Underclocking PT2399

[Taylor]

I built an Echo Base using a 1Meg pot for time (subbed for the spec'ed 50k). This gives up to 10 seconds of (increasingly bit-crushed, bandwidth-limited) delays, which I had a lot of fun with.

Until the delay cut out and I was left with just clean signal. A bit later (many minutes later) the delay cut back in. I'm pretty certain this is not an issue with my construction, so I'm thinking there is an inherent issue with underclocking the PT2399. I asked about it in the Echo Base thread, but no luck. I also think it's better for questions to have their own thread than being buried in a 30-page thread.

Anybody have any thoughts as to why this would be? I remember it was unclear how the chip derives its delay time from the resistance at pin 6, but I can't figure out why this would be bad, and especially why it stops working, then starts working again with no change in settings, etc. Resetting power does not seem to fix it either.

[earthtonesaudio]

Probably the oscillator just stops until enough charge accumulates on the internal capacitors to start it up again, but that's just armchair speculation with no basis in anything proven.

Fascinating though!

Did you happen to note at what resistance value this occurs/the exact time it takes to start working again?  I don't know how helpful that info would be, but it may reveal more about the mysterious inner workings of the PT2399 chip.


[Side note: what are/ what can you do with the CC1, CC2 pins of the PT2399?  I've never seen anyone do anything with them except exactly what's shown in the datasheets.]

[Taylor]

Reading about, seems CC0 and CC1 are for "mod" and "demod". Not sure what this means in this case. Perhaps someone could try swapping the caps there and see what happens?

Found some mention of my problem here. Problem is that they're using a different schem and PCB than I am. Also, in my case, the delay doesn't just stop at a certain resistance. I can turn it up all the way to 1M and it delays for some time, then stops.

[Mark Hammer]

The Princeton chips either come with on-board DRAM, or else address lines for using external DRAM.  For those of you too young to remember, DRAM requires "refreshing".  That is, unlike static RAM or flash memory, it does not simply "hold" the data as long as power is supplied.  The data needs to essentially be continuously rewritten in order to exist.  My guess is that when you underclock enough, at a certain point you go past the maximum acceptable refresh period, and lose all data stored in RAM.

[earthtonesaudio]

Thanks Mark, that makes perfect sense!

[Taylor]

The Princeton chips either come with on-board DRAM, or else address lines for using external DRAM.  For those of you too young to remember, DRAM requires "refreshing".  That is, unlike static RAM or flash memory, it does not simply "hold" the data as long as power is supplied.  The data needs to essentially be continuously rewritten in order to exist.  My guess is that when you underclock enough, at a certain point you go past the maximum acceptable refresh period, and lose all data stored in RAM.

Interesting, but in my case, once the signal cuts out, I can turn the delay knob back up to a very short time, and it still doesn't work. It won't allow me to write new information at a high speed once the delay signal stops, at least for a while (maybe 10 minutes, but always different).

[Mark Hammer]

I imagine you've disturbed the timing overall.

[R O Tiree]

Could someone do me a favour and measure the voltage at pin 6 with a 50k pot at 0% (shorted) and 100% (fully open), please? I have some thoughts...

Thanks.

[frequencycentral]

Could someone do me a favour and measure the voltage at pin 6 with a 50k pot at 0% (shorted) and 100% (fully open), please? I have some thoughts...

Thanks.

Very very little difference at all. A few millivolts.

[R O Tiree]

Thanks, Rick.