☠ Guitar Pre. Clean With Some Sizzle.

[MJL21193]

Some tinkering I've been doing:
Guitar Preamp.pdf
Taking advantage of the things I learned from my recent guitar amp build, I have come up with something a bit different. It combines features of the 2 channels from the Ampeg VH140 into one, mostly with respect to tone control.
Basically a clean preamp, it has some edge when the gain is turned up. Or, at least that's how it looks in simulation.
Suggestions for improvement welcome.

One thing I want to do is a side by side comparison of the actual results to the simulated, as the build progresses. First thing I did was to verify the amount of gain I had from the FET/BJT combo input stage, as the sim didn't give me the expected results. Since I'm not overly experienced with jfets, I breadboarded the circuit:

111.png

I discovered that the simulation model I have for the J201 is a bit off, giving a lower than expected gain for the stage. In fact, there isn't a jfet model in the MS database that gives near to an accurate result for the gain through this stage. I find this a bit perplexing TBH.
To make the stage simulate the correct gain, I had to change R24 to 880 ohms (gain for the stage is actually supposed to be ~R17/R24).

112.png

[kenpeter]

Feed a fraction of the collector voltage back to the source makes a linear voltage amp.
Transforms horizontal curves of your JFET + BJT Sziklai pair to vertical curves of a triode.
Reminds me of O.H.Schade's Fig#35 (search it on the tube forum). And you have done
so entirely with direct coupled local feedback, very cool indeed!

But also reminds me of my own project, PNP concertina with equal impedance split.
Was already 1/100 voltage feedback from plate=emitter and Mu, wanted to balance
that with a similar fraction of voltage feedback from the collector side.

You could increase your JFET's open loop gain using VBE as a constant current source.
Use a larger value R1 in parallel with this current if you still want some resistive slope.
Ref to your R1, not mine...

[kenpeter]

Properly selected values of R17 and R24 might make R25 unnecessary?
Try 820 and 120

[kenpeter]

Dunno if this model for j201 is right or not? Found by searching this forum...

I am thinking neither circuit drawn here is optimally biased yet.
The collector would want to idle a few volts higher???

[kenpeter]

Would you believe R3 (ref post #4) is totally unnecessary too?
I reduce the value to .001, nothing changes...

R1, R2, and Threshold of J201 so strongly set the operating point

[kenpeter]

Everything readjusted so's current in JFET and battery are same as your original.
Sorry, seems gain had to go up if we center the output for maximum headroom.
Not sure if that has now over-linearized the thing in some harmful way?...

[MJL21193]

Originally Posted by kenpeter

Properly selected values of R17 and R24 might make R25 unnecessary?
Try 820 and 120

Thanks for your help! I have adopted your change to eliminate R25 and adjust R17, R24 to compensate. I have left R2 in place for now - the degeneration may make the stage more stable.
Interestingly enough, the changes have brought my J201 model closer to reality, but still off. A much better fit now, giving nearly dead-on results, is the 2SK170 model. The model you posted for the J201 gave me worse results than the model I have. Same as the model I got from Fairchild - crap results.

On the face of it, it appears the models are overly influenced by how much current passes, as demonstrated by your changes and my new results.

111.png

With the 1K resistor back temporarily (now R5), I have nearly exact match with the actual circuit.

Originally Posted by kenpeter

Everything readjusted so's current in JFET and battery are same as your original.
Sorry, seems gain had to go up if we center the output for maximum headroom.
Not sure if that has now over-linearized the thing in some harmful way?...

The gain for the stage needs to stay relatively low, with just enough headroom to cover the typical output of a guitar being played fairly aggressively. This, according to my measurements, is around 1Vpp. The idea is that clipping will happen in the first gain stage and not in the following stages (they clip, but it's controlled via diodes and not rail to rail on the opamps). That's the primary purpose of this first stage, or at least how I envisioned it.

[MJL21193]

Some detailed looks.
Despite the problems with accurately modeling the fet/bjt combo, I expect the opamp stages to sim within reason. In particular, the amount of gain and the quality of the distortion, with regard to wave shape.
After the input stage, there are 2 'clipping' stages. The first has the high tone control included in its feedback loop. What I have tried to do is limit the clipping in this stage to ~50% of the gain control and up. So, with the gain set at 25%, there should be no clipping:

111.png

resulting in a very clean output:

113.png

With the gain set to 100%, there will be clipping:

112.png

resulting in a nicely distorted output:

114.png

This is a balancing act and it took some considerable fiddling with levels to get it here. Hopefully, the actual circuit mirrors the simulated performance. Based on the measurements I made on the distorted channel of my build, it shouldn't be far off.

[MJL21193]

I need to build a prototype of the whole circuit before I can test the accuracy of the data in the previous post. That will be a matter of doing a rough layout, etch that board, stuff and test. If my previous experience has counted for anything, that first board won't be the final one.

On to the main reason for doing this. In my original build, I followed the Ampeg schematic to the letter, except for one small detail: I didn't notice that there were log pots (and a couple of reverse log pots) on the schematic. I used linear pots only. This was more of a problem for the clean channel, as it uses more log pots. There is a rather abrupt transition in adjustment of the tone controls on the clean channel.
Here, I set out to minimize that, and still use linear pots. I think I've done a pretty good job, if the frequency response of the real preamp matches the FR of the sim.
Here is the response with the bright switch off. The 5 traces represent the response of the preamp with the tone controls set to 0%,25%,50%,75% and 100%, with 0% being the bottom (green) trace:

111.png

Not too bad, with more than 10db of boost over the pots travel.
This is the response with the bright switch on:

112.png

High shifted up 5-6db, pulling the mid up slightly too. Not too shabby.

I did try to jack up the mid a bit more but it killed that deep null when the pot is at 0. I think that deep null might be more desirable than a hump, especially as the speakers have the highest output at that point. Could be interesting, again, if the real circuit matches the sim.

[MJL21193]

I have the test board layout done:

1.png

took a while longer than I expected as I'm coming to dislike PCB layout...

Ready to etch at any rate and I can do some testing once it's stuffed. One slight problem is that I don't have a 25K pot for the high tone control. I could wait or just use a 50K instead for testing. As I'm nearly certain this will not be the 'final' board revision, I think I'll press on and use the 50K pot.