Flatline Optical Compressor - Blend Mod verification (not 4:1)

[slacker]

3-  Feed IC1A(CompOpamp) with signal before IC2A(FirstBuffer) and raise the gain of IC2A. I think that will work, BUT it will increase noise 

I'd do this, which I think is what Merlin meant, any increase in noise shouldn't be anything to worry about.

[samhay]

If you don't like Merlin's suggestion, do you have room for a dual-gang blend pot?

If so, you could set up the blend as two opposing volume pots - something similar is used in some dual pickup guitar wiring schemes - and then use different value resistors from the 2 wipers (smaller for dry signal) into the mixer op-amp.

[LaloFP]

3-  Feed IC1A(CompOpamp) with signal before IC2A(FirstBuffer) and raise the gain of IC2A. I think that will work, BUT it will increase noise 

I'd do this, which I think is what Merlin meant, any increase in noise shouldn't be anything to worry about.

Yes he edited his comment

Update: Soo, I tried that option. Its true, the noise shouldnt bother, we need to amplify the signal just a bit.

But, for the moment I failed. I can get an equal sum of the signals, both with same volumes, and that is awesome.

BUT, Im not sending the input to the CompOpAmp properly. I get less compression or I get weird highs reduction.

Adjust gain of IC2A until you have equality.



EDIT: I meant to say IC2A

If I Just run a wire, the compressed signal is almost not compressed. If in that configuration, I cut the connection to pin 3 of CleanOpAmp (so it removes the CleanOpAmp), I get the compression and more loudness. So I assume that the CleanOpAmp is stealing signal from the CompOpAmp (and I dont know how to balance it right)

I also tested running the same C1(220n) and R2(1M) before C4(input cap), but that gives me compression with loss in highs. I also tested values for R2 up to 10M and it seems to get better, but is all ear speculation.

I googled a lot yesterday about splitting signals to two opamps and biasing two opamps and I could find something that apply to this scenario. I dont want to add a transistor or another opamp stage.

What can I look at to get it right "in numbers"? I mean, I want to understand and adjust not by ear first, but with the understanding of the theory and calculations involved here.

In which way can I configure the splitting of the signal to feed both OpAmps right?

Thanks guys!!!!!!!

UPDATE: So I was looking at this and after a lot of search and test, I came to this conclusion: (please tell me if Im wrong!!)
The guitar pickups are a high impedance source to just split it in two opamps with plain wire and expect it to work right.
The opamps inputs are very high impedance too, so I supposed it could work.
Maybe it works but due to the pickups high impedance the opamps need to be setted to have greater gain. But if I do that and then I put the Pedal after other effects with low impedance it would rice the levels too much.


Is that right?

So, what Im planning to do now:
1 - Improve the pot-between-two signals to reduce the Wet signal before Blend allowing 100% Wet signal mix
2 - Improve the Panning for Fun to not get louder at Half Blend (but I think its less likely)
3 - Feed IC1A(CompOpamp) with signal before IC2A(FirstBuffer) and raise the gain of IC2A. I think that will work, BUT it will increase noise
4 - Move the final opamp between the CleanOpamp and the Blend pot, the get the same volume as the Wet signal (I think I will have enough volume to put a passive volume control after that).

What do you think?? I will test it in a second

[aion]

I'm working on a version of the Flatline very similar to yours in concept, with a clean blend. I needed 2 opamps for the compression portion, one for the clean side and one for the blender. To avoid a third DIP8 package, I used a JFET buffer at the beginning to split the signal. It works great.

[merlinb]

UPDATE: So I was looking at this and after a lot of search and test, I came to this conclusion: (please tell me if Im wrong!!)
The guitar pickups are a high impedance source to just split it in two opamps with plain wire and expect it to work right.
The opamps inputs are very high impedance too, so I supposed it could work.

The TL072 input impedance is so high you can feed several and the pickup can't tell the difference. No problem there. (Although it might be a good idea to reduce R9 to 4.7Meg or even 1Meg)

Maybe it works but due to the pickups high impedance the opamps need to be setted to have greater gain.

The stock flatline goes straight into the gain-controlled opamp, and that's what you're doing here too, so that signal path should work fine.
You also have another bog-standard opamp for clean gain, so that signal path should also work fine. And you're blending the outputs together. It couldn't be simpler.
The fact that it doesn't work right tells us you have done something wrong. Time to measure voltages; are the opamp outputs both at 4.5V?

[LaloFP]

Adjust gain of IC2A until you have equality.



EDIT: I meant to say IC2A

So I tested it again and it worked with 1M on R9 and 220n in C4 (I made a mistake before with C4), but If I just cut the connection with the CleanOpAmp I get a bit more output, and I can measure it in the DIY Vactrol looking at the minimum resistance when hitting hard the guitar.

But I cant resolve these two concerns:
1 - What are the advantages of buffering the input before splitting? The Splin´n´Blend and the PanningForFun implementation do that. Is it just necessary when you dont know the input impedance of the effect where the signal is going? So in our case because are only two opamps with high impedance isnt necessary? Isnt Aion´s JFET adding any advantage?

I'm working on a version of the Flatline very similar to yours in concept, with a clean blend. I needed 2 opamps for the compression portion, one for the clean side and one for the blender. To avoid a third DIP8 package, I used a JFET buffer at the beginning to split the signal. It works great.

2 - I know the benefits of summing signals with an Opamp, but I red somewhere (I cant find it again) that when you are summing two opamps outs, the simple potblend method works because one opamp is like virtualground to the other opamp. Is it right?

-----

I get a sense that splitting without a buffer isnt good and summing without an opamp isnt good, so to make it perfect in paper it needs 5 opamps (like aion´s). So if  I use just 4 Im going to sacrifice the summing or the splitting. I dont want it just to work, I want to understand it 

[blackieNYC]

I have a compressor I'm very happy with after putting in a blend control for only the highs of the dry signal. Keeps the signal from getting dark-sounding while still giving you the option of a fully compressed tone. Is this what most people are really looking for when they consider a blend of the dry signal. 

[merlinb]

1 - What are the advantages of buffering the input before splitting? Is it just necessary when you dont know the input impedance of the effect where the signal is going?

Basically yes, or when you know the signal is going to a low impedance that the pickup can't drive.

So in our case because are only two opamps with high impedance isnt necessary?

Correct. Both those opamps are also acting as buffers. It might matter if you were using a low input impedance opamp like the NE5532, but the TL072 is measured in tens of megohms. Feeding two TL07x straight from the guitar is easy peasy for the guitar pickup.

One thing does catch my attention though: you currently have R8 connected to Vref. That could be a problem since Vref is shared with other parts of the circuit -you could get interaction. It's better to disconnect R8 from Vref and instead give it a a dedicated cap to ground (10u or more is fine).

[LaloFP]

Sooo... I have some weird good news:

The little dip in volume I was hearing when I disconnected the CleanOpamp´s input (when using just the input to feed both opamps directly) was REAL, BUT: the CleanOpamp was defective/fake. Really... it was another TL072 chip from ST that looked different, and I realized that it wasnt the real deal when I swapped it with the other chip for other test. Its gain was way higher (so the compressor was hitting even if not playing the guitar (I could meassure the vactrol´s resistance)) And I suppose that the input impedance was smaller too, and that made the drop in volume for the other CompOpamp.

I swapped it with another TL072 and after a lot of test, I wasnt able to detect if the CleanOpamp´s Input was connected or not. So, like you repeated to me a million times, for the guitar is the same to feed 1 or to feed 2 TL072. (I say it clearly so anyone that reads this thread doesnt get confused with my previous statement about recognizing a drop on volume. It was real, but because of a faulty part)

Last question, So I can draw the schem and post it:

Now, I have 2 opamps outputs, and tested it with the simpler blend:


Im happy with the results, and with the pot to 100%wet I can´t hear bleed of the clean signal (with a SPDT I toggle it out and in and can´t hear a difference).

But I want to know:
1 - I red somewhere (cant find where) that one opamp´s output is virtual ground for the other opamp. Is that true? Is true for inverting and non-inverting configurations?
2 - In paper and theory, what advantages could we get of adding a summing opamp after the pot? I have one opmamp left to use... but I want to know what Im missing if I dont add the output summing buffer.

Thankss!!!! 

[merlinb]

1 - I red somewhere (cant find where) that one opamp´s output is virtual ground for the other opamp. Is that true? Is true for inverting and non-inverting configurations?

Yes it's true for both.

2 - In paper and theory, what advantages could we get of adding a summing opamp after the pot? I have one opmamp left to use... but I want to know what Im missing if I dont add the output summing buffer.

Normally it would provide a super-low output impedance, but in your case you have a 100k volume pot at the output which defeats the low output impedance. So basically, the last opamp is giving you next to nothing of value. Maybe you can find a better use for it?

[PRR]

> what advantages could we get of adding a summing opamp after the pot?

The pot wiper impedance varies from zero to 2.5k. If you load it with 25k or less, there will be a dip in the middle.

Since nearly all guitar-chain inputs are 50k or more, you don't need it.

[LaloFP]

2 - In paper and theory, what advantages could we get of adding a summing opamp after the pot? I have one opmamp left to use... but I want to know what Im missing if I dont add the output summing buffer.

Normally it would provide a super-low output impedance, but in your case you have a 100k volume pot at the output which defeats the low output impedance. So basically, the last opamp is giving you next to nothing of value. Maybe you can find a better use for it?

I tested moving it before everything to push harder to the compressor, like a pre-gain to squash it, but it didnt sound pretty. Im going to leave it unconnected (or use a TL071). If I use it I have to add a resistor... I dont see the point. Maybe in the future it can be used to make an 2 band compressor

> what advantages could we get of adding a summing opamp after the pot?

The pot wiper impedance varies from zero to 2.5k. If you load it with 25k or less, there will be a dip in the middle.

Since nearly all guitar-chain inputs are 50k or more, you don't need it.

Yeah, with a 100K pot it was noticeable, and with a 500K pot the dip was huge. I did it just for testing.

Here is the final schem!! Thank you guys for your huge help! I learned a lot with this and Im really pleased with the result (in the proto). All the Blend positions have the same output level (when hitting the strings with medium force) and is super practical

UPDATE: R8 is 22k, the image is wrong



Its almost the same as the original, so Im going to try to modify the veroboard that its in http://tagboardeffects.blogspot.com/2012/02/john-hollis-flatline-compressor.html

[rankot]

You have one unused opamp, so it's much better to use it for Vref stabilization, than to leave it unused 

Like this:


PS: Are you sure R8 in your IC2A feedback loop is 44k, since 43k is a standard value?

[LaloFP]

You have one unused opamp, so it's much better to use it for Vref stabilization, than to leave it unused 

Like this:

Nice! And no need of extra components 

PS: Are you sure R8 in your IC2A feedback loop is 44k, since 43k is a standard value?

Ops!! Yeah, its 22k

[samhay]

Or use the op-amp as an output buffer.
Also no extra components.

[LaloFP]

Or use the op-amp as an output buffer.
Also no extra components.

I thought I would need to decouple the signal before and after the buffer and bias it with the offset voltage.
Im going to test it

[samhay]

Output of blend pot is at half supply.
You can feed this straight into the (+) input of the op-amp and use  it as a unity gain buffer.
This is not as good a mixer as an inverting topology, but if you're happy with the current performance, it should be adequate.

[LaloFP]

Output of blend pot is at half supply.
You can feed this straight into the (+) input of the op-amp and use  it as a unity gain buffer.
This is not as good a mixer as an inverting topology, but if you're happy with the current performance, it should be adequate.

Thats great! I wasnt sure If it was better to filter the offset and add a fresh one, but I guess its pointless.

Since without the buffer it works ok... can the stable vref be better?  This kind of decisions, when it almost doesnt matter, drives me nuts