BSIAB 2 Analysis in SPICE

[iainpunk]

i think i found a way to have the triode artifacts you mention, like a curve/asymmetric transfer function of the triode.
you mention the cathode and anode resistors making a difference in a triode's sound, but what is basically does is changing the bias, which is also doable with CMOS gain stages, using asymmetric loading, which reproduces triode sounds quite convincingly.


the cathode follower's sound can be done with a capacitively loaded Jfet Buffer.

a good frequency response can be fully designed with opamps.

3) Mu Amp, about the same as a mosfet, but prone to sagging when heavily distorted.
Also I have often seen reports that the Mu Amp does not simulate a Class A amplifier, but rather a Push-Pull amplifier.

you imply that 'push-pull' and 'class A' are mutually exclusive, which is false, lots of push pull amps operate in class A (just look up the Zen Amp, which is a class A push pull ''''HiFi'''' amplifier). the MuAmp and the SRPP topological family are fully class A, since all devices used are in conduction during the whole 360° of the signal, unless cilpping occurs, when it turns in to class C operation. (a triode clipping to the rails is also operating in class C)

i don't think that if you design a pedal, you have to include the tone stack, since the amp you put that pedal in front of already has a tone stack. this also counts for the phase inverter and the power stage.

i mean my, project isn't to recreate a tube circuit or my amps sound, but to come close to the overarching ''vibe'' of my amp's clean channel cranked, changing a bunch of stuff to sound and feel better to my taste, so my methods might not apply in the process of cloning a marshall's circuit, but its food for thought.

cheers

[niektb]

So I've been toying with this idea to roughly match a transfer function of a mu-amp:


It's a combination of a single-ended soft-clipper and dampened single-ended hard clipper (the later because the mu-amp doesn't clip as hard as a diode).
I've deliberately added some high-end roll-off to keep it smooth
Also, probably use a low slew-rate opamp as it will likely clip

[Vivek]

So I've been toying with this idea to roughly match a transfer function of a mu-amp:

Wow that's great !!!

Please post the transfer function that you acheived.

Let's collaborate to build a new BSIAB with Opamps and Diodes

I too was about to start trying to use Opamps and diodes to approximate the transfer function of the Mu-amp

I was thinking of using Diode Function Generators

- Vivek

[Vivek]

It's a combination of a single-ended soft-clipper and dampened single-ended hard clipper

Are the diodes pointing the correct way ?

I want a circuit that does not depend on rail saturation of the Opamp

[POTL]

iampunk I see such circuits as an imitation of an amplifier or preamp. I think these pedals should be versatile, sound good in a clean channel, sound good in a power amp or regular speakers (using a cabinet simulator). Today I simulated mosfet, jfet, muamp and tube, I set them to the same aspiration, the same frequency response and looked at the oscilloscope. transistors connected in the triode simulation mode (class A amplifier) ​​had a similar distortion shape, it was asymmetric. Mu Amp had symmetrical clipping, the signal was compressed on both sides. when there is time, I will add Cmos to the comparison.

[iainpunk]

yeah, i think you're right about the implied versatility of an ''amp in a box'' pedal, which isn't my personal goal, i'm designing  a good sounding overdrive, that happens to be inspired by my tube amp's clean channel.

yes, normal transistors are quite tube like, especially with appropriate miller caps. i must say that to my ears, the bias of a transistor can make it morph between pentode and triode sounds, but i haven't checked on a scope.
CMOS is somewhat tube like, but it is just to symmetrical in self-bias mode, i like using external bias voltage to get a touch more gain and way less symmetric amplification/clipping.
must say that if you consider the mu-amp with jfets, you should also try that topology with triodes. while you are at it, take a look at SRPP topology, its simpler than - and superior to mu-amp topology.

one thing nit-picky thing tho:

transistors connected in the triode simulation mode (class A amplifier)

the class A part, while its probably correct that its operating in class A, i think you mean 'single ended', because this is also class A:

cheers, Iain

[niektb]

It's a combination of a single-ended soft-clipper and dampened single-ended hard clipper

Are the diodes pointing the correct way ?

I want a circuit that does not depend on rail saturation of the Opamp

You were right! So I inverted one end! I also softened the soft-clipping stage slightly! I can't find the transfer function option 
But I cán plot some graphs like this. This is with a 1kHz wave. Inputs are: 5mV, 10mV, 25mV, 50mV and 100mV. Gain is maxed out.

[Vivek]

I made transfer function graph using excel

By taking readings at about 20 points in between -1200mvp and +1200mvp inputs from the simulation software and typing it into XL


Does your software allow BEHAVIORAL VOLTAGE SOURCE where voltage = sine wave  x time ?


I assumed that 1200mvp would represent approximately the largest guitar signal.

Even for transient analysis graphs like the one you posted, I normally use 10, 50,100, 200, 500, 1000 and 1200mvp as inputs

[niektb]

I appears I actually cán do X-Y plots in TINA-TI! 
So this is what I came up with. I lowered the gain a bit to match that of a mu-amp (27dB). The 120pF in parallel to the diode wasn't doing anything, increased that to an Fc of around 3kHz. The low-end roll-off matches that of a tubescreamer. Add a volume pot (and maybe a tone pot) and you should have a nice simple overdrive on it's own
0

[Vivek]

In the BSIAB2

First Mu amp had gain of 19.5

I did not understand your transfer response graph properly

Your waveshape graph does not match mine.

[niektb]

In the BSIAB2

First Mu amp had gain of 19.5

I did not understand your transfer response graph properly

Your waveshape graph does not match mine.

My bad, got the wrong online calculator (power instead of voltage), but the gain at 1kHz is 25.86dB which is 19.6x.
The transfer response plots both the upgoing (the right curve) ánd the downgoing (left curve) flank (similar to an eye diagram). As you can see, there is a bit of skew...

I know, you can play around with the diodes (should probably LEDs) and the series resistors to get the exact clipping response you want...

[Vivek]

Dear niektb

Could you send me a table in mvp

Input, output
-1200
-1100
-1000
.
.
-100
0
100
200
300
.
.
.
1100
1200

[niektb]

There is an export to text option in graphs in TINA-TI so here you go

Code Select Expand

IN OUT
-1.19948686640175 -1.99172831120686
-1.18903511572659 -2.03126952638068
-1.15598285617605 -2.02599553702016
-1.09157628729542 -1.97192473513099
-1.00029150598102 -1.86805974649348
-0.884376228958682 -1.7164942072197
-0.746684678641574 -1.51916223416013
-0.590607282127369 -1.2716353831398
-0.51791956573537 -1.1215739668619
-0.495495115473969 -1.06622336963506
-0.475765732250903 -1.01059522403624
-0.449178032528607 -0.921253796095772
-0.421439703392049 -0.810524809331847
-0.397326650713596 -0.705874645706524
-0.387202009946283 -0.660812189527395
-0.379039781821198 -0.62452228584848
-0.368298300708477 -0.577234458472926
-0.357674616349312 -0.531143559692211
-0.346080903818751 -0.481675745796077
-0.33063025567917 -0.417110810867088
-0.305877591726118 -0.316821746555281
-0.255963445025229 -0.125548673814609
-0.178279004130879 0.14670077310334
-0.05716587475708 0.522063581972764
-0.024502881647189 0.613813337272989
0.001508367517196 0.683223979898092
0.024044418585108 0.739535804230277
0.069083066403761 0.83671101664916
0.112455125435754 0.910079472880794
0.166504512361057 0.977265078299259
0.242093971159183 1.03808670419737
0.352184709596205 1.08472056039832
0.510379219072326 1.10974625305615
0.673931730028114 1.1138029810462
0.820890018091734 1.11108066493533
0.947635372347442 1.10768129106118
1.05104681033356 1.10472410616672
1.12857797635659 1.10206997492198
1.17831979701013 1.09953933629877
1.19904746745494 1.0970455679434
1.19025060624046 1.09455923609091

[Vivek]

Here's what I see from your data values



I would say that the basic concept is present, but work is needed to fine tune the diodes and gain determination resistors to get closer to the intended transfer function

Its very interesting that your graph does not go through (0,0). Any reason why you designed it with that offset (It could be wonderful if that offset could follow the integrated envelope !!!)

Could you try D3 going to ground via a decoupling cap, or even going to vref via a decoupling cap ?


Is there DC buildup on the caps C4, C5, C8 ?



Also, your positive side clipping is a bit more severe ie chopping to 1 V output while BSIAB does around 2 volts

On positive swings, BSIAB shows some compliance after knee, and does allow the wave to get a bit bigger. Yours is less forgiving.

Please tweak your transfer function by changing type/number of diodes and the resistors


Thanks

[niektb]

It's a combination of a single-ended soft-clipper and dampened single-ended hard clipper

Are the diodes pointing the correct way ?

I want a circuit that does not depend on rail saturation of the Opamp

You were right! So I inverted one end! I also softened the soft-clipping stage slightly! I can't find the transfer function option 
But I cán plot some graphs like this. This is with a 1kHz wave. Inputs are: 5mV, 10mV, 25mV, 50mV and 100mV. Gain is maxed out.

Yeah I also stared at that offset a bit and I'm not sure yet what's causing it! But it does seem to be related to the ingoing signal amplitude, as is also shown in the image in the quoted post (notice how the signal spends more time in the positive half when the input signal gets larger).

[Vivek]

Its DC offset
Maybe due to DC buildup on caps
Due to asymmetric clipping


When the signal gets larger and starts to clip, the charge starts to build up and the wave shifts up


If you check the voltage on the caps, you might find that they build up in the first few ms

That's really great, if its controlled and not too huge a DC shift

But so far, we are not close to our target

[Vivek]

Regarding the filter and drive controls in between the First Mu-Amp stage and second Mu-Amp stage



I saw that many Tube Amp circuits have a treble bleed circuit near the volume pot

It appears that is good if the earlier stage did not create too much distortion and is quite clean

It means that the circuit is trying to correct for the Loudness perception as per Fletcher- Munson curves.

How it does that properly, I do not understand, since the volume control in the pedal of Amp does not know about the final loudness curve created by the PA system.


However if there is too much distortion and high harmonics created by the earlier stage, there is a need to cut the treble after the distortion.

[marcelomd]

Regarding the filter and drive controls in between the First Mu-Amp stage and second Mu-Amp stage[/b]



I saw that many Tube Amp circuits have a treble bleed circuit near the volume pot

It appears that is good if the earlier stage did not create too much distortion and is quite clean

It means that the circuit is trying to correct for the Loudness perception as per Fletcher- Munson curves.

How it does that properly, I do not understand, since the volume control in the pedal of Amp does not know about the final loudness curve created by the PA system.

However if there is too much distortion and high harmonics created by the earlier stage, there is a need to cut the treble after the distortion.

I 'm not sure I understand the question.
The treble bleed is there because the impedance of the volume control, combined with the impedance of the next stage (they are in parallel, so lower volume == lower impedance), may cause treble attenuation. The capacitor is there to bypass this, so the treble range sees the full impedance of the next stage.

Someone please correct me.

[iainpunk]

that DC shift is a result of asymmetric amplification/clipping.
the function of capacitors to put in there to remove DC offset, makes the area under the waveequal on both sides, this shifts the 0-crossing towards the more narrow side.

hope it makes sense.

cheers

[niektb]

that DC shift is a result of asymmetric amplification/clipping.
the function of capacitors to put in there to remove DC offset, makes the area under the waveequal on both sides, this shifts the 0-crossing towards the more narrow side.

hope it makes sense.

cheers

But thats the thing! I dó have an output capacitor?