Discrete Hotcake

[daeg]

Had the idea, threw a schematic together. Used the Boss BD-2 as the point of reference. Have not breadboarded it yet.

[daeg]

My thinking with this is that not only is the discrete opamp thing kinda cool, but it will probably perform even worse than a TL071, which could mean less high-end fizz.

[Rob Strand]

Something to worth trying anyway.

You need a resistor from Q3 collector to ground - see Boss stuff.

Also a big cap from the opamp side of the 220 ohm to ground.

[daeg]

You need a resistor from Q3 collector to ground - see Boss stuff.

Good catch. Will add it.

Also a big cap from the opamp side of the 220 ohm to ground.

You mean a filter cap on the V+ rail? Yeah, I'll add that too. The Hotcake schematic I was going off didn't have it.

[aron]

I can't remember if we tried this, but for a while I was using Joe Davisson's discrete op amp in a number of the circuits I had. I had it on a DIP8 socket.

[daeg]

Updated. How is it looking?

[daeg]

I can't remember if we tried this, but for a while I was using Joe Davisson's discrete op amp in a number of the circuits I had. I had it on a DIP8 socket.

I like that idea. Perhaps I'll build a DIP-8 module with a TL071 compatible pinout.

I could see this circuit as having the most impact from the discrete Opamp, considering it's one of the only distortions that uses 100% Opamp clipping. Most users only like these into already dirty amps because of high-end fizz and decay artifacts.

[Rob Strand]

Updated. How is it looking?

It looks good now.

The 470pF across the output might cause problems.    It could cause oscillations and/or affect the sound differently on the discrete opamp compared to the original opamp. It's not normal to shove a 470pF across the output of an opamp.

No clear way how to deal with that.   The easy way out is to just be prepared to pull it.   Beyond that
you might need to add a 100ohm to 1k resistor on the track between Q3's collect and R9.

I can't remember if the 470pF was on all versions of the hotcake.

[daeg]

What is C3 (the cap across the BJT) doing? It looks like Miller capacitance but the PNP config has me uncertain.

Could we up it from 47p to 470p then remove C5 (the 470p to ground)?

[Rob Strand]

That is C3 (the cap across the BJT) doing? It looks like Miller capacitance but the PNP config has me uncertain.

Could we up it from 47p to 470p then remove C5 (the 470p to ground)?

The PNP acts provides the gain for the discrete opamp, so it behaves more like the opamps we are used to.

C3 prevents the amplifier from oscillating.    It is acts like a miller capacitance (part of a pole-split compensation).    There is usually a "right" value for a given feedback configuration.   You will find Boss uses slightly different values here and there.    The Waza series of Boss pedals use discrete opamps also and they mess with that cap value in different parts of the circuit.  IIRC in one case they added a resistor in series with the cap.

C3 is quite different to C5.    You might be able to adjust C3 so the circuit doesn't oscillate with C5 present.    However, that's more of a "you could do it" kind of thing.   It's probably not the right thing to do.    When you stick a cap across the output of a given opamp it does "something" to the high frequency response.   That something is different for each opamp.   The output stage of a TL072 and most opamp is a class AB stage with a low impedance and usually some current limiting circuits.   The output stage of the Boss discrete opamps is very different indeed, nothing like a class AB output stage.   So that means the whole effect of C5 is going to be different.

IMHO, playing with the value of C3 has experimented with enough.  It could turn out that value can have positive and negative effect on the tone of the discrete opamp.      For example the Boss BD-2 has 47pF and the Boss OD3 has nothing.    Is C3 contributing to the weird fizz in the tail of the note on  the BD-2?  The OD3 doesn't have that fizz.   It's something I've planned to checkout for ages but haven't done it.

[daeg]

What about connecting C5 (470p) to the R7-R9-C4 junction instead of ground?

This circuit is already prone to oscillation and pulling C5 gives instant oscillation on my breadboard.

[Rob Strand]

What about connecting C5 (470p) to the R7-R9-C4 junction instead of ground?

This circuit is already prone to oscillation and pulling C5 gives instant oscillation on my breadboard.

I think the whole idea of C5 doesn't apply to to the discrete opamp.   The output stages are too different.
I've actually got doubts about how much is achieves on the original 

[daeg]

Reporting back Rob. Breadboarded the circuit above. Results are...

1.) Much less gain available.
     * With the Drive knob dimed you still have to dig to get clipping.
2.) Gain rises and falls with supply voltage.
     * But then so does headroom... No free lunch I guess.
     * Tested at 3.3v, 5v, 9v, 18v. 18v clipped slightly more than the others.
3.) Gain affected by transistor characteristics
     * J201 amplify much more than MPF102 ... shocker I know.
3.) The fizz and noise artifacts are gone.
     * No crackle even at the same gain level.

[Rob Strand]

Reporting back Rob. Breadboarded the circuit above. Results are...

Cool stuff.

The 2SK184GR JFETs that Boss use have quite different characteristics to "normal" JFETs.
The 2SK184GR should give a bit higher gain than the common JFETs.

Another way to get more gain is to decrease the 4.7k source resistor, and
then decrease the 2.2k drain resistor (roughly by the same factor).    Suppose you change the 4.7k to
2.2k, you measure the source voltage then set the drain resistor to,

      Rdrain    =  2* 0.66 / (Vsource / Rsource) = 2 * 0.66 * Rsource / Vsource

So suppose you set Rsource = 2.2k, and measured Vsource = 6V,  Rdrain = 480 ohm which is about 470 ohm.

The  2SK184GR JFET change is probably going to be more noticeable than the resistor changes, just like the MPF102 vs J201.

Also you can't go crazy and reduce the source resistor to very small values.  For each JFET type you can only go so far.


You might notice my formula doesn't give the same Rdrain as Boss.   Not sure why, I'd have to check all the details.

[11-90-an]

Forgive me, but I couldn't resist... 


Just for fun...

p.s. I have a slight feeling of some toasted JFETs though...

[Rob Strand]

Forgive me, but I couldn't resist...

Strictly you need to adjust the drain resistor when you vary the current.   A current mirror between the two drains would work for all currents.    And guess what, the current mirror would increase the (open loop) gain by 6dB. 

We have to be careful about the cranking the gain too far.  The amp might go unstable.  Tweaking the collect-base cap might be possible to some degree.


Here's an example on how to do the mirror,
https://3shared.wordpress.com/2010/07/24/discrete-preamp-class-a/

Don't worry to much about the other stuff.

[GFR]

There's a version of the hotcake that uses the IC's offset compensation pins to generate the clipping.

https://www.diystompboxes.com/smfforum/index.php?topic=125157.0

https://www.ti.com/ds_dgm/images/fbd_slos080n.gif

[daeg]

The  2SK184GR JFET change is probably going to be more noticeable than the resistor changes, just like the MPF102 vs J201.

I'm going to shelf this project until I get a hold of some more appropriate JFETs. Breadboard is getting tossed in a ziplock bag. "Tag it and bag it".

Thanks for all of your help Rob. You're a wealth of knowledge.

[Rob Strand]

I'm going to shelf this project until I get a hold of some more appropriate JFETs. Breadboard is getting tossed in a ziplock bag. "Tag it and bag it".

Thanks for all of your help Rob. You're a wealth of knowledge.

Boss obviously did their homework when they chose those JFETs.