[sdiy] uLaw DACs in the modern era?

[Richie Burnett]

Probably any of those techniques you mentioned would work to some degree Tom, subject to things like component tolerances and temperature drift. But a digital conversion or LUT would be more precise and repeatable than a log amp.

I'm not sure what magic punch you think it might add to sounds though? Mu-law and A-law companding were introduced because they perceptually sound largely transparent. You will certainly hear the effect of the variable size quantisation steps on something like a sinewave test tone, but on more complex sounds it is quite subtle.

And yes, these companding techniques were invented by the telecoms industry to allow wide dynamic range audio to be passed through 8-bit channels with otherwise limited dynamic range. Think of it as instantaneous dynamic range compression at the sending end and expansion at the receiving end, applied on a sample by sample basis with no memory.

-Richie,

Sent from my Xperia SP on O2

---- Tom Wiltshire wrote ----

> Yeah, recreate it is what I’m thinking. I don’t want to starting mining for unobtanium. A LUT+linear DAC is the obvious/simplest way for sure, but I was wondering about other more interesting (if probably less practical) possibilities like a non-standard resistor string or R/2R DAC (or combination thereof) or a exponential amplifier to convert a linear DACs step into a “expanded” range like a u-Law decoder.
> 
> Give it another fortnight and Alfa Rpar or Behringer will probably re-release the AM6070 anyway!! ;)
>  
> 
> > On 19 May 2018, at 20:21, paula at synth.net wrote:
> > 
> > Tom,
> > 
> > Why go for a hard to find DAC? why not recreate it?
> > 
> > I "recreated" the uLaw using a 12 bit DAC and an 8 bit LUT. Then I was able to play back the LInn samples directly.. sounded pretty nice too IMHO. Ok, yes I should've used 13bits, but I didn't have any 14bit DACs to hand, but I did have 12 bit, so I went with that.
> > 
> > Paula
> > 
> > On 2018-05-19 11:08, Tom Wiltshire wrote:
> >> Hi All,
> >> Over the last couple of days I’ve been studying early drum machines,
> >> mostly the Linn LM-1 and the Oberheim DMX.
> >> Both of these machines use 8-bit memory, but manage a 12-bit dynamic
> >> range by using a u-Law coding and outputting samples via the AM6070
> >> u-Law DAC. Datasheet here if you’re interested:
> >> http://synfo.nl/datasheets/AM6070.pdf
> >> Does anything like this exist any longer? It seems to me that the
> >> unusual DAC response is big part of the “thump” of these early
> >> machines, since it has considerable distortion, but manages to spread
> >> it to areas where it’s perceptually less noticeable (so higher level
> >> signals have less detail than lower level signals).
> >> Alternatively, would it be possible to build such a device? The
> >> combination of a 4-bit value plus some gains and offsets doesn’t seem
> >> impossible to manage, but I’d be interested to hear people’s ideas for
> >> how you’d do it.
> >> Thanks,
> >> Tom
> >> ==================
> >>       Electric Druid
> >> Synth & Stompbox DIY
> >> ==================
> >> _______________________________________________
> >> Synth-diy mailing list
> >> Synth-diy at synth-diy.org
> >> http://synth-diy.org/mailman/listinfo/synth-diy
> 
> 
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy