[sdiy] Bubble memory analog delay?

Richie Burnett rburnett at richieburnett.co.uk
Wed Jul 10 00:59:24 CEST 2013

> Note - magnetic tape domains don't reflect the
> signal they represent in their degree of magnetization, either. The
> domains are created artificially with the bias oscillator.

The analogue signal being recorded _is_ represented by the degree of 
magnetisation.  However, due to magnetic hysteresis (a type of non-linear 
behaviour,) a bias oscillator is employed to try to linearise the response. 
The key difference is that with audio tape the signal only gets mangled once 
by what's left of the non-linearity during the recording/playback process. 
If you try to shuffle analogue "bits" down a magnetic delay line I think the 
cumulative non-linear effects will quickly wreck your analogue signal.  I'm 
open to being proved wrong of course, if you try it!

>> Much better to use modern DRAM
>> derivatives for digital audio delays in my opinion.
> How do you mean? Just DAC -> DRAM -> ADC? Something like that?


The RAM addresses are generated by a digital counter that you can clock from 
a fixed or variable oscillator as you wish.  You should be able to find 
examples of digital audio delays using either SRAM or DRAM chips on the 

You usually have to take special precautions to periodically refresh DRAM 
(every few milliseconds,) however for a digital audio delay where all the 
rows are accessed sequentially anyway, this can serve to provide the refresh 
function automatically.

I designed a product for a customer about 15 years ago that delayed stereo 
audio by 90 seconds using a 16MB EDO DRAM SIMM of the type used in PCs back 
then.  The application was a "profanity delay" for outside broadcasts, live 
radio phone-ins, that sort of thing.


> you want to somehow modulate the clock at which the ADC and DAC work?
> Perhaps SRAM is a better choice in that case.
> Speaking of which, I was wondering if there was something like a 2D
> CCD where you could move the image either horizontally or vertically.
> I wonder if one could possibly implement simple convolution with such
> an item.
> Cheers,
> D.
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy 

More information about the Synth-diy mailing list