Rediculously modded PT2399 Delay

NOC3 · Post by **NOC3** » 05 Feb 2009, 01:06

FF's magnus modulus thread inspired me to mess around with the pt2399 circuits again. I drew up this schem a long time ago but today I updated it with some of the new ideas. I thought that the trem on the magnus modulus was a bit strange because it affected both the delay and dry signals, effectively acting as a tremolo after the entire circuit. I thought it would be cooler to make it so that the tremolo only affected the wet OR the dry signal. The trem in this circuit only subtracts from the signal though, so it may seem that it drops the level a bit...maybe that can be changed somehow, I chose to just keep it simple though.

Inspired by just about every PT2399 delay out there, companding like the PT80, Trem and Modulation like the Magnus Modulus and SWTC2 as found on AMZ.

Features:
Up to 1 sec of delay
Companding for more authentic analog sound and lower noise
Charge pump w/12 volt regulator for higher headroom.
Tone control for repeats
Pitch modulation for repeats
Expression pedal jacks for repeats and level
Buffered FX loop for clean or delayed signals, FX send can also be used as a stereo out
Tremolo switchable for only clean or delayed signals

Enough talk here's the schem. I left out the bypass switching.

Please let me know if there are any errors etc. and I'll fix em!

NOC3 · Post by **NOC3** » 05 Feb 2009, 01:24

Oh yeah, I forgot to mention I'll be working on a PCB layout for this tonight also. Board mounted pots and toggles, jacks and bypass switch will be off board. It will need a big box.

It will be an ExpressPCB design, double sided and plated through. Sorry to those who like to make thier own boards, I'm spoiled with these double sided PCBs. At least you have a schem!

A couple notes on the schem:
The Mod and Trem Depth pots may need to be increased to 250K since they are in parallel.

You can alter R21 if the trem is too strong at max, and the LDR should have a high off resistance of 100k or bigger and a low on resistance around 10k. You could also use a fet but I don't like them...However if you have a good way to use a fet in this application, send me a circuit snippet so I can incorperate it in the PCB design as an option.

NOC3 · Post by **NOC3** » 05 Feb 2009, 07:03

AH,
I see now that tremolo probably won't work right as is...It will do a trem sound, but the wet/dry trem signals will interact with each other which isn't what I was going for. I've fixed it and will update the schem here once I'm done redrawing it.

briggs · Post by **briggs** » 05 Feb 2009, 12:38

Looks very impressive NOC3. I've played with the pt2399 quite a bit too and found that integrating it with a sequencer was one of the strangest sounds I've ever created! I look forward to seeing the updated scheme.

pz · Post by pz » 05 Feb 2009, 18:49

Why are you loading the charge pump output with +5V regulator and 2399 chip?
You could just feed the 7805 with 9V and leave the charge pump and +12V line only for the audio path.

NOC3 · Post by **NOC3** » 06 Feb 2009, 01:56

pz wrote:Why are you loading the charge pump output with +5V regulator and 2399 chip?
You could just feed the 7805 with 9V and leave the charge pump and +12V line only for the audio path.

Good catch, fix'd.

OK, I've updated the schematic. Everything should work right now.

I changed up the switching for the trem, ditched the LDR and used FF's original Magnus Modulus trem circuit with the JFET. But then I had an extra half of the op-amp to use so I replaced the discrete FX loop buffer with an op-amp buffer. Then while I was at it I figured what the hell, why not give the tremolo its own LFO? So now you can have a subtle modulated delay and set the trem as deep as you want!

The layout is coming out pretty well, looks like this is going to be a hammond 1590D sized monster.

NOC3 · Post by **NOC3** » 07 Feb 2009, 07:40

Ok, so I think I'm done with the layout. I just have to double check EVERYTHING which could take a while. If you would like to donate the use of your eyes, mine could use a rest. As you can see the schem has a couple updates, I numbered all the pins of the various connections to correspond with the layout silk screen... I had some room left over so I added a couple parts to make it stereo and I added pads for an oscillation switch.

Here's the mechanical drawing of the top of the board with all of the pads and silkscreen. The circles on top are just to display where the top pots will be, the other 5 pots will be underneath the board. The pots are 16mm Alpha horizontal board mount style available from smallbear. The DPDT toggle switches are board mounted, Mouser P/N: 108-1MD1T2B3M2QE-EVX

I'm thinking these boards will be about $15 each. The PCB is 6.5" long by 1.75" tall.

tuemmueh · Post by **tuemmueh** » 09 Feb 2009, 00:04

soulsonic · Post by **soulsonic** » 09 Feb 2009, 01:28

How does the delay time modulation work? I'm looking at it and I don't get it. It just has the LFO connecting to pin 6, but how does the delay time get modulated if there isn't a variable resistance? Am I missing something? Is just injecting a signal there enough to make it modulate?

Also, why run it on only 12v? The charge pump is making 16 to 18-ish volts depending on the diodes used; why not just use that and not throw away extra energy on the regulator? Or, if you really want it regulated; use lowest-drop diodes and tantalum caps for the voltage doubler with a 15v LDO regulator so you get so you can squeeze 3v more headroom.

earthtonesaudio · Post by **earthtonesaudio** » 09 Feb 2009, 01:42

The LED is the "variable resistance." Not the most efficient way of doing things, but it should work.

soulsonic · Post by **soulsonic** » 09 Feb 2009, 02:09

Okay, I was wondering if that might be it... thanks!
Yeah, that wastes alot of current, doesn't it?

earthtonesaudio wrote:The LED is the "variable resistance." Not the most efficient way of doing things, but it should work.

earthtonesaudio · Post by **earthtonesaudio** » 09 Feb 2009, 03:07

I meant inefficient not so much in terms of current consumption, but the fact that you have an oscillator running at 9V to control a pin which can't respond to anything except 0-2.5V. It looks like a square wave oscillator so it seems like you're generating moderately large transients unnecessarily.

Actually I was wrong about the LED acting as a variable resistor. (Fiddlesticks!) I think it's just there for a rate indicator, the LFO is coupled through the 10uF capacitor and the AC waveform is superimposed on whatever DC level is present at the delay time pot. No need for a voltage/resistance or voltage/current transformation. AC works just as well as DC when you're talkin' LFOs.

soulsonic · Post by **soulsonic** » 09 Feb 2009, 03:42

I don't get it. The datasheet I have doesn't seem to describe the workings of Pin 6 well enough for me to understand how injecting an AC signal into it is supposed to modulate the delay time. If it were that easy, why don't hardly any of them work that way? This is the first one I've seen with an AC signal connected directly to that pin. Every other one I've seen has used the LFO to drive a variable resistance like a transistor.

soulsonic · Post by **soulsonic** » 09 Feb 2009, 03:53

Okay, I see that the Magnus Modulus has it done like that.
But, I still don't get how it works.

earthtonesaudio · Post by **earthtonesaudio** » 09 Feb 2009, 04:45

Well, Princeton Technologies is unlikely to disclose enough about their chips to get people like us tweaking them till they burn up, so they just say "use a resistor of x value." But from measurements people have taken and posted online, one could surmise that it's a medium impedance current source. The pin sits at 2.5V nominally, but if you sink enough current from it, the voltage drops (but not right away, and not dramatically). Because of this, my opinion is that a transistor current sink is the most elegant way to control this pin, so long as you know the current limits (which PT is rather quiet about, again, to keep us tweakers out of trouble). If you took this approach, the modulation could be much lower-voltage/lower power and still get the same effect (and indeed possibly be even more flexible).

But that's neither here nor there. For adding modulation, AC-coupling an LFO seems to work because pin 6 is primarily current controlled. So long as the coupling capacitor is large enough for the LFO, and the driving stage is low enough impedance, the PT2399 chip shouldn't care if it's alternating or direct current.

By the way, props to NOC3 for tackling, and sharing, such a massive project. I wouldn't critique it as I am, if I didn't like it. Oh, and C31 is backwards.

KB · Post by KB » 09 Feb 2009, 09:33

Hi

This is an interesting project..
I am certainly no EE, but I am with soulsonic on this one....

soulsonic wrote:Every other one I've seen has used the LFO to drive a variable resistance like a transistor.

I also have only seen modulation of a PT2399 via an LDR or transistor. Look at the Echo Base for a transistor driven modulation that has been shown to work... I am not saying directly driving pin 6 from the modulator will not work... I will try it out, when I get a chance.

Cheers

Kevin

~arph · Post by **~arph** » 09 Feb 2009, 12:53

Well either way seems to work fine, just listen to the magnus modulus. I actually think that it sounds better using an AC coupled LFO..

NOC3 · Post by **NOC3** » 09 Feb 2009, 18:02

Thanks for the replys guys.

Concerning the modulation LFO, I lifted it from the magnus modulus after hearing the clips.

The LFO is a modified triangle wave, not a square wave. If you wanted it square just reconnect pin 5 of U2 to pin 3 of U2 on the other side of that 470k resistor. Tremulus lune LFO. C32 and C46 may need to be 100uf to alter the range depending on how slow or fast you'll want it...you could also use bigger rate pots

As far as the voltage goes, check the datasheet for the SA571. Maximum voltage 18v. I don't like running things close to their absolute maximum ratings. It might be ok, it might not, but one surge and you are done for sure though. 12v is still a big improvement over 9v, it would take a pretty big signal to cause any clipping. If you REALY wanted 18 volts though, just jumper pins 1 and 3 on the 12v regulator.

A couple more things I'll add will be:
1- Pads for a "Dry kill" switch, that will cut the dry signal coming out of the mono output.
2- A sister board for an option to add a second PT2399, with pads for connecting it to the main board, running them in series will give much cleaner long delays. Just use a fixed resistor for the second PT2399, half the "course" delay time pot value to 50k, and replace the "fine" pot with a switch to bypass the second PT2399 and you have long and short times that sound the same with no further degradation of the repeats!

noelgrassy · Post by **noelgrassy** » 09 Feb 2009, 22:49

Thanks for all your hard work and especially sharing!!!

I'll gladly buy at least 2 boards when you go into production. I'm just a solder jockey so I can't
critique diddly. Dammit I know good worksmanship when I see it though.

Analoguru coined that Solder Jockey phrase in what was a rare condescending moment I'm sure.

Lemme know,

NG.

NOC3 · Post by **NOC3** » 10 Feb 2009, 02:15

Believe me I'm just as excited about this as you guys.

Schematics are updated once again....now I'm DONE! Unless there is an error somewhere, which I don't think there is. Please check it over anyway though.

The Long/Short Delay switch works by putting another PT2399 in series with the first one and taking the output from the second. If the delay time pot resistances are the same, the long setting will be almost exactly twice as long as the short setting, usefull for switching from quarters to 8th notes. It does require a dual ganged pot but that is really the easyest /most usefull way to control the delay times in this setup. I've used 50k, as that still gets you about 1100ms delay time...but you could go up as high as 100k there for insane delay times (however the delays will get really dirty when they are that long). The biggest advantage though is that because the delay time is shared between two chips, the long repeats are just as clean sounding as the short ones, or you can even leave it on "long" but set the delay time short for really clean delays.

I'll probably use footswitches for the delay bypass, modulation bypass, trem bypass, and a momentary footswitch for the oscillations. But you can put whatever you want on toggles. So 4 footswitches spread out over 7.3 inches...not too cramped.