PT2399 based delay. Decrease delay output

[nonost]

Hi! I've increased the cap value at pins 9&10 from 100nF to 470nF. That way, the signal noise ratio is better. However, the delay output is LOUD. So loud I have to keep the repeats pot at minimum with a millimetric range of use. If I turn the pot up just a bit, it goes into oscillation like crazy with infinite repeats collapsing.

Here the schematic



I guess I have to decrease the output delay level at the PT2399 mixer, right before the final TL072 mix stage. What resistors should I try with? R12, R13 & R14 look as good candidates to me, but I'm not sure at all.

Thanks!

[nonost]

R9?

[ElectricDruid]

If the delay output is too loud, increase R7. If the Repeat pot causes oscillations too early, increase R9. Or both, if that helps!

[nonost]

 Ey Tom thanks!

Yep, increasing R9 did the trick with the repeats, now it is tamed. The wet signal is louder than the dry signal, so I'm increasing R9 a bit.

Do you know how to increase the effect output signal? The bypass signal is a bit louder and it's annoying me a bit. I'm also hearing the bypass signal somehow brighter, but that might be because it's louder. Besides the 100pF caps, is there something else I can tweak to make the dry signal brighter?

Cheers!

[ElectricDruid]

The wet signal is louder than the dry signal, so I'm increasing R9 a bit.

You mean R7, but yes.

Do you know how to increase the effect output signal?

Increasing R5 would boost the level of the whole effect, dry+wet, everything.

They've set it up with 22K input resistors (R4+R7) against a 10K feedback resistor (R5), so the gain is slightly under x0.5. The logic is that two signals at half the amplitude add to make one at full amplitude. Except it doesn't work quite that well when one signal is delayed.

I'd guess you need something around 15K for R5 to get unity-gain-ish.

The bypass signal is a bit louder and it's annoying me a bit. I'm also hearing the bypass signal somehow brighter, but that might be because it's louder. Besides the 100pF caps, is there something else I can tweak to make the dry signal brighter?

100p/330K on the input op-amp IC2_A is a lowpass rolloff at 4.8KHz, so no wonder it sounds dull. Try 47p or 33p instead.

The 100p cap on the output mixer could even be *larger* without affecting anything. 10K/100p is a cutoff at 159KHz, which is way too high. You could change that for 470p without any issues, and perhaps with less high frequency digital noise in the output.

Aside from those 100p caps? No, you're in the right place, that's where you need to be looking.

Cheers!

You're welcome!

[nonost]

Yes, you are right. I meant R7 for the delay output level.

Increasing R5 to 15k gives you a slight boost over the bypass signal. The 10k is a better match, though. I guess a 12k resistor would be ideal but I don't have that value at this moment.

Ok. I will try with these 100pF caps.

BTW, the tone control barely works. It does nothing until around 3:00, at this point the signal gets darker and repeats go crazy into oscillation. I think I have to adjust another value around the tone control section, I guess the 470nF at 9&10 that I'm using is the cause. I've  built this circuit in the past(stock) and the tone control worked pretty good.

Thanks a lot Tom! This thing is almost completely tamed! 

[anotherjim]

I think the Tone pot should be 100k. As it is, C14 works with R15s 10k at pot minimum or 20k with the pot maximum. That'll be a very small sweep over the range of the pot.  Also, C14 at 100n is going to be pretty dark. This may be intentional but I would try 10n or lower with the 100k pot.

[nonost]

Hi Jim! I built this circuit stock in the past and the tone pot seemed to work pretty good. But with the big 470nF at pins 9&10, somehow it does affect it and it's useless now.

I will follow your advices, though. And let you know.

Thanks!

[anotherjim]

Yeh, the tone filter could be meant to be cutting treble from a very low roll-off frequency as drawn. It's only a single-stage filter and might be simulating a worn-out tape echo.

[duck_arse]

http://www.valvewizard.co.uk/smalltime.html

valve wizard has an augmented datasheet for the PT2399 linked on this page, it mentions those integrator caps and values.

[nonost]

I think it's intentional, Jim. It sounds quite dark with the 100nf, but it's a great sound. Not defined but cozy. Pretty useful for more ambient sounds.

Which resistor should I increase in order to lower the repeats level, R12 or R13? At this point the signal is quite above unity and I want it to reduce the level before it gets mixed at the pins 13&14 of the PT2399. I know I can do it at R7 but I'd like to do there also.

Cheers!

[anotherjim]

R8 and R9 are mixing the input with the repeats into the delay. I'd try increasing R9 for weaker repeats. The mix is unlikely to match with R8 & R9 at equal values as they don't come from equal sources so R9 could be OK for you at 22k for all I know.

[nonost]

There are a couple things I don't undestand...

The input opamp gives the dry signal a gain factor of 2. 360k/180k = 2. Okay. It goes to the output mixer.

But the dry signal that goes to the PT2399 is taken back to unity gain right at the pins 15&16. I put the values in this calculator:

http://sim.okawa-denshi.jp/en/OPttool.php

The 22k, 10k, 10k, 10nf & 2.2nf give -0.45 gain and cutoff at 3.4khz.

The thing gets more interesting because at pin 11&12 the processed signal is amplified by a gain factor of 2. I don't get it. At this point the signal has been already converted to digital and then back to analog...Of course this x2 wet signal goes to the mixer. In this very mixer, both dry&wet are reduced by half to return the global signal to unity gain.

I don't understand the point of amplifying the signal ONCE it's already wet. If I'm not wrong what makes sense is to amplify before the AD/DA proccess and then reduce the signal afterwards. So you get a better signal noise ratio.

What am I missing here?

[ElectricDruid]

You're assuming that the design is entirely logical, which may not be the case. Often we see circuits that have been built out of parts taken from elsewhere, or circuits that have been heavily tweaked in ways the original designer never intended. So (for example) someone sticks a filter in, without realising that the filter has a gain of x0.45, and then finds the output is too quiet, so they tweak some resistor values on the input or the output to boost it. Whereas if they'd designed the filter from the ground up, they would probably have used a unity-gain filter.

So...I'm not surprised if it doesn't make a lot of sense, and I don't think you're missing anything.

The question is how much work do you want to do to get it how you want it?

[nonost]

I see...

At this point I'm just learning and trying to understand stuff more than anything else. I didn't have basic knowledge about opamps and filters that I'm starting to get it now. Reading and searching a lot about the pt2399 and so on. BTW, thanks Tom. You are helping me quite a lot, and the info in your web is great. The one in electrosmash is top as well.

I'm making the first op amp a non-inverting buffer and then tweak the resistors accordingly. Maybe I put a bit of gain right at the PT2399 input and then reduce it at the output.

Is there any difference between inverting and non-inverting buffers regarding impedance or such? I know it would 2 more resistors to make an inverting buffer, but besides that, any advantage? I know that in that case the resistor in series with the inverting input should be low value in order to not introduce noise. Since the opamp output goes right to the PT2399 input, I wonder what would be the best option in this terms (impedance or anything magical else that I'm not aware )

Cheers!

[antonis]

I know it would 2 more resistors to make an inverting buffer,

Actually, it would need one resistor less..
(A big value resistor between VR and non-inverting input which, in parallel with R1 should set  the input impedance..)
Unless, of course, you wish to take care of inputs offset current..

Since the opamp output goes right to the PT2399 input, I wonder what would be the best option in this terms (impedance or anything magical else that I'm not aware )

Both configurations output impedance should be considered negligible..

[ElectricDruid]

Is there any difference between inverting and non-inverting buffers regarding impedance or such?

In general, it's much easier to make non-inverting buffers with a high input impedance than inverting buffers. If you increase the resistor values for the inverting buffer (say, to 1M) things start to get noisy, so it's not a great idea.
However, this only matters at the very front of an effect pedal circuit. Somewhere in the middle, it won't matter much since you won't need super-high impedance and you could easily do the inverting buffer with 10K resistors.
The output impedance of either is very low (it's straight out of an op-amp's output in either case) so there's nothing in it from that point of view.

PS: Thanks for the kind words about the website. Glad you've found it useful.

[anotherjim]

Yes, they all seem to start copying the datasheet examples for the PT2399. Some extra bits added are lossy and need some make-up gain. We might not be able to put more gain in the input to avoid clipping it, especially when feedback is added in - it only has 5v supply so less headroom.

[nonost]

You are right antonis, I forgot the typical 1M resistor!

I'm going to try the led limiter from pin 7 to gnd.

There are a couple filters I don't get. I have problems regarding where the signal goes...R14(1k) and C10(10nf) form a low pass filter. Does R12(22K) also counts here? If it doesn't, the filter is quite useless since 1k + 10nf is centered at 15k.

R16(1k) forms another LPF with C12(47n). Does the value of the pot Feedback count as well? I think it doesn't, but I'm not sure. Also, each repeat will darken more than the previous one, because of that filter. Which is center at 3.39kHz. So at some point, there won't be anything above that 3.39khz.

[nonost]

Hi. I've been studying. I've notice that people are using the MFB at pins 13&14 and 15&16 in order to add/reduce gain as well as low pass filters. Without having into account the Q factor of the the filter itself.

The first MFB at the De Profundis shows a Q = 0.86. And the one at pins 13&14, Q = .58. The latter is Bessel, whereas the former is somehow a Cheby with 1db ripple or so.

More than .707 is going to add gain to the frequencies before the fc.

I wonder if this was on purpose or it was not...I doesn't make a lot of sense to me. The first filter "boosts" the range close to the fc, which is 3.4khz BTW. That doesn't sounds good.

The second one looses a bit of good midrange...

I have a doubt regarding how to reduce signal gain. I'm using a big cap across pins 9&10, this cap boosts the signal. How can I reduce the repeats gain without using the MFB filter? I don't want to mess with the response. I need -12db so my filter would make a big hump in the midrange close to the cut off frequency. Not nice. I tried a voltage divider before and after the feedback pot, and although it works (4k7 with 2k to ground) it somehow alters the pot sweep and I only get 3 repeats with the feedback pot halfway. In the last 20% is where all the fun stuff lives.