the 
resulting minimum pulse width you're aiming at. Switching frequency
must be above audio range, which sets the *maximum* pulse width.
40kHz - Tmax = 25us. For a 3 decades sweep, your minimum 
pulse width is 25ns - not impossible, but not easy to achive without
jitter and with good temperature stability. Ideally, you'd want to
modulate the PW with an expo characteristic - not easy, as your
most temperature sensitive branch of the expo courve has to deal
with the smallest pulse width.
The Roland JP-4 uses this on the VC Envelopes, and the ETI vocoder
on VC Slew. But there it's not in the audio path, so the clock frequency
can be much lower and the timing constraints are not so severe.

IMO, that whole PWM phaser concept is no good idea altogether.
It surely works, sort of, but to get a good sweep range you must
invest more than for other phaser concepts.

JH.




More information about the Synth-diy mailing list