PIC based delay

[gez]

Interesting circuit in this month's (July 2005 edition) Everyday Practical Electronics magazine.   It's a 'Dalek voice emulator' called the 'Cybervox'.  Not that interesting in its self, but part of the circuit utilises a PIC based 'echo'!

A PIC18F252 device is used, which has a memory bank of 1536 bytes.  The code is really simple, as is the circuit, and it kicks out around 24.6ms delay with a sampling rate of 64kHz and 565ms with a sampling rate of 2.4kHz (according to the blurb).  Not very hi-fi at the lower end of things, especially as the ADC is 8bit, but with a little thought you could probably get a half decent chorus effect from this chip.

I’ll have to dust down my intro to PIC books…

www.epemag.co.uk

[amz-fx]

I suspect that it is using 8-bit A/D which will limit the dynamic range, i.e. the noise floor will be higher than you would like.

I subscribe to the digital version of EPE but it is not online yet so I cannot look it over.

http://www.epemag.com/

regards, Jack

[puretube]

R.G. will like this!  

[ExpAnonColin]

I suspect that it is using 8-bit A/D which will limit the dynamic range, i.e. the noise floor will be higher than you would like.

Who said 8-bit noise is a bad thing?

Coming soon, the lo-fi 8-bit delay...

-Colin

[R.G.]

I suspect that it is using 8-bit A/D which will limit the dynamic range, i.e. the noise floor will be higher than you would like.

Yep, it would, and that's why you don't already have a single-PIC delay design from me.

As a rule of thumb, each bit of digital precision adds 6db of signal to noise.

So with one bit quantization - a comparator! - you actually get 6db signal to noise. You can actually hear and understand speech which has been "quantized" with a comparator, for instance.

But 2 bits = 12db, and so on to 8  bits, which is 48db s/n. This is noticeably music, but ugly.

60 db s/n is what you got from cassette tape without fancy noise reduction, and that's just tolerable for on-stage music. That's ten bits. Anything less than 14 to 16 bits is unusable for recording.

In case you watch the PIC specs, there are cheap ($2) PICs with ten-bit A/D, but the sample and conversion time as well as the overhead of manipulating external memory in isochronous code segments is not currently do-able. But they will get there some day.

I recently found an A/D that costs under $4.00 that can be controlled by a PIC. It needs some static RAM and some external logic chips, but I think it will get to 16b at 32kHz sampling.

The AC97 A/D/A codecs are possble, but as a rule, they cannot eat their own output bit frames, so they're not currently suitable for direct write/read memory delays.

I've ..um.. been watching this corner for quite a while now.  

[gez]

I recently found an A/D that costs under $4.00 that can be controlled by a PIC. It needs some static RAM and some external logic chips, but I think it will get to 16b at 32kHz sampling.

EPE mag had a PIC based chorus/flange etc design a few years ago that used external memory for the delay.  It was called the 'Polywhatsit' if I recall.  That was also pretty lo-fi.

Yup, the Cybervox is just 8bit (as I mentioned in my intitial post - pay attention at the back you boys!   ) and considering it's being used in a Dalek emulator circuit you wouldn't expect it to be hi-fi, but it's really simple and, for the likes of me, something to whet my appetite to learn more about PICs.

[Gladmarr]

Y'know, Microchip makes the new DSPic.  I think that may make a single pic-based delay a more feasible reality.  I don't know anything about them, but I have seen some info on them, and they do list music effects in their propaganda for them.....

[puretube]

anybody else here missing Peter S., too :?:

[gez]

anybody else here missing Peter S., too :?:

I believe the Australians refer to his condition as '**** struck!' (in love) :lol: