Troubles with Baxandall in LTspice (LTspice noob)

[Rob Strand]

his is the current schematic (removed shunting cap at input altogether).

What you need is a cap between the wiper of the volume pot and the 1M resistor to Vref (the 1M that's before the buffer/tone control).

[hans h]

 Aha, so I missed a coupling cap. I guess that the way it is now, the volume pot is screwing up my bias?

[hans h]

Hi Vivek,

That's too much for me to type in one post. Let's start with filtering. I have a 100 pf in the feedback loop of the first two opamp. My previous  experience was that these cut quite a bit of the high end hiss (sometimes even too much). However, for the current project I do not yet know whether it's enough, I still have to get my breadboard version up and running.

With regards to the coupling caps: simply 1/(2pi R C). The 1k for clipping is based on one of the comments in this thread, the 50 k is the minimum impedance of the volume pot, and 1m is the input impedance of the Non-inverting opamp. See chart and schematic below:





To recap: the cap for the clipping stages (330n) is still unsure, I'll have to experiment ( as was pointed out to me in this thread).

[hans h]

With regards to the gain pre-clipping: 1+rfeedback/rtoground => first stage has a minimum of 3.2 and maximum gain of 103.2. For the second stage I wanted lower minimum gain so I used a 1k instead of the 2k2. After all, the signal is already boosted by the first stage. Idea behind this is to get maximum flexibility in combining gains. I should be able to do low into low, high into low or high into high gain.

[hans h]

With regards to overall frequency response: Primary sound is (or should be) Mesa. I therefore stuck a few Mesa tones tacks in Duncan tsc and tried to emulate what I was seeing using ltspice. I found that (with bass and treble at 70 percent) a 47nf bass cap and 2n7 treble cap got me really close. However, I also needed bass rolloff at approx 27hz.thats where the input cap comes into play (5n6 with 1m impedance).

Then for switching of caps I have: treble cap a fixed 2n7. Bass cap values: 15 nf (voxish/ big muffish), 15 nf parallel with 12 nf (marshall) and 15 nf parallel with 33 nf (Mesa).

[Rob Strand]

Aha, so I missed a coupling cap. I guess that the way it is now, the volume pot is screwing up my bias?

Yes, the bias voltage will be low because the volume pot loads down the 1M feeding the opamp, and the
bias will vary from 0V to to maybe 0.5V depending on the volume pot position.

[PRR]

...........volume pot is screwing up my bias?

That's what I see. (my) U2 and U3 both.

[hans h]

In the latest schematic there is now a cap between the volume pot and u3 (10nf). However, I see I forgot the cap after the first clippers, before u2 (suggestion Antonis). I will place another 10nf cap there. So much to think about! 

[Vivek]

DC paths

[hans h]

Thanks for the sketch on the DC paths. I should have these kind of sketches for all kind of impedance situations 

I still have to answer your other questions Vivek:

A) First off, if I need a SWTC on top of the baxandall, I think I'd rather go with some extra simple filtering like negative feedback or cap to ground. After all,  I already have: A cap switch for the baxandall, a boost switch, potentially an asymmetry pot, two gain pots, a volume pot and a bass and treble pot 

F) Same goes for the Wampler pre-dist bass controll. If I find that stuff gets too flabby from too much bass going into the distortion stages, I'll just drop the bass to ground using a smaller input cap and potentially also a smaller cap for C1

B) Vref circuitry: it is a bit unclear as I added the diode protection at a later stage. But it goes: series diode (whatever type, I don't mind a bit of voltage drop as I can always use 18V to power the pedal), 100 r series resistor. After that I have the 9V node which is filtered by a 100 uf cap. After that a voltage divider with 10 k series followed by 10 k to ground. This 4.5V node is filtered by a 47uf.

C) This sounds really interesting. So If we bias positive, the "positive clipper" will start to conduct earlier than the "negative clipper" and therefore we get asymmetric clipping. If you have a bias knob installed here, this will also give you a "asymmetry control". Is this in any way better, or just different from the asymmetry control I dreamt up (increase resistance in one of the two clipping paths). I haven't really tested the original asymmetry control much yet, so cannot really tell you how good or bad it works.

D) not anymore, see added 10 nf cap after the volume pot in the new schematic

E) Thanks for the tip, I'll add it to the schematic. I would guess it should go in series between the last opamp and the 2u2 output cap?

G) For now I am going for clean opamps, with all dirt coming from the clipping pairs. So based on your comment I should probably start at 18V. These drain-gate connected mosfet clippers already have a very soft clipping knee, so in that sense similar to overdriving jfets or tubes for example. If I find that the distortion/ overdrive doesn't sound complex enough, I might start to experiment with purposefully overdriving the opamps.

H) I have already answered most of this point. However, I havent said anything yet about signal level. Similar mosfets (BS-170) apparently clip at 1.5 to 3V. Let's compare this to the potential signal level in the pedal:

So after the initial attack we are left with say 400 mv peak to peak of signal from the guitar (http://www.muzique.com/lab/pick.htm). In the first stage, this will be amplified to a minimum of 3.2*0.4 = 1.28 V and a maximum of 103.2*0.4 = 41.3V (WOW!). I think the lower limit is quite alright (only a bit of hair when digging in). However, it seems the maximum level is on the high side. Maybe I should use a 10k gain pot? 13.2*0.4 = 5.28V. Do you have any idea why they use a 100k gain pot in the Barber dirty bomb? Or am I calculating gain the wrong way?

Lets say our first clipper clips of everything at 3V. The second opamp will then indeed saturate badly, even with a tiny gain of the second gain knob.

So does this mean the setup I had envisioned doesn't work at all? Should I add a voltage divider prior to U2? Hmmm, I guess your question does confuse me, but in a good way 

I still have to add the additional coupling caps on the breadboard (post clipper and post volume knob). Let's hope it works regardless of the high signal potential :p.

[Vivek]

A) Normally, there is a Low pass filter after each stage of multi-stage clipper

I had suggested SWTC with internal trimmer after stage 1 and also stage 2, so that you can sent to your liking and forget.

I feel your circuit will benefit from some low pass filters, such that response at 5K is lets say 6 to 12 db down from 1K, and response at 10K is lets say 12 db down from 1K

for example, here is a rather extreme filter in a Marshall solid state combo, after distortion stage and before power amp stage :



The whole idea is to remove high frequency fizz/harmonics/inter modulation products.


B) Normally those DC paths should not exist

C) Normally, the max total "gain" of such pedals is around 200- 400 or so. Normally, the gain is somewhat equally distributed between the two stages

Normally if you want the harmonic benefits of high gain on one stage, but not necessarily have high output levels extremely overdrive the next stage, it is possible to have inter-stage attenuators

For example, have gain of about 60
attenuate to about 1/10th
Have gain of about 30



But "Normal" is not what inventors and experimenters do !!!

Make your own new path, brother !!!!

[hans h]

Hi Vivek,

Thanks for the explanation. One thing though: why not just a cap in the feedback loop instead of a cap to ground? I already have 100 pf caps in the feedback loop, but I realize these only dump the very high frequency stuff.

Also: you were completely right. Tested the pedal yesterday and it was way too fizzy (and quite uncontrollable due to the extreme gain). I am now experimenting with getting the gain/ volume under control, and after that I'll place some extra low pass filters. At the moment I have a 100k gain pot for the first stage and a 10k for the second that is already better, but I think I still have a long way to go

[hans h]

with regards to the "dc paths should not exist": thanks, I already included extra caps to block DC flowing into the clippers and the volume pot after what Rob and Paul mentioned so that should be ok now. I see I also forgot the coupling cap going into the baxandall stage. That should be 1uf based on LTspice, but I completely forgot it in the breadboard version (and got some really weird sounds because of it)

I also found out that the gain is WAY too much in the baxandall. Pre-baxandall is quite okay with 100k gain pot followed by 10k gain pot followed by non-inverting buffer. If I put the two gain pots to their minimum I get clean sound throughout the circuit, untill signal gets into the baxandall stage: lots of volume increase there and a lot of clipping (apparently from the opamp). I am trying to get the gain of the baxandall down while keeping the same characteristics as the tone stacks i am emulating, but it is difficult. I still have about 7db gain with pot settings at 0.8 of the sweep.....

Edit: I found that some series resistance before the baxandall lowers the overall gain (gain is Rfeedback/Rin, should have known). So now I have this:

(oops can't add picture in edit. See next post:)

[hans h]

[PRR]

....gain is WAY too much in the baxandall....

If you use linear pots and center them, the Bax is unity gain. So what is wrong?

[Rob Strand]

I am trying to get the gain of the baxandall down while keeping the same characteristics as the tone stacks i am emulating, but it is difficult. I still have about 7db gain with pot settings at 0.8 of the sweep.....

You should confirm your built circuit is actually producing unity gain with the tone pots centered.

The way to knock down the gain is to put a resistor between the clipper and the volume pot (clockwise/pin3).   A 100k will knock the level down by 6dB.

The other question is do you need all the boost/cut from the tone control.   You can simply redesign it to have less boost/cut.   Set the maximum boost cut with the lower value input/output resistors on the tone control then tweak the caps so the response is like the existing circuit with the pots rotated only a small amount.