[sdiy] Can google's free* 180nm OSHW foundry be used for synth parts?

[Mike Bryant]

On your points :

Noise figure of a good audio transistor will be about 2.5dB.
Raw flicker noise from some digital transistors can be as high as one-quarter of the voltage rail - so almost -10dB down.  This is why you sometimes have small capacitances inside digital logic devices - usually implemented as capacitance to substrate along the tracking so not always shown on schematics.  Usually not quite that bad, but digital transistors aren't designed for low noise, just fast reliable, bounce-free switching.

Most audio is on +/-15 (or 17 for pro-audio).  Add in some stress due to capacitors on the inputs and outputs bootstrapping the voltages and you're soon at 50volts. 

Bipolar isn't essential - plenty of good CMOS audio products.  But many run on lower voltages such as 5V and you have to avoid the crossover point distortion so best to run op-amps in inverting mode.  All ADCs run off 5V (or less) and most DACs do as well but some generate a negative supply line internally.  (OMG - that's an SMPS - can't possibly be used for quality audio :-)

Are there any 5V only analogue synths ?  Probably not but no reason one couldn't be created.

Could you build an OTA using CMOS ?  Need to think on that one.  Certainly would need something other than the linearizing diode on the input :-)

Main problem with having 64 OTAs would be connections.  Most OTAs drive a capacitor which would be off chip so definitely need lots of pins.  And lots of grounds !  
Better to think of common modules that could be integrated with fewer pins, rather than discrete components.  Also hate to think what the crosstalk would be on so many analogue devices.  Dual and quad op-amps go to huge efforts to isolate from each other with all their pins separated from other blocjs, but once you get to the point of having to route tracks between analogue blocks to get to the periphery for bonding it could get horrible.  There have been FPAA (field programmable analogue arrays) - some friends of mine did a startup doing one over a decade ago - but they never seem to take off for this very reason - the noise figures are never as good as you really need.



-----Original Message-----
From: cheater cheater [mailto:cheater00social at gmail.com] 
Sent: 06 August 2022 14:19
To: Mike Bryant
Cc: synth-diy at synth-diy.org
Subject: Re: [sdiy] Can google's free* 180nm OSHW foundry be used for synth parts?

Hey Mike, thanks for the email.
"and noise figures in single dBs, not single dBs down from the voltage rail" - hmm, mistake perhaps?
"high voltage 50V transistors" - why is that necessary? Most synth chips are more like 15V rail chips, I don't think I've seen synth chips that run off 50V rails, what am I missing here?

A friend who works at a semiconductor fab told me that the process might not support bipolar transistors (he just doesn't know if it does), but I'm not sure how necessary those are for the purpose of creating audio chips?

Peter,
I'm sure once people have 64 OTAs on a chip they'll find a use for them :)

Best regards


On Fri, Aug 5, 2022 at 8:36 PM Mike Bryant <mbryant at futurehorizons.com> wrote:
>
> It's a digital process with components optimised for digital 'analoguey' applications such as line drivers.  Analogue to the quality levels you are talking about need dielectric isolation, high voltage 50V transistors and much higher gains so you can apply some feedback to get distortion down, and noise figures in single dBs, not single dBs down from the voltage rail.
>
> I think it's also compulsory to include their microprocessor core on 
> the die - it definitely was on early runs
>
>
> -----Original Message-----
> From: Synth-diy [mailto:synth-diy-bounces at synth-diy.org] On Behalf Of 
> cheater cheater via Synth-diy
> Sent: 05 August 2022 19:16
> To: synth-diy
> Subject: [sdiy] Can google's free* 180nm OSHW foundry be used for synth parts?
>
> Google are running a program where if you submit open source chip designs they'll produce them either cheaply or for free on a 180 nm process.
>
> "We sponsored six shuttle runs over the course of two years, allowing the open source community to submit more than 350 unique designs of which around 240 were manufactured at no-cost."
>
> Is this a process that could be used for analog synth parts? We could use new OTA chips, and I think I wouldn't mind a chip with 64 monolithic OTAs on it, or chips with full voices, etc. With this feature size, leaps and bounds beyond the last time synth chips were being miniaturized, we could go wild with designs.
>
> Apparently they already have solutions for on-chip passives, single transistors, and relatively "high" voltages (10V), so the next question is how they would handle something like an OTA, VCA, op amp, etc, since most users seem to be focusing on digital chips.
>
> "The GF 180nm technology platform offers open source silicon designers new capabilities for high volume production, affordability, and more voltage options. This PDK includes the following standard cells:
> - Digital standard cells libraries (7-track and 9-track)
> - Low (3.3V), Medium (5V, 6V) and High (10V) voltage devices
> - SRAM macros (64x8, 128x8, 256x8, 512x8)
> - I/O and primitives (Resistors, Capacitors, Transistors, eFuses) cells libraries"
>
>
> https://www.phoronix.com/news/Google-GloFo-GF180MCU
>
> https://opensource.googleblog.com/2022/08/GlobalFoundries-joins-Google
> s-open-source-silicon-initiative.html
>
>
> If we can figure out how to take advantage of that, synth chips could become very interesting.
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy
> Selling or trading? Use marketplace at synth-diy.org
>