[sdiy] Low voltage synthesis?
mbryant at futurehorizons.com
Fri Jan 14 04:36:52 CET 2022
Using currents rather than voltages is how it's done inside linear ICs. I think it's explained in Art of Electronics.
Nowadays mobile phone amplifiers are all digital - the digital audio (I2S possibly) digital signal is converted to a PDM signal which drives a class D amplifier and a simple LC reconstruction filter on the output.
Some years ago they did use DACs (e.g. Wolfson devices) followed by a Class D amp, until someone realised going digital -> analogue -> digital -> analogue was a nonsense.
By "linearly rather than exponentially" I mean that most synths use an exponential convertor to convert from a voltage input to a current output, which then drives the Moog ladder, or some forms of VCO, etc. The Blackmer cell also uses currents internally.
> Wouldn't audio / synth circuits create much less power dissipation if they used power (impedance-matched) signalling rather than voltage signalling (low into high impedance)? I haven't worked much with impedance matched audio, but I believe it should lower power losses in circuits, right? Could that be one way to do "colder" synths? In which case you could keep the "high" rail voltages and keep the benefit of outpacing resistor noise.
Actually it doesn't affect power dissipation much, but if you're in the Rupert Neve camp of how to do things, it is the correct way and you do get slightly better noise figures. But the industry moved away from that approach some years ago with low output impedance, medium-high input impedance becoming the norm, though every manufacturer has a different in and out impedance. Music Tribe even have differing impedances across their brands.
From: cheater cheater [mailto:cheater00social at gmail.com]
Sent: 13 January 2022 23:45
To: Mike Bryant
Cc: synth-diy at synth-diy.org
Subject: Re: [sdiy] Low voltage synthesis?
On Thu, Jan 13, 2022 at 6:07 PM Mike Bryant <mbryant at futurehorizons.com> wrote:
> resistor noise still eventually wins.
I am aware of resistor noise as one source of issues in a low voltage system. I am also wondering what other problems there might be. Brian talked about voltage regulator dropout and opamp rail margin, and those are pretty valid, but I wonder what's being used in places like mobile phones or wireless earbuds for example, they can't be running
+/-5V op amps or LV op amps with major crossover distortion.
> Techniques to look at to reduce problems further include using currents rather than voltages wherever possible.
What sort of noise does that reduce and how does it do that? It makes sense to differentiate between voltage and current noise. For example, the resistor in an RC high pass filter will introduce noise in the voltage, since the resistor is across the signal to ground. However, in an RC low pass filter, the resistor will introduce noise in the current.
> Of course many synth circuits do use currents - the Moog ladder filter for example - but you'll need to work linearly rather than exponentially.
What does "linearly rather than exponentially" mean? Are you saying the moog vcf somehow works "exponentially using currents" and I would need to somehow make it work "linearly using currents"? I don't know what that means - would you mind explaining a bit? Thanks.
> The other issue you run against is you need to use rail-to-rail op-amps, but most of these are specified for 5V/+-2.5V operation. Also these opamps are often CMOS based and have what I term "input crossover distortion" as they switch from the N to P type devices conducting. Negative feedback reduces this, but it's always slightly there. There are a few bipolar R2R opamps without this problem - Rohm BA2510 is my favourite.
Thanks for the suggestion on the op amp.
BTW, I have planned to read (some of?) Low-Noise Electronic Design by Motchenbacher in order to help with figuring out how to do noise. (I think the book title changed a little later).
I've browsed the book just now for a bit and I wonder about one thing.
Wouldn't audio / synth circuits create much less power dissipation if they used power (impedance-matched) signalling rather than voltage signalling (low into high impedance)? I haven't worked much with impedance matched audio, but I believe it should lower power losses in circuits, right? Could that be one way to do "colder" synths? In which case you could keep the "high" rail voltages and keep the benefit of outpacing resistor noise.
> -----Original Message-----
> From: Synth-diy [mailto:synth-diy-bounces at synth-diy.org] On Behalf Of
> cheater cheater via Synth-diy
> Sent: 13 January 2022 16:21
> To: synth-diy
> Subject: [sdiy] Low voltage synthesis?
> Synths generally run off 15V or 12V power rails, which among others helps with SNR. However, it creates a bunch of power dissipation, plus you can't use some of the tiniest components. Does it make sense to build synthesizer circuits that run on much lower voltages than that?
> Say, 1.5V rails? Are there techniques of suppressing noise that are better than just ramping up the voltage rails? Have the parts progressed enough nowadays that we can do this?
> Never mind stuff like interfacing with existing modular synthesizer parts etc. Let's say you're building a modern 8-voice polysynth with 4 VCO, VCF, VCA, ENV, LFO per voice. Obviously that would generate loads of heat and possibly tuning issues given the constraints of a keyboard, plus it would be huge, especially given the size of capacitors vs their rated voltage.
> Where would noise creep in first if you started lowering rail voltage in your synthesizer designs? What other disadvantages in terms of signal integrity come from using low voltage rails for synths?
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> Selling or trading? Use marketplace at synth-diy.org
More information about the Synth-diy