[sdiy] Analysis of the TB-303 CPU timing

Richie Burnett rburnett at richieburnett.co.uk
Thu Mar 16 21:47:22 CET 2017

I couldn't hear any difference there, but that's quite a slow leisurely 
pattern.  You're more likely to hear a given amount of ms of jitter on a 
faster pattern, so try a pattern with loads of repeating 16th notes at a 
frantic tempo with generous use of accents and slides.  That might be more 
revealing.  (Also use higher pitched notes to reveal timing detail, because 
it's easier to perceive onset timing for notes with higher fundamental 

BTW, I seem to remember that one of the biggest causes of "latency" in the 
TB-303 is the "gate processing circuitry" between the micro and the VCA. 
The onset of notes gets delayed by some number of milliseconds due to the 
charging of some capacitor in the gate processing circuit.  From what I 
remember this also delays the rising edge on the instrument's gate output. 
So the gate output lags the micro's internal gate output, but TB-303 appears 
to play in time if you just compare the audio output with the gate output. 
It's only when you look at the gate out of the micro or the tempo clock that 
you can see that everything is playing so many milliseconds late due to a 
combination of software latency+jitter *and* hardware delays.


-----Original Message----- 
From: Julian Schmidt
Sent: Thursday, March 16, 2017 5:45 PM
Cc: synth-diy at synth-diy.org
Subject: Re: [sdiy] Analysis of the TB-303 CPU timing

hmmmm, my first try says it's probably not worth the hassle.
Or can you hear a difference?


CPUs used in random order:
QS, RE-CPU, TB, RE-CPU with extra sloppy timing (double the latency double 
clock polling time (3.6ms)
All normalized to -1dB


Am 16.03.2017 um 13:32 schrieb Tom Wiltshire:

On 16 Mar 2017, at 11:43, Julian Schmidt <elfenjunge at gmx.net> wrote:

So I could record a few patterns with different CPUs in the same synth 
hardware to exclude influences by differences in the analogue part of the 
machine or different knob settings.

Would be interesting to find out
A) If people can hear the difference between original sequencer, emulated 
sequencer and a tight sequencer
B) What listeners prefer if they are able to distinguish it.

maybe even add a 4th audio example to the test where the jitter and 
latencies are increased slightly?

I think this would be very interesting. There's a lot of hand-waving and 
magic around this little silver box and not much in the way of hard facts, 
so some serious research like you've been doing is extremely welcome.

If you be bothered, setting up the test as a blind study so people don't 
know which file they're listening to, and then asking which one they think 
they heard and how much they like it would definitely be the best way to 
proceed, but I realise the work involved.

Thanks very much for what you've done so far. It's fascinating.


Synth-diy mailing list
Synth-diy at synth-diy.org

No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.8007 / Virus Database: 4756/14126 - Release Date: 03/16/17

More information about the Synth-diy mailing list