Some useful data on the PT2399

[merlinb]

Today I spent some time testing the PT2399. In particular, I was interested in the VCO control pin interface. It turns out to be current sourcing, and it can be driven directly from the output of an opamp or transistor. I also did tests of headroom (with no input/output filters applied), so that optimised filters can be devised. I've compiled the data in the following PDF, in case it is of use to anyone:
http://valvewizard2.webs.com/PT2399_Data_Notes.pdf

Also, I believe it used delta-sigma ADC and DAC, but I could be wrong. Does anyone know what purpose pins 7+8 might serve? (other than just being decoupled!)

[slacker]

Interesting stuff thanks for taking the time to do that. I think it uses Delta modulation rather than delta-sigma, there's thoughts on pins 7 and 8 here http://www.diystompboxes.com/smfforum/index.php?topic=86297.msg724260#msg724260

[earthtonesaudio]

I also appreciate the idea of putting all the pertinent info for hardcore DIYers into one document.  Thanks for getting the ball rolling, Merlin!

Today I spent some time testing the PT2399. In particular, I was interested in the VCO control pin interface. It turns out to be current sourcing, and it can be driven directly from the output of an opamp or transistor.

This page shows one way of implementing control over delay time, using a voltage-controlled current source.  It's a bit more geared toward synthesizers but still interesting.

Also, I believe it used delta-sigma ADC and DAC, but I could be wrong.

I agree with Ian.  The placement of the integrator makes this a Delta Modulator.

Does anyone know what purpose pins 7+8 might serve? (other than just being decoupled!)

My belief* is that the "mod" and "demod" sections (connected to pins 7 and are spdt switches and therefore when capacitors are connected from pins 7/8 to ground, they form switched-capacitor resistors.  These synthesized resistors in turn are used with their respective op-amps to make integrators.  The value of these resistors is approximately R=1/(f*C) where f is the clock frequency and C is the value of the capacitor.  You can make these capacitors smaller to reduce the gain of those op-amp sections or lower the average cutoff frequency of the integrators.
From here the voltage at pin 9 is used as a reference for the input comparator in the ADC.

I also did tests of headroom (with no input/output filters applied), so that optimised filters can be devised.

I'm also curious what the clock frequency was when you measured the audio frequency response.  The nature of the chip is to reduce high frequency gain as the clock frequency decreases, so the bandwidth is not a constant.

Delta Modulation
Switched capacitor resistors


*This is just a guess but I think it makes sense.

[merlinb]

I'm also curious what the clock frequency was when you measured the audio frequency response.  The nature of the chip is to reduce high frequency gain as the clock frequency decreases, so the bandwidth is not a constant.

That didn't occur to me at the time. I think it was running about 2MHz. I will do more tests at other settings!

[puretube]

I wouldn`t wanna switch 100n  22 million times/second...

[anchovie]

I don't know if this will give any further clues over its internal workings, but with the Noise Ensemble (where I use OPAMP1 as a comparator) I had to leave out the cap on pin 8 otherwise there was an increase in noise. Same with the filter cap between 9 & 10.

[merlinb]

I'm also curious what the clock frequency was when you measured the audio frequency response.  The nature of the chip is to reduce high frequency gain as the clock frequency decreases, so the bandwidth is not a constant.

That didn't occur to me at the time. I think it was running about 2MHz. I will do more tests at other settings!

OK, I have repeated the test at different clock frequencies, and althoug the bandwidth does indeed shrink as as the clock reduces, the differences are almost negligible in the 1-20MHz range.
Even at 0.58Mhz the headroom is much the same, except that the output bascially vanishes (suddenly) for audio above about 5kHz.

[merlinb]

I have updated the data notes to reflect my latest experiments.
http://valvewizard2.webs.com/PT2399_Data_Notes.pdf

[Brymus]

Thank You Merlin.

[caress]

Thank You Merlin.

thumbs up!

[asatbluesboy]

Quite an old thread, I know, but I've been wondering... According to these notes I don't need to use pins 13/14, now do I? I ask because so many people complain about how noisy PTs are, and how that is partially due to all its sections getting Vref from the same place and its throwing all sorts of crap on it, that I thought maybe an extra voltage divider and an extra single op-amp could help taming it a bit -- obviously at the expense of a tighter layout, a current draw increase as well as an increase in overall complexity which could make it even noisier or add problems of its own. I'm an idiot with an experience in electronics which is more empirical than anything, so I could just be being stupid -- I am actually likely to. Thoughts?

Edit: as a matter of fact, I could just skip pin 16 as well and use one of a low-noise dual op-amp's inverting inputs and hook its corresponding output to the PT's pin 15, couldn't I?
Edit 2: if that is the case, should I ground pins 13, 14 and 16 or NC is enough?

[syntaxera]

wow!  thanks a lot for this....   should come in handy

[earthtonesaudio]

Unless you're willing to explore uncharted territory I would stick with using pin 16 as the input.  It *may* be possible to use pin 10 or even pin 8, but that's not certain.

You can easily avoid using the op-amp between pins 13/14 though.  Since it's an op-amp, it shouldn't be left floating or tied to ground, but its inverting input should be tied to its output.  While you can usually short these 2 pins safely, it's a more conservative idea to use a small-value resistor.  That way, if you etch a board but later change your mind and want to use that op-amp for something, you have 2 pads instead of a trace, which makes after-the-fact additions much easier.

[asatbluesboy]

Thanks a lot, Alex!