LG1 Gain block for "Unit Process / Basic Block" AIAB

[Vivek]

Earlier, I had made some posts positing the concepts of Amp in a Box AIAB design using

Unit process
and
Basic "Lego block" concept for stages

The concept was that an Amp in a Box AIAB is nothing more than a bunch of "unit processes" in a row
for example, input buffer, gain stage, tone controls, gain stage, cab sim, output buffer

and those unit processes are constructed with nothing more than lower level blocks that are quite modular and generic. 


Which lead to the idea that, if there are sufficient number of interconnectable blocks available and they are sufficiently tweakable, it is possible to generate a myriad number of valid Amp in a Box designs by choosing and tweaking the blocks, some of which might somewhat emulate actual existing amps and many of which help you realise the sounds in your head.


Recently I saw the schematics of the various AMT L2 series of pedals.

PDF download link : https://guitar-gear.ru/forum/index.php?app=core&module=attach&section=attach&attach_id=39108




Each pedal in the series uses same input block, same clean channel, same cab sim
but different gain channel
and the gain channels are very similar, just minor parts changes between them

Yet, that simple basic architecture allows AMT to claim they have models for Bogner, Diezel, Engl, Krank, Marshall, Orange Peavey, Mesa ...

and this basic architecture can also be used to design/ customise totally new Amps that dont yet exist in tube form.


all of which strengthened my belief on the concept of "Unit Processes" and "Building Blocks" as something useful in the DIY world


I am inclined to eschew FET and rail saturation, since those depends too much upon the physical characteristics of the device.


I seem to prefer Opamp based, non rail saturating situations for their greater conformity and repeatability.


So, dear brothers and sisters of DSB, please help me to design the first building block

Low to medium gain stage LG1

Proposed Specifications :
Opamp based, not rail to rail saturating

1. Input impedance above 100K
2. Frequency response and gain not dependent on the block that came before
3. Output impedance below 5K
4. Frequency response and gain not dependent on the block that comes after
5. Adjustable HPF before distortion with corner from 20 Hz to 6Khz. 6db/oct
6. Also a fixed HPF before distortion, 6dB/Oct. Calculator to determine parts to have corner anywhere from 10 Hz to 300 Hz
7. Adjustable LPF after distortion with corner adjustable from 2Khz to 10Khz, 6 db/Oct
8. Also a fixed LPF, 6db/Oct. Calculator to determine parts that have corner anywhere from 6 Khz or 12 Khz

9. GAIN
9.1 Based on Opamps
9.2 No rail to rail saturation (or rail saturation only reached for input signals more than 1.2Vp)
9.3 "Tube like" transfer curve with 2 - 5 knees, similar to Line6 patent 5,789,689:


-- with that red point adjustable from around 10mvp to 600mvp of input
-- with the slopes after start of clipping to be adjustable from rather benign to medium crunchy (blue line)
-- with output for 1Vp input to be adjustable from 200mvp till 3Vp (Possible with 9V supply)

The idea is to build Amps with different possible structures, including :

Input buffer with gain -> LG1

Input buffer with gain -> LG1 -> output buffer

Input buffer with gain -> LG1 -> LG1 -> output buffer

Input buffer -> LG1 -> Tone Control -> LG1 -> LG1 -> output buffer

Input buffer -> LG1 -> -> LG1 -> Tone Control -> output buffer

Or even a parallel path


                     -> LG1 - >
Input Buffer                          Output buffer
                     -> LG1 - >

and ability to state something like :
"Oh wow, I got a great Plexi tone with
Trimmer 1: 3 O clock
Trimmer 2: 5 O clock
Trimmer 3: 2 O Clock
Trimmer 4: 8 O Clock"

Or, "I needed more pre-distortion Bass for that fuzzy sound, so I set trimmer 2 to 12 o clock instead of 10 O clock"

(if there are just too many trimmers in the design, we can instead hardwire some parts that can be easily determined by an online calculator)


I look forward to your posts of basic Opamp topology to achieve the stated design goals for basic building block LG1 for low to medium gain AIAB (or comments to improve the design goals of this building block)

later on, we can add to the block library by designing the input and output buffers, offering 8 types of tone control stages, adding other filters like a notch etc.

and design some PCB for 1 gain stage, 2 gain stage or 3 gain stage AIAB where builder can create his custom design by choosing how to populate that board / setting the trimmers.

Thanks !

[Steben]

What you are asking is fairly easy to concieve.
Except for the slight non linearity in the transfer curve in the negative half. You can easily design soft clipping with loops, but not negative clipping.
It would need a DC bias shifting the axis.

[Vivek]

Dear Steven,

Please propose a circuit diagram

Thanks

[Vivek]

I was thinking about this as the starting point for a simple DFG

[Steben]

I was thinking about this as the starting point for a simple DFG

It would make a useful semi symmetric clipper.
Yet the middle line region is still quite straight = linear.
Question whether you are in need for that slightly bend line. Does not add much in (gainy) overdrive applications.

[amptramp]

There may be a few other things to specify:

I would guess you would want to specify whether the unit process is inverting or non-inverting.  I would make the standard block non-inverting because it is easier to get a high input impedance in a single stage.

Do you want a bipolar ± supply or does one supply of 18 volts work?  If you can use a single supply, there are 18 volt batteries for power tools and 19 volt batteries for laptop computers that work.  If those work, you can have a power source that would last through a complete tour rather than just a gig and the power tool ones com with a charger.  Computer batteries can be charged but there may be some circuitry needed for that.  Batteries will not need filtering for noise.

If you are using op amp design, your input buffer can also be your gain stage.

If you are playing live, you don't need a cab sim.  This is where modular design works well - leave the cab sim out for playing live and in for recording.

Do you intend to have any feedback between stages or multiple inputs?  This is where an inverting input might be used.

It would use a lot of parts to have the tube emulator in each block and not every block would need it.  An analog clamp circuit on one polarity only might do the job with variable resistance in series and voltage trip point being the variables.

The LEGO block concept would use more parts since each block has to maintain high input impedance and low output impedance whereas an amplifier with all functions integrated in one unit would not need this.  It is necessary to keep the bandpass of basic stages high and have lowpass and highpass filters used only once.

I completely support the idea that non-linearities should not come from the devices used such as FET and rail saturation.  The design should be op amps used in their linear range and characteristics should be repeatable, like McDonald's hamburgers, the same everywhere you get them.  There shouldn't be a situation like a fuzz face where you need one for cold temperatures and another one for hot temperatures like an outdoor stage in order to get the sound right.  Top name players can use a fuzz face because the music store salesman will be only too happy to let him check out a bushel basket of them to find one that works right.  The rest of uis don't get that consideration.

[Steben]

I completely support the idea that non-linearities should not come from the devices used such as FET and rail saturation.  The design should be op amps used in their linear range and characteristics should be repeatable, like McDonald's hamburgers, the same everywhere you get them.  There shouldn't be a situation like a fuzz face where you need one for cold temperatures and another one for hot temperatures like an outdoor stage in order to get the sound right.  Top name players can use a fuzz face because the music store salesman will be only too happy to let him check out a bushel basket of them to find one that works right.  The rest of uis don't get that consideration.

Now you mention it, I am sure good analog opamp design can even mimick "a" Fuzz Face. You know all what is generic about a Fuzz Face. Not about that specific Germanium one or "mosfet and darlington" stuff, I mean the basics like impedance reaction, asymmetry and so on and so on.

[Vivek]

There may be a few other things to specify:

I would guess you would want to specify whether the unit process is inverting or non-inverting.  I would make the standard block non-inverting because it is easier to get a high input impedance in a single stage.

I thank you for your comments

Yes, Agree on non-inverting topology.

Do you want a bipolar ± supply or does one supply of 18 volts work?  If you can use a single supply, there are 18 volt batteries for power tools and 19 volt batteries for laptop computers that work.  If those work, you can have a power source that would last through a complete tour rather than just a gig and the power tool ones com with a charger.  Computer batteries can be charged but there may be some circuitry needed for that.  Batteries will not need filtering for noise.

I feel that if I want max 3Vp signal for flat tone controls, and if the tone control can boost 10 dB = 3.16x and I dont want tone controls to create more distortion,

Then I am looking at 10Vp signal

and I would need 24V supply.

Maybe a charge pump.

Are the calculations correct ?

If you are using op amp design, your input buffer can also be your gain stage.

Yes

If you are playing live, you don't need a cab sim.  This is where modular design works well - leave the cab sim out for playing live and in for recording.

There could be two outs, one with cab sim and one without

Do you intend to have any feedback between stages or multiple inputs?  This is where an inverting input might be used.

I did not think about feedback between stages. What could be the benefits ?

I was thinking of one input. But whole idea of block by block system is that anyone who needs different stage can design it / use it.

It would use a lot of parts to have the tube emulator in each block and not every block would need it.  An analog clamp circuit on one polarity only might do the job with variable resistance in series and voltage trip point being the variables.

Tube emulator only in the gain blocks, not in the input, filters, tone, output blocks

The LEGO block concept would use more parts since each block has to maintain high input impedance and low output impedance whereas an amplifier with all functions integrated in one unit would not need this. 

Yes, there could be some redundancy to ensure interoperability,

But I feel LG1 can be achieved by one Opamp.

It is necessary to keep the bandpass of basic stages high and have lowpass and highpass filters used only once.

I feel there is a need for filters between stages
So input will be 20 - 20K
Gain stages will have HPF and LPF
output buffer can be wide bandwidth, maybe till 10K

I completely support the idea that non-linearities should not come from the devices used such as FET and rail saturation.  The design should be op amps used in their linear range and characteristics should be repeatable, like McDonald's hamburgers, the same everywhere you get them.  There shouldn't be a situation like a fuzz face where you need one for cold temperatures and another one for hot temperatures like an outdoor stage in order to get the sound right.  Top name players can use a fuzz face because the music store salesman will be only too happy to let him check out a bushel basket of them to find one that works right.  The rest of uis don't get that consideration.

I did a study on change in transfer function of a FET stage based on trim pot setting, and I found it be be very wide in range. Slight changes in trim pot can change the way that FET stage works.

I am hoping for consistency and repeatability with Opamp and Diodes.


Could you please propose a Non-inverting topology that accomplishes the design goals for the gain stage LG1. Thanks

[Steben]

You need to define again what those design goals are. Is LG1 a triode sim?

[Vivek]

LG1 is supposed to be a low to medium gain stage that creates gentle to mild crunch distortion when used singly, but reaches metal territory when 2 or 3 of them are in series

It can be 100% clean at low gains
Fetzer like at slightly more gains
and tube like at its higher settings when used singly

It does not claim to be an emulation of a triode

it however has gentle, asymmetric clipping which is somewhat similar to the transfer curves of a triode.

[ElectricDruid]

A transistor differential pair gives the best "low gain" drive sound I know about. This example is used as a triangle->sine waveshaper, but the circuit is exactly the same used as an overdrive:



You can ignore the IC1 ramp->triangle stuff, and you'd probably want to replace IC2.1 with a standard non-inverting buffer-with-gain amp stage. You can tweak PR2 for the degree of asymmetry you want.

[Steben]

A transistor differential pair gives the best "low gain" drive sound I know about. This example is used as a triangle->sine waveshaper, but the circuit is exactly the same used as an overdrive:



You can ignore the IC1 ramp->triangle stuff, and you'd probably want to replace IC2.1 with a standard non-inverting buffer-with-gain amp stage. You can tweak PR2 for the degree of asymmetry you want.

The opamp is buffering with probably higher headroom than the transistor pair?
Perhaps not much difference with those Boss style circuits right?

[Vivek]

A transistor differential pair gives the best "low gain" drive sound I know about.

Whenever Tom ELECTRICDRUID drops a hint, I listen deferentially !

[teemuk]

The opamp is buffering with probably higher headroom than the transistor pair?

Most likely. The differential clips very softly so it starts to distort and "round up" the output signal quite early. Do note that the circuit effectively amplifies the difference of the differential output.

Perhaps not much difference with those Boss style circuits right?

Boss effects drive a common emitter amp from differential input and employ global NFB. Boss circuit is basically a very simplified discrete operational amplifier and as such there is a major difference in comparison to a plain differential stage operated in open loop.[/quote]

[ElectricDruid]

The opamp is buffering with probably higher headroom than the transistor pair?

You could adjust that with the value of PR1. If the input is going to use all of the headroom of the op-amp, you'd want to set the PR1 value so that you got some hard clipping from the transistors before the op-amp starts to clip.

Whenever Tom ELECTRICDRUID drops a hint, I listen deferentially !

LOL! That's really not wise!!

[Steben]

Any existing circuit that uses differential distortion?

[Vivek]

The differential clips very softly so it starts to distort and "round up" the output signal quite early. Do note that the circuit effectively amplifies the difference of the differential output.

Do you mean that the output is not linear ?

Output of differential pair not equal to (input 2 - input 1) * X ?

but instead depends a lot on |(input 2 - input 1)| ?

and how is that different than a single transistor stage entering saturation ?

[Vivek]

Any existing circuit that uses differential distortion?

https://www.diystompboxes.com/smfforum/index.php?topic=93613.0

And didn't Boss BD-2 do this ?




Aren't all Opamps differential ?

[PRR]

...Output of differential pair not equal to (input 2 - input 1) * X ?  ...

Only for very small signals. Even at 20mV it is not clean. 50mV is distinctly distorted.

The ~~50k/2k divider before the pair makes that 500mV out of the opamp.

[Steben]

Any existing circuit that uses differential distortion?

https://www.diystompboxes.com/smfforum/index.php?topic=93613.0

And didn't Boss BD-2 do this ?




Aren't all Opamps differential ?

BD-2 is a very rough mini opamp as teemuk described.