Rat with Dual Opamp – mind helping with the last section?

[antonis]

...and one more question – will a simple low pass on the clean work/be enough?

1. Why you think you need a 30kHz LPF on a buffer output..??
2. IC2.B LPF corner frequency is 1.6MHz..!!
3. Is there any particular reason for C16 to be wired to Vref instead of GND..??

[T-Dub]

Hopefully this is cleaned up a bit 


Really appreciate you taking the time here.

I *think* that the clean blend is what you suggested now – 1k off each output, R15 and R14. I'm guessing I should get rid of c12 so the mixing IC doesn't need a bias?

R15/C16 were mean to be a low pass at ~200Hz on the clean so it would slope off the high frequencies if I was using this for bass...fixed the dodgy RC there. Was hoping to put the C16 on a switch to engage for chopping the top off, or even putting a clean tone pot to blend between caps there.

Also changed c13 to be an LPF at ~8K to smooth the top end of the drive out...that's what I'm hoping anyway!

This is slowly becoming less alchemy and science so I highly appreciate your much more trained mind on this one.

[ElectricDruid]

Looking much better.

You've got bias for the final blend buffer IC1B coming from the input buffer IC1A, even with C12 in place. There's a direct DC path from the output of that chip (which is biased at Vref)  through R15 and the Blend pot to the +ve input of IC1B.

R15/C16 will do what you want, but they're also going to hit the volume, especially for higher notes. That might make using the blend tricky, since lower notes will see less cut than higher ones (-6dB/octave). You'll have to try it and see for that. The science can give you the numbers, but the numbers don't tell you whether it sounds how you want it!

[antonis]

Looking much better.

Indeed..!! 

BUT...

IC2.B might be permanently hitting the positive supply rail..
@T-Dub: I let you find the reason and heal it..
(hint: You have to roll-off DC gain to unity..)

[ElectricDruid]

BUT...

IC2.B might be permanently hitting the positive supply rail..

Ooh, yeah, he's right. I missed that.

[GGBB]

Also changed c13 to be an LPF at ~8K to smooth the top end of the drive out...that's what I'm hoping anyway!

As I mentioned earlier, this is not very effective - because the gain of that stage is only 2 so C13 can't reduce gain of high frequencies very much.

If you want to do this more effectively, add the 18n cap after R14 to ground.

[GGBB]

Looking much better.

Indeed..!! 

BUT...

IC2.B might be permanently hitting the positive supply rail..
@T-Dub: I let you find the reason and heal it..
(hint: You have to roll-off DC gain to unity..)

Mentioned earlier:

R8 and R10 are unnecessary. There's no harm in R8 except that it robs about 10% of your output. Without R10, POT1 (volume) can be smaller e.g. 10k (R10 and POT1 in parallel form a single resistor). You probably don't need to add gain to the output buffer, but if you do, R11 and R12 should be bigger like 100k and should go to ground through C15 which should be ~220n. If the plan is to tame high end with C13, it does very little because gain is only 2. But making it significantly larger e.g. 10-100n will reduce most other frequencies as well (not dramatically because gain is low). As you raise the gain of this stage, C13 will have a bigger effect and will have less effect on lower frequencies.

As an alternative to adding C15 (from the original schematic version not the latest) - connect R12 to VR instead of ground.

[antonis]

Mentioned earlier:

But not utilized..

P.S.
Let's be forgiving with OP 'cause, IMHO, he tries to implement discrete building blocks without understanding the way they work..

[T-Dub]

Having a hard time quoting here but yes, patience is greatly appreciated and I am doing my best to understand the building blocks properly

[antonis]

I am doing my best to understand the building blocks properly

IMHO, that's an essential requirement for intentionally making them working improperly..
(e.g. for making an asymmetrically clipping amp you have to know when and how it clips..)

[T-Dub]

Guilty as charged with using building blocks, without quite knowing what's going on yet    I'm learning heaps here though



Ok, so when it was mentioned to roll off DC gain to unity, I *think* that means to add a capacitor going to ground (c13 here), which I think is the same bit of info I accidentally skipped over that GGBB mentioned (I've implemented his suggestion there). Am I right that instead of placing the capacitor to ground I could have put R12 to Vref?

I'm interested by the R12 100k/ C13 220nf combination in values, would that make a LPF at 7.2HZ? Also, what would be the difference between using 100k vs 1k in the R12/R11 positions?

Thanks so much!

[antonis]

Ok, so when it was mentioned to roll off DC gain to unity, I *think* that means to add a capacitor going to ground (c13 here), which I think is the same bit of info I accidentally skipped over that GGBB mentioned (I've implemented his suggestion there). Am I right that instead of placing the capacitor to ground I could have put R12 to Vref?

Yes but only in case of exactly the same voltage on both R12 legs..
(i.e. absolutely no voltage drop across R9..)

[ElectricDruid]

Ok, so when it was mentioned to roll off DC gain to unity, I *think* that means to add a capacitor going to ground (c13 here), which I think is the same bit of info I accidentally skipped over that GGBB mentioned (I've implemented his suggestion there).

Yes. The cap makes a lowpass filter for the feedback, and since it's negative feedback, that has a *highpass* effect on the signal, so DC gain is unity.
Another way to look at it is to think about it "from DC to daylight". At DC/0Hz, all caps are an open connection, so the amp is effectively a unity-gain buffer with the output connected back to the -ve input. At "daylight", infinity-Hz, all caps are shorts, effectively a wire, so we can ignore them and calculate as if they weren't there.
That gives you a quick sketch of what the end points of the frequency response look like. The bit in the middle is what the maths is for!

Am I right that instead of placing the capacitor to ground I could have put R12 to Vref?

From a DC bias point of view, yes, you could do that instead, since the +ve input has that level already, so there'd be no problem. However, that doesn't fix the DC gain problem. It gives you a working amplifier with gain all the way down to DC.

I'm interested by the R12 100k/ C13 220nf combination in values, would that make a LPF at 7.2HZ?

Yes, it would.

Also, what would be the difference between using 100k vs 1k in the R12/R11 positions?

If you change R12, the cutoff frequency would go up by 100K/1K = 100, so from 7.2Hz to 720Hz. The classic Tubescreamer value. If you change R11 as well, the gain doesn't change. So you can tweak gain or the frequency response by adjusting one or the other or both.

[antonis]

I'm interested by the R12 100k/ C13 220nf combination in values, would that make a LPF at 7.2HZ?

Yes, it would.

I think Tom is sloppy enough lately..  

[ElectricDruid]

It's in a feedback loop, so call it whatever you like. It's either a lowpass because that's what it does to the feedback, or it's a highpass beause that's what it does to the signal.

My point was that T-Dub's cutoff frequency calculation is correct, which it seems to be.

[T-Dub]

It's in a feedback loop, so call it whatever you like. It's either a lowpass because that's what it does to the feedback, or it's a highpass beause that's what it does to the signal.

My point was that T-Dub's cutoff frequency calculation is correct, which it seems to be.

Really blew my mind that (of course) the lpf acts as a hpf due to it being negative feedback.

[antonis]

Really blew my mind that (of course) the lpf acts as a hpf due to it being negative feedback.

Same as C8/Dist pot HPF, acting as LPF there..

P.S.
It should be more convenient to memorize the most frequently existed filters around an op-amp configuration..



edit: C6/R5/R4 HPF might seem puzzling (same for R7/C5/Gain pot lug3-2) but simply add resistor values for equivalent R in 1/2πRC corner frequency formula..

[ElectricDruid]

Really blew my mind that (of course) the lpf acts as a hpf due to it being negative feedback.

Yeah, me too! The first time I actually understood what was going on there was definitely one of those "Oh wow!" moments.

[PRR]

one of those "Oh wow!" moments.

Nobody remembers the reason they call them Operational Amplifiers.