first use of PWM techniques for a variable gain block?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

bcarso

Well-known member
Joined
Feb 20, 2005
Messages
4,055
Location
San Fernando Valley, Los Angeles
I'm thinking that pulse-height/pulse-width multipliers would qualify, and those date way back. Much more recently I've seen PWM switched resistors, but surely they would be considered a subset of the multiplier.

Any historians in here know?
 
I did breadboard a simple working unit some time ago, but never fully developed it. It used readily-available devices, -I'd love to see what you made of the sketch, Brad...

-There's probably a dozen stupid things that I did wrong. -PRR did make some very helpful observations, and greatly changes my mental approach to how well (or rather, not well) op-amps reject high-frequency schmutz, but I've not revisited that sketch for some time.

Keith
 
There's a Grass Valley PWM audio limiter with 7/64 on the drawing. That's the earliest I've seen produced commercially.
 
Thanks to all.

I'm working with a new client who has patent paranoia. While some caution and diligence is appropriate in this biz he seems to be just totally over the top. I haven't quite figured out why, but in the meantime I'm obliged to either assuage the fear by citing prior art, or suggest that we patent some aspect of anything we do as a protective measure.

I hate to be cynical but it's pretty easy to do the latter by just making the claims sufficiently restrictive. It makes for a fairly worthless patent but still can be a deterrent. But of course patents take a long time to get and cost a lot, and then you have to pay to maintain them.

I recall once when Bose sent a letter threatening suit to the automotive group at Harman. H's response was to cite one or more patents and other prior art and point out that if Bose pursued litigation, H would see to it that their specious patent cited would be struck down.

H never heard another word. But had this been Bose and you or me, they would likely have just crushed us, knowing that we wouldn't have the resources to defend ourselves.
 
[quote author="SSLtech"]I did breadboard a simple working unit some time ago, but never fully developed it. It used readily-available devices, -I'd love to see what you made of the sketch, Brad...

-There's probably a dozen stupid things that I did wrong. -PRR did make some very helpful observations, and greatly changes my mental approach to how well (or rather, not well) op-amps reject high-frequency schmutz, but I've not revisited that sketch for some time.

Keith[/quote]

I see that thread started before my time here, but also that I still managed to stick my oar in late in the game.

Looking it over now, it seems fine. As usual PRR's comments are well-considered.

Probably a few things I would do a little differently, as among other things I've never been a big fan of the 555, and I'd have a teensy bit of hysteresis applied around the 319, etc. etc. There might be a first-order compensation of the FET gate-channel-coupled switching charge, or some other differential scheme.

Indeed the thread and discussions are good and reminded me of the need to do a thorough job of bandlimiting going into any chopper as well as when coming out of it. Another consideration: in addition to the L-C filtering adjunct suggestion, note that our favorite Sallen-Key filters have some departures from ideal rolloff when you get out to high frequencies. There was some app note/blurb about that recently, and I forget what the author recommended.
 
Ah the simpler times when we weren't concerned about rogue clock frequencies everywhere.

I don't have any early cites but believe PWM is older than dirt in heater controllers and such.

One clever duty cycle trick I recall seeing was a complex multi-pole filter around a sampled delay line, where the resistors in the filters were chopped (by cmos transfer gates) to make the filters vary and track the desired anti-alias, anti-image tuning for a given clock frequency. IIRC the chopper used a one shot for on time, with off time varied by the clock wavelength minus the one shot time. At high clock frequency you had the high on time and higher pole freq, at lower clock frequency you had less proportional on time and lower pole frequency.

This was not a published design but probably 20 years old..

JR
 
LOL.. I just remembered an old (bad) idea I had for cheap digital snake.

Use a HF chopper to merge a number of analog input onto a single wire. At the receive end de-chop these to S&H circuits at an appropriate refresh rate.

like I said not a great idea.... :grin: I've had lots of those.

JR
 
[quote author="JohnRoberts"]LOL.. I just remembered an old (bad) idea I had for cheap digital snake.

Use a HF chopper to merge a number of analog input onto a single wire. At the receive end de-chop these to S&H circuits at an appropriate refresh rate.

like I said not a great idea.... :grin: I've had lots of those.

JR[/quote]

Time-domain multiplexing/demultiplexing like that shows up as an example in some texts I've seen, in fact.
 
There are some great compressors that use this technique.
The EMT 156 has already been mentioned.
I aquired one of these from eBay a couple of years ago. the first big studio I worked at had one of these in every room and I remember how transparent it sounded.
Also, the Pye compressor - a very useful workhorse.
I think these would be difficult to build now - a lot of the parts being custom made. But, yes, an interesting technique that seems to give good results.
 
Thanks again for those references.

It turned out the client had thought that I was proposing something along the lines of the Anadigm stuff, which is mostly switched-capacitor architectures. I think I've convinced him that the PWM resistors are not a concern.
 
Back
Top