Yet another boring AIAB

[POTL]

Hello everyone, inspired by the New BSIAB thread, I decided to take the time to compare all (or almost all) of the existing ways to simulate an amp in a box. I have studied the following ways
1) CMOS
2) Classic MOSFet (bs170 / 2N7000)
3) JFet (j201 and others)
4) Muamp
5) SRPP
6) Op amp
7) Depletion MOSFet
In the competition for the best contender, many dropped out.
CMOS - low flexibility and high noise
Classic MOSFet - Big Noise, Large Input Capacitance
Jfet - low flexibility, a huge range of parameters that do not allow good repeatability without selection of components
Muamp - High Output Impedance Problems

3 applicants left
1) Op Amp is the most flexible and most stable until only one theoretical problem is solved - how realistic is it to make a "correct" waveform using diodes and how it will sound compared to discrete circuits. Which is better to take, a classic bipolar transistor amplifier, an amplifier with a field effect transistor input or a mosfet opamp.
2) MOSFet in depletion mode, everything looks promising, the simulation shows a good frequency simulation with a minimum of circuit changes. However, simulation says it's not very good for clean sound (though I don't have an LND150, only a BSS129). The question is, how reliable is the use of such a transistor, in terms of repeatability?
3) SRPP looks promising like MOSFet in depletion mode, I took Catalinbread products as a basis and added a feedback resistor to reduce the gain, instead of J201 I used mmbf4393, they showed better dynamic range. The main question is, how stable is this device due to the huge spread of JFet parameters, will it require the selection of transistors or the circuit will work with the closest possible results (gain, frequency and dynamic ranges) without selection of components?

Update
I decided to compare one more nuance, Mu amp, Mu amp with output buffer and SRPP, and install a small boot resistor. SRPP article promises good results, but Mu amp with buffer works well unlike SRPP interesting discovery.

[vigilante397]

So.... what about tubes?

[POTL]

So.... what about tubes?

A rhetorical question. We will not be able to replace the lamps with these devices, but the goal is different, to find the most similar in sound way of emulation without DSP.

[ThermionicScott]

A rhetorical question. We will not be able to replace the lamps with these devices, but the goal is different, to find the most similar in sound way of emulation without DSP.

Heh, I was just about to tease that you and Vivek seem to be on a mission to "model" tubes with the most generic and character-free SS devices out there.  Guess I don't have to worry about offending after all.   

[FiveseveN]

At some point it becomes less of a hassle to just use some tubes.
If authenticity is the goal, build a tube preamp in a box.
If flexibility is the goal, use DSP.
If it's academic and you're disregarding practicality, sky's the limit on the resolution of details you could address. Are you familiar with Kempf, Danyuk, Azevedo etc.?
In deciding how similar the emulator sounds, how are you planning to compare it with the amp? Or on a related note, which functions of an amp is it looking to emulate (not the moving speakers part, I presume)?

[iainpunk]

the problem/question i have with very premise of the idea of AIAB.
wouldn't it be better to focus on a ''perfect sounding overdrive'' instead of a clone of something someone else came up with?

another question: where do you draw the line of replication?
do you design it for use directly into a power amp or FX Return? front of the clean channel? into a DI box?
what do you design it towards? if you do tube emulation, are you cloning a certain amp or is it a better idea to design an ''amp like overdrive'' that has some characteristics of general tube amps, or are you copying a specific amp with ss components?

don't get me wrong tho, i really like your efforts of tube emulation, and i love reading these threads!

cheers, Iain

[POTL]

the problem/question i have with very premise of the idea of AIAB.
wouldn't it be better to focus on a ''perfect sounding overdrive'' instead of a clone of something someone else came up with?

another question: where do you draw the line of replication?
do you design it for use directly into a power amp or FX Return? front of the clean channel? into a DI box?
what do you design it towards? if you do tube emulation, are you cloning a certain amp or is it a better idea to design an ''amp like overdrive'' that has some characteristics of general tube amps, or are you copying a specific amp with ss components?

don't get me wrong tho, i really like your efforts of tube emulation, and i love reading these threads!

cheers, Iain

Hello. I don't want to frame the effect. in fact I am into AMT effects, which can be used both in the input of an amplifier, as a distortion pedal, and in an effects loop, as a preamp. Ideally, use the pedal in conjunction with an impulse cabsim to play in a line. When it comes to playing the loop, I'm not sure if 18 volts will be enough for a good sound, so I'm basically looking to use this effect into the amp's input. I tried many methods several years ago, now I decided to delve into the theory and understand why certain methods were better or worse.

It remains to understand the theoretical moment with Mu Amp, how capable of repetition it is, in contrast to single jfets. How much dB of gain from each stage will be really useful. As far as I remember, the 4 jfets stages (4x15dB minus the voltage dividers) around 30dB gain were just as useful as the 70dB from the 3 2n7000 stages in the Zvex BOR clone. By example, I found out that the pure sound at a voltage of 18 volts will not exceed 26-28 dB, and the distorted sound can be adjusted to gain similar to a tube preamp.

After all the simulations of Mu amp and SRPP, the big question is, if they are bad for repeatability, I will not consider them (they already require more detail than any other way to simulate). Mosfet and Op Amp look like clear favorites.

[iainpunk]

nice,

have you looked in to Mu Amp and SRPP using depletion mosfets, like the LND150? they omit some of the variability drawbacks of JFETs while still having superior characteristics of those topologies.

i think the Mu Amp is more affected by mis-matched Jfets than the SRPP is, but i might be wrong here.
also, the SRPP can be trimmed with the resistor between the two FETs to get the output at about 65% of Vcc, trimming the mu amp is more difficult, so matching is more needed in them.

cheers

[ThermionicScott]

I am into AMT effects

Well, that was an interesting Google search.   

Hello. I don't want to frame the effect. in fact I am into AMT effects, which can be used both in the input of an amplifier, as a distortion pedal, and in an effects loop, as a preamp. Ideally, use the pedal in conjunction with an impulse cabsim to play in a line. When it comes to playing the loop, I'm not sure if 18 volts will be enough for a good sound, so I'm basically looking to use this effect into the amp's input. I tried many methods several years ago, now I decided to delve into the theory and understand why certain methods were better or worse.

I think Iain and others of us are curious because a given pedal output will sound very different going into the input of an amp, the effects loop, or into a DI.  It's a pretty safe bet that if you optimize it to sound good in one of those cases, it'll sound bad in another.  If you are trying to incorporate a full tone stack and cab sim EQ into the sound and then run that into an amp's input, you're likely to double-boost the bass and treble.  The Boss Metal Zone is a good (cautionary?) example:

And I think I didn't love the EHX English Muff'n for the same reason, you end up with two preamps trying to determine the EQ.

So it's not just to pin you down.  The term "amp in a box" is definitely overloaded, but I tend to think of it not literally but: what distortion pedal can give me the illusion of switching from my clean Fender to a cranked Marshall or Vox AC30?  In those cases, the pedal's EQ circuitry needs to complement what your clean amp is doing, not just duplicate the tone stack from a Marshall or what have you.

[POTL]

nice,

have you looked in to Mu Amp and SRPP using depletion mosfets, like the LND150? they omit some of the variability drawbacks of JFETs while still having superior characteristics of those topologies.

i think the Mu Amp is more affected by mis-matched Jfets than the SRPP is, but i might be wrong here.
also, the SRPP can be trimmed with the resistor between the two FETs to get the output at about 65% of Vcc, trimming the mu amp is more difficult, so matching is more needed in them.

cheers

No, I simulated Mu Amp and SRPP on regular Jfets J201 and mmbf4393. I considered these topologies for two reasons, there is no need to adjust the bias of each transistor and good repeatability, but as I understood from previous comments, circuit repeatability is not much better than regular single jfets.

simulation showed that SRPP has a number of disadvantages
1) It is still dependent on the value of the load resistor, the situation is slightly better than the classic Mu Amp, but much worse than the Mu Amp with buffer.
2) Excessively high gain (for someone this is a plus), about 50 dB, against the conditional 17 dB for single Jfet and about 30 dB for a mosfet. It is difficult to build multi-stage circuits, with 4 or more consecutive amplification stages
3) To reduce the gain, you need to use either a small resistor in the feedback of the lower transistor, which greatly reduces the input impedance (this is critical for the first stage) or use a large source resistor of the lower transistor (100 kΩ or more).
4) Distortion is symmetrical and it is very difficult to simultaneously adjust the asymmetry and filters.
5) the capacitor between the gate of the upper transistor and the drain of the lower transistor should be of great importance, otherwise the amplifier will cut off a lot of low frequencies, I managed to get a flat response at 22uF, with small values ​​(47nF-100nF) it looks like a treble booster. Having learned this and the effect of the feedback resistor in the first stage, I understood why Catalinbread products sound so "square", because a lot of high and low frequencies are lost in the design.

[POTL]

I am into AMT effects

Well, that was an interesting Google search.   

Hello. I don't want to frame the effect. in fact I am into AMT effects, which can be used both in the input of an amplifier, as a distortion pedal, and in an effects loop, as a preamp. Ideally, use the pedal in conjunction with an impulse cabsim to play in a line. When it comes to playing the loop, I'm not sure if 18 volts will be enough for a good sound, so I'm basically looking to use this effect into the amp's input. I tried many methods several years ago, now I decided to delve into the theory and understand why certain methods were better or worse.

I think Iain and others of us are curious because a given pedal output will sound very different going into the input of an amp, the effects loop, or into a DI.  It's a pretty safe bet that if you optimize it to sound good in one of those cases, it'll sound bad in another.  If you are trying to incorporate a full tone stack and cab sim EQ into the sound and then run that into an amp's input, you're likely to double-boost the bass and treble.  The Boss Metal Zone is a good (cautionary?) example:

And I think I didn't love the EHX English Muff'n for the same reason, you end up with two preamps trying to determine the EQ.

So it's not just to pin you down.  The term "amp in a box" is definitely overloaded, but I tend to think of it not literally but: what distortion pedal can give me the illusion of switching from my clean Fender to a cranked Marshall or Vox AC30?  In those cases, the pedal's EQ circuitry needs to complement what your clean amp is doing, not just duplicate the tone stack from a Marshall or what have you.

In fact, the equalizer affects significantly, there are many different and they behave in different ways. If you are plugging a distortion pedal into a clean channel, you will most likely want the flexibility of a distortion pedal (clean channels are different), and most importantly, you will want to change the clean channel EQ settings for the sake of overdrive, thereby abandoning the previously set settings for your clean channel. A cabsim to an amplifier is useless, but as a switchable option or a separate output, it can be useful.

[iainpunk]

simulation showed that SRPP has a number of disadvantages
1) It is still dependent on the value of the load resistor, the situation is slightly better than the classic Mu Amp, but much worse than the Mu Amp with buffer.

maybe compare with srpp with buffer, to make it more fair, haha, but yes, the load impedance has effect on the output, but it should be less of an effect than with a single Jfet or mosfet gain stage. load dependancy van be a good thing too, but more on that later

2) Excessively high gain (for someone this is a plus), about 50 dB, against the conditional 17 dB for single Jfet and about 30 dB for a mosfet. It is difficult to build multi-stage circuits, with 4 or more consecutive amplification stages

gain can be reduced by a resistive divider up front, or with negative feedback. look at the feedback network like you would at a inverting opamp NFB loop.

3) To reduce the gain, you need to use either a small resistor in the feedback of the lower transistor, which greatly reduces the input impedance (this is critical for the first stage) or use a large source resistor of the lower transistor (100 kΩ or more).

the point above or use lower gain devices like LND150 mosfets instead of Jfets, or use a low value load impedance.

4) Distortion is symmetrical and it is very difficult to simultaneously adjust the asymmetry and filters.

load impedance can be done with a resistor to either ground or Vcc, along with a capacitor coupled impedance. this ''asymmetric'' loading makes clipping asymmetric as well.

5) the capacitor between the gate of the upper transistor and the drain of the lower transistor should be of great importance, otherwise the amplifier will cut off a lot of low frequencies, I managed to get a flat response at 22uF, with small values ​​(47nF-100nF) it looks like a treble booster. Having learned this and the effect of the feedback resistor in the first stage, I understood why Catalinbread products sound so "square", because a lot of high and low frequencies are lost in the design.

i have no idea what cap you are talking about, true SRPP designs only have 2 caps, input and output caps.

hope i helped

I am into AMT effects

Well, that was an interesting Google search.   

Hello. I don't want to frame the effect. in fact I am into AMT effects, which can be used both in the input of an amplifier, as a distortion pedal, and in an effects loop, as a preamp. Ideally, use the pedal in conjunction with an impulse cabsim to play in a line. When it comes to playing the loop, I'm not sure if 18 volts will be enough for a good sound, so I'm basically looking to use this effect into the amp's input. I tried many methods several years ago, now I decided to delve into the theory and understand why certain methods were better or worse.

I think Iain and others of us are curious because a given pedal output will sound very different going into the input of an amp, the effects loop, or into a DI.  It's a pretty safe bet that if you optimize it to sound good in one of those cases, it'll sound bad in another.  If you are trying to incorporate a full tone stack and cab sim EQ into the sound and then run that into an amp's input, you're likely to double-boost the bass and treble.  The Boss Metal Zone is a good (cautionary?) example:

whai i'd do is use a ''flat'' tone stack, like a James or Bax. one straight output jack for front of amp and put a mid scooping boost unit on another output for straight into the power amp.

cheers, Iain

[POTL]

Hi Iaen,
1) maybe I confused something, here is the circuit which I call SRPP the capacitor, about which I spoke is the only one in this diagram, its value is 100nF. The diagram you posted is new for me, I have not tried it. I understand correctly that the upper transistor does not require a voltage divider at the gate, as is done in the Mu Amp circuit.
2) Sound idea about the symmetry / asymmetry of clipping and setting the voltage divider of the upper transistor, I saw this on Catalinbread, but my simulator could not show the correct result, I will try again.
3) About EQ, I don't like Baxandall / James circuits, at least with a guitar, but they work well with bass. I like the midrange cut in classic tonstacks.

[niektb]

Hi Iaen,
1) maybe I confused something, here is the circuit which I call SRPP the capacitor, about which I spoke is the only one in this diagram, its value is 100nF. The diagram you posted is new for me, I have not tried it. I understand correctly that the upper transistor does not require a voltage divider at the gate, as is done in the Mu Amp circuit.
2) Sound idea about the symmetry / asymmetry of clipping and setting the voltage divider of the upper transistor, I saw this on Catalinbread, but my simulator could not show the correct result, I will try again.
3) About EQ, I don't like Baxandall / James circuits, at least with a guitar, but they work well with bass. I like the midrange cut in classic tonstacks.

This circuit is the one that I know as the Mu-Amp

[iainpunk]

like niektb says, the image you posted isn't a normal SRPP.
the idea behind the SRPP is biasing through the voltage over the resistor coming off of the emitter. bothe E-resistors are the same, with the same current generate the same voltage over them, making the Jfets conduct the same. if you wiggle one of the voltages, you'll change the conduction of the lower transistor and through the difference in conduction between the two transistors youll amplify the input wiggle.
the basis for srpp is not needing to bias externally like the mu amp, due to a really nice self bias system. this slightly ups the gain of the stage as well as greatly improving linearity.

cheers