opamps vs transistors

[JonFrum]

Thanks for the info guys. The main reason I want to try to swap in some transistors is to have some more control over the "voice" of the gain. To whoever said "it is not possible with your current level of understanding" all I can say is my level of understanding is changing by the second so that seem like a fairly ridiculous comment. It IS possible for me to do, and I WILL do it! Thanks

[R.G.]

To whoever said "it is not possible with your current level of understanding" all I can say is my level of understanding is changing by the second so that seem like a fairly ridiculous comment. It IS possible for me to do, and I WILL do it! Thanks

That would be me.  And all I have to say for myself is...

Kewl. So not only was the statement accurate, it has inspired you to work hard at learning. Talk about a win-win!

Keep digging! 

[wampcat1]

Thanks for the info guys. The main reason I want to try to swap in some transistors is to have some more control over the "voice" of the gain. To whoever said "it is not possible with your current level of understanding" all I can say is my level of understanding is changing by the second so that seem like a fairly ridiculous comment. It IS possible for me to do, and I WILL do it! Thanks

That would be me.  And all I have to say for myself is...

Kewl. So not only was the statement accurate, it has inspired you to work hard at learning. Talk about a win-win!

Keep digging! 

Great answer RG

Burningwater, RG isn't being sarcastic... what he's saying is this. What you are asking is analagous to a someone asking a mechanic how to convert his bicycle into a car since they both do the same basic thing...that is, move the wheels. See what I mean? The answer to your question would involve directing you to geofex.com and having you read every tutorial there so you can get more of a grasp on just what a transistor is and just what an opamp is.

No offense, just trying to explain!

bw

[burningwater]

sfr's link is a great article

mac's link to davidsons work is exactly what I am looking to do

Thanks yall.

[wampcat1]

then you want this schem:
http://www.diystompboxes.com/analogalchemy/sch/diodeopamp.html

however, if you want to build a single opamp circuit to a t, then you want to build the schem in this doc:
http://www.st.com/stonline/books/pdf/docs/2296.pdf

[Steben]

Maybe I haven't read it enough between the lines, yet I want to add some little things.
1. Transistor characteristics vary a lot. Germanium bipolar, Silicon, FET's... all respond different. If not used in complicated circuitry, they all add a fairly large amount of distortion. In simple "discrete" amplifiers, if you put input voltage next to output, you will never get a straight line (= no distortion at all), but a curved one. "Linearity" often means "least non-linearity". Most tubes offer in fact the most linear response. As we speak of clipping on the other hand, bipolars get often nasty, especially the ones ment for "audio". FET's are more gentle.
2. OPAMP characteristics vary very little. At least those we are interested in. They all add a very small amount of distortion, most very little noise. Opamps are almost linear, and for guitar use they are perfect linear. As we speak of clipping, most opamps are worse than bipolars. There are some, however, that clip sharp, yet not chaotic and are used in distortion boxes.
3. Some circuits act a little as opamps (stress on external component to determine gain etc...), yet offer better musical qualities. Discrete feedback designs, CMOS inverters,... All are a bit used as opamps, but get nice clipping.

[Jay Doyle]

Maybe I haven't read it enough between the lines, yet I want to add some little things.
1. Transistor characteristics vary a lot. Germanium bipolar, Silicon, FET's... all respond different. If not used in complicated circuitry, they all add a fairly large amount of distortion. In simple "discrete" amplifiers, if you put input voltage next to output, you will never get a straight line (= no distortion at all), but a curved one. "Linearity" often means "least non-linearity". Most tubes offer in fact the most linear response. [quote[
2. OPAMP characteristics vary very little. At least those we are interested in. They all add a very small amount of distortion, most very little noise. Opamps are almost linear, and for guitar use they are perfect linear. As we speak of clipping, most opamps are worse than bipolars. There are some, however, that clip sharp, yet not chaotic and are used in distortion boxes.
3. Some circuits act a little as opamps (stress on external component to determine gain etc...), yet offer better musical qualities. Discrete feedback designs, CMOS inverters,... All are a bit used as opamps, but get nice clipping.

OK, I've got to disagree with some of the things above:

Germanium bipolar, Silicon, FET's... all respond different.

Agreed, with the caveat that you end that sentence with 'when overdriven'. I would defy you to be able to pick out each of those with a full spectrum response and a gain of 5, blind.

If not used in complicated circuitry, they all add a fairly large amount of distortion.

The complication of the circuitry surrounding them has nothing to do with anything. The distortion, when operating within the headroom limits, is pretty low for a single Class A stage.

In simple "discrete" amplifiers, if you put input voltage next to output, you will never get a straight line (= no distortion at all), but a curved one.

In any situation with any kind of gain be it voltage or current you are NEVER going to get a straight line as you call it. There is no 'straight wire with gain' EVERYTHING imparts it's own distortion onto the signal, this is true of opamps and yes, tubes as well.

As we speak of clipping on the other hand, bipolars get often nasty, especially the ones ment for "audio". FET's are more gentle.

This is a vast generalization. Look at Gus Smalley's NPN Boost for a bipolar that gets great distortion and it is simply clipping, then try a JFET with a CCS load and I would reverse your generalizations. It all depends on the way the stage is set up if you are comparing a JFET stage with a gain of 20 and a BJT stage with a gain of 200, yeah, sure the JFET is going to seem more gentle, but you are asking something completely different.

They all add a very small amount of distortion, most very little noise.

Here I have to wholeheartedly disagree with you. Opamps are much more noisier than single discrete circuits. More active devices in an opamp = More noise. And the small amount of distortion comes from feedback, which isn't unique to opamps, you can improve the linear response of any circuit by adding negative feedback.

Opamps are almost linear, and for guitar use they are perfect linear

This has nothing to do with the opamp, but the negative feedback used, see above. And if they are 'perfectly linear' does that make them better than your statement that: "Most tubes offer in fact the most linear response"?

As we speak of clipping, most opamps are worse than bipolars. There are some, however, that clip sharp, yet not chaotic and are used in distortion boxes.

Again, a vast generalization. And can you please point me to the opamps that are used for their clipping in distortion boxes? And I mean the clipping, not the exploitation of certain adjustments available to the opamps used (the slew rate distortion of the 308 in the Rat, or the bias trickery that the Hotcake uses?)

Some circuits act a little as opamps (stress on external component to determine gain etc...), yet offer better musical qualities. Discrete feedback designs, CMOS inverters,... All are a bit used as opamps, but get nice clipping.

No, some circuits have the external components determine the gain but offer better musical qualities (in your opinion). Having the external components determine the gain does not an opamp make.

[Ben N]

And can you please point me to the opamps that are used for their clipping in distortion boxes? And I mean the clipping, not the exploitation of certain adjustments available to the opamps used (the slew rate distortion of the 308 in the Rat, or the bias trickery that the Hotcake uses?)

Various Tech21 pedals use opamp clipping, IIRC, as does the Tube Driver (ironic). I'm pretty sure the XXL does. I'm not familiar with "the bias trickery that the Hotcake uses" but always open to learn!

Ben

[alextheian-alex]

There are a lot of options out there.  Pedals that were designed around opamps will only sound the same with opamps... discreet or IC.  Same goes for plain old single fets and BJTs.  you can alter the tone slightly with different opamps or discreets, or by subbing out vacuum tubes though... and that is where the fun starts!!  If you wanna build up a few different discreet opamps... go for it!  If you want to make them out of vacuum tubes, even more fun!  if you want to substitute an opamp for a single BJT or FET... SWEET!  try it!
Here is another perspective on this:

There are a zillion commercial discreet opamp designs out there, and i have seen heard and built as many as i could... many of which are very very good, like the API2520 and JE990, which have been in service for about a zillion years... Ok maybe 20 or so.  The benefit is not necessarily "plug and play" but discreets are fun to tinker with and let you shave away what you don't want, as well as tweek what you DO want.  The models that i mentioned are used in high-dollar studio applications and low level audio gain stages where the IC opamps on the market just were not good enough.  Lemme say now though... when you get into the "money is no object" world, then a .01% increase in function costing a 1000% increase in cost is no biggie, so don't neccessarily expect to be TOTALLY blown away by a discreet opamp.  however, many old pedals had seemingly complicated amplifying stages, but when you follow the signal through the circuit, you see that they are just primative discreet opamps, and i say primative because they usually lack the current sinks and mirrors and other trappings that give modern ICs a lot of their stability, so it is not unheard of or even uncommon in pedals.

The caveats: Most IC opamps run an AB PP output which is biased to be linear withing reason, but many pedal circuits push it well beyond that region... often out of AB and into class B--and that is where things get ugly.  That is usually my biggest problem with them.  In many cases, you can force them into class A--although you have to watch your dissapation--which is worth experimenting with if you are annoyed by the upper odd order harmonic artifacts... or FIZZ as it is known more commonly. But even though you are forcing class A, the output us still PP, so you are cancelling out any even order harmonics that the output stage generates.

Another problem with op-amps is that they have incredibly high open loop gain... that does not sound too bad, but to stabalize all that gain, you have to use insane amounts of feedback.  So if you are generating odd order harmonics, canceling out even order, and then feeding the signal back into the input... well you can see what happens there.  The harmonics are lower in proportion to the fundamental, but the Fourier series tapers off at a gentler slope than just a single transistor... which means that those upper harmonics are just hanging out.  They are usually inaudible in the linear region, but when you clip it... then you are essentially bringing them up compared to the lower order harmonics and the fundamental... ICK!  That is one of the reasons for little treble bleed snubber band-aid capacitors all over.  A 741 wired up like an MXP Dist+ with a 1v RMS input signal will clip like mad, and give you odd harmonics up into the ridiculously high orders.  When i measured it, without filtering, the 11th harmonic was only around 2-3dB lower than the 2nd harmonic!  A discreet opamp has the same issues, but you can optimize it for minimal odd harmonics if you are slick about it.

now talking about gain.   Yes, opamps will give you a ton of gain, but we don't really use it all.  Look at the tubescreamer, the first stage opamp is configured for a gain just a bit over 100x at max... you don't need an opamp for that.  You can get that from ANY fet or BJT and many vacuum tubes.  A dist+ is just over 200x at max, again, you can easily milk that from any fet or BJT or vacuum tube signal pentode.  So that is just to say that you have options.  You CAN drop a transistor or tube in the place of an opamp in a lot of cases.  heck, I have seen univibe clones using only tubes!  I'll probably build one eventually.

As far as the tube opamp thing goes... yeah, it can work.  Is it worth the hassle... i dunno.  I have done it in the past and will in the future do it again, but for very specific reasons.  There are a lot of tube preamps and drivers out there that are just discreet opamps, they are just not called that: as in a differential input system with a follower output using global and  local feedback to stabalize the high open loop gain.  The benefit of a tube is the transfer function, which empahsizes even order harmonics.  In an opamp form, the feedback cancells out a lot of that 2nd order since that is part of it's purpose (usually THD measurements only take the 2nd harmonic into account with the figures).  if it uses a push pull follower output stage, then you are not gaining any "sweetness" there either, and at hard clipping, with all that feedback, you are not really any better off than a transistor opamp.  But if you only use the feedback that you need, and you don't use a PP follower, then they can be cool.

better use of a tube is where you are amplifying intermediate-gain signals with no feedback, that will be re-amplified later on.  like an input buffer (remember, tubes have a nearly infinate input resistance... good for buffers) or early gain stage.  Then let the solid state components take over and do their thing.

So how would I use a tube in an established circuit... a pentode in the place of the 741 in an MXR definately.  Also as the input and output buffers of a tubescreamer.   in the amplifying stages of modulation effects most certainly.  And don't forget about tube diode clipping!

[R.G.]

Alex, maybe I'm running slow tonight, but I don't understand some of what you said. To wit:

Most IC opamps run an AB PP output which is biased to be linear withing reason, but many pedal circuits push it well beyond that region... often out of AB and into class B--and that is where things get ugly.

I don't know of any way other than an internal bias shift to get a Class AB circuit into pure Class B. Tube circuits have bias shift, OK, but in solid stage AB output stages, this is quite difficult to do. Can you explain how this happens?

In many cases, you can force them into class A--although you have to watch your dissapation--which is worth experimenting with if you are annoyed by the upper odd order harmonic artifacts... or FIZZ as it is known more commonly. But even though you are forcing class A, the output us still PP, so you are cancelling out any even order harmonics that the output stage generates.

Presumably this is the old resistor-to-Vminus trick? Cancellation of even order harmonics only happens as a sure thing to my knowledge in transformer coupled stages. Class A stages may or may not cancel, depending on the matching of the output devices and how well the low current curves for the top device mismatch the high current devices for the lower device, and vice versa.

I'm not saying it can't happen, I just can't see how. Could you explain this?

Another problem with op-amps is that they have incredibly high open loop gain... that does not sound too bad, but to stabalize all that gain, you have to use insane amounts of feedback.

Actually, stability is not related to open loop gain so much as it is to phase shift and the Nyquist criterion of less than 180 degrees of phase shift before gain is under unity. High gain makes it harder to get gain down to unity sometimes, but more feedback makes it harder to maintain stability, not less. We were taught that stabilizing a gain-of-one opamp is harder than a gain of five. Some industrial opamps are internally stabilized for gains above X (often 5 or 10) but not unity. So could you explain needing to use more feedback to gain stability?

So if you are generating odd order harmonics, canceling out even order, and then feeding the signal back into the input... well you can see what happens there.

Actually, I can't. Can you help?

The harmonics are lower in proportion to the fundamental, but the Fourier series tapers off at a gentler slope than just a single transistor... which means that those upper harmonics are just hanging out.  They are usually inaudible in the linear region, but when you clip it... then you are essentially bringing them up compared to the lower order harmonics and the fundamental... ICK!  That is one of the reasons for little treble bleed snubber band-aid capacitors all over. 

Actually, overdriven single transistors flat top pretty dramatically too. The raw sound of overdriven bipolars was a classic sound of the 60's fuzz boxes.

A 741 wired up like an MXP Dist+ with a 1v RMS input signal will clip like mad, and give you odd harmonics up into the ridiculously high orders.  When i measured it, without filtering, the 11th harmonic was only around 2-3dB lower than the 2nd harmonic! 

Was the opamp driven into clipping at that input? Opamps which bang onto the power supply rails are using their tremendous open loop gain to keep the output linear until they just can't any more, I thought. So the clipping is very, very sharp cornered, and that means very high ordered harmonics to construct that sharp corner. I guess I don't understand your premise here. Of course overdriven opamps generate huge high order harmonics - that's how they're designed, I thought.

A discreet opamp has the same issues, but you can optimize it for minimal odd harmonics if you are slick about it.

How would I do that?

now talking about gain.   Yes, opamps will give you a ton of gain, but we don't really use it all.

But don't they use it all the time? It's the high open loop gain that's fed back to make the linear region really really linear. It's because the gain can be thrown away that they get really linear, isn't it?

heck, I have seen univibe clones using only tubes!  I'll probably build one eventually.

Good for you. I have built one. Go for it.

(usually THD measurements only take the 2nd harmonic into account with the figures).

Really? All of the distortion analyzers I've seen that are not Fourier series based work by cancelling only the fundamental as carefully as they can, so the residual is only the distortion - all of it, including all the harmonics. In fact, I thought that the term "Total Harmonic Distortion" arose because it was simpler with analog instruments to measure all of them together by canceling the fundamental, leaving only the stuff that was not the fundamental. There used to be "wave analyzers" which were very narrow band filters to measure specific harmonics, which they had to do one at a time, by hand. So do modern THD analyzers measure only second harmonic? How do they do that?

Could you help me out with these?

[brett]

So do modern THD analyzers measure only second harmonic?

T=Total. 
But T might also stand for Two or Twice or Twin.....  Not.   

[alextheian-alex]

Alex, maybe I'm running slow tonight, but I don't understand some of what you said. To wit:

1)Most IC opamps run an AB PP output which is biased to be linear withing reason, but many pedal circuits push it well beyond that region... often out of AB and into class B--and that is where things get ugly.

I don't know of any way other than an internal bias shift to get a Class AB circuit into pure Class B. Tube circuits have bias shift, OK, but in solid stage AB output stages, this is quite difficult to do. Can you explain how this happens?

2)In many cases, you can force them into class A--although you have to watch your dissapation--which is worth experimenting with if you are annoyed by the upper odd order harmonic artifacts... or FIZZ as it is known more commonly. But even though you are forcing class A, the output us still PP, so you are cancelling out any even order harmonics that the output stage generates.

Presumably this is the old resistor-to-Vminus trick? Cancellation of even order harmonics only happens as a sure thing to my knowledge in transformer coupled stages. Class A stages may or may not cancel, depending on the matching of the output devices and how well the low current curves for the top device mismatch the high current devices for the lower device, and vice versa.

I'm not saying it can't happen, I just can't see how. Could you explain this?

3)Another problem with op-amps is that they have incredibly high open loop gain... that does not sound too bad, but to stabalize all that gain, you have to use insane amounts of feedback.

Actually, stability is not related to open loop gain so much as it is to phase shift and the Nyquist criterion of less than 180 degrees of phase shift before gain is under unity. High gain makes it harder to get gain down to unity sometimes, but more feedback makes it harder to maintain stability, not less. We were taught that stabilizing a gain-of-one opamp is harder than a gain of five. Some industrial opamps are internally stabilized for gains above X (often 5 or 10) but not unity. So could you explain needing to use more feedback to gain stability?

4)So if you are generating odd order harmonics, canceling out even order, and then feeding the signal back into the input... well you can see what happens there.

Actually, I can't. Can you help?

5)The harmonics are lower in proportion to the fundamental, but the Fourier series tapers off at a gentler slope than just a single transistor... which means that those upper harmonics are just hanging out.  They are usually inaudible in the linear region, but when you clip it... then you are essentially bringing them up compared to the lower order harmonics and the fundamental... ICK!  That is one of the reasons for little treble bleed snubber band-aid capacitors all over. 

Actually, overdriven single transistors flat top pretty dramatically too. The raw sound of overdriven bipolars was a classic sound of the 60's fuzz boxes.

6)A 741 wired up like an MXP Dist+ with a 1v RMS input signal will clip like mad, and give you odd harmonics up into the ridiculously high orders.  When i measured it, without filtering, the 11th harmonic was only around 2-3dB lower than the 2nd harmonic! 

Was the opamp driven into clipping at that input? Opamps which bang onto the power supply rails are using their tremendous open loop gain to keep the output linear until they just can't any more, I thought. So the clipping is very, very sharp cornered, and that means very high ordered harmonics to construct that sharp corner. I guess I don't understand your premise here. Of course overdriven opamps generate huge high order harmonics - that's how they're designed, I thought.

7)A discreet opamp has the same issues, but you can optimize it for minimal odd harmonics if you are slick about it.

How would I do that?

8)now talking about gain.   Yes, opamps will give you a ton of gain, but we don't really use it all.

But don't they use it all the time? It's the high open loop gain that's fed back to make the linear region really really linear. It's because the gain can be thrown away that they get really linear, isn't it?

9)
heck, I have seen univibe clones using only tubes!  I'll probably build one eventually.

Good for you. I have built one. Go for it.

10)(usually THD measurements only take the 2nd harmonic into account with the figures).

Really? All of the distortion analyzers I've seen that are not Fourier series based work by cancelling only the fundamental as carefully as they can, so the residual is only the distortion - all of it, including all the harmonics. In fact, I thought that the term "Total Harmonic Distortion" arose because it was simpler with analog instruments to measure all of them together by canceling the fundamental, leaving only the stuff that was not the fundamental. There used to be "wave analyzers" which were very narrow band filters to measure specific harmonics, which they had to do one at a time, by hand. So do modern THD analyzers measure only second harmonic? How do they do that?

Could you help me out with these?

WOW!  Quite a long reply... thanx for taking the time to care, honestly... cudos to you.  i numbered the replys instead of creating an endless run of quote boxes.  Lemme see if i can clear up what I was talking about:

1. It does happen.  I have a paper somewhere that outlines the crossover distortion in poor op-amps, specifically the 741 and it's variants. It is known as dynamic crossover distortion.  It happens with large signal changes in slower chips with poor compensation networks.


2.  Yeah, you can do it with a resistor, but a current source is much better.  Any stage with a symmetrical transfer function will only cancel even order harmonics that are generated internally WITHIN THAT STAGE.  It will not cancell the harmonics of the input signal, if that is what you were getting at--probably jst poor wording on my part.  And yes, you are right on that the better the balance of the devices, the better the cancellation.

3. Sorry about the mixup there... bad wording. I should not have referred to open loop gain.  I was talking more about internal feedback.

4. Well... feedback is used to linearize the circuit the improved transfer function will cause a decrease in THD, but it is much more effective at squelching low level harmonic distortion, so since 1)we are driving the output into saturation 2) we are cancelling out even harmonics that are generated within the clipping output stage and 3) lower order harmonics are getting cancelled more effectively than higher order, then we are left with primarily higher order, primarily odd order harmonics.

5. True, but if we are talking clipping a single ended circuit vs clipping a complimentary pair, then there will always be more even order harmonics in the signal, AND if there is no feedback, then we are not squelching the more pleasing lower order harmonics that are generated.

6.  I think we are saying the same thing here.  I was just pointing out that the higher order harmonics were nearly as loud as the lower order ones.

7. By designing the circuit with tight tolerance components, that are matched where they need to be, and adjusting each stage to it's optimal operating range, perhaps using a single ended output, all to minimize distortion operate it a in as linear a way as possible so that you can reduce the neccessary feedback.  You can't open up an IC to make sure all is well in there, and there is nothing you can do about the contact noise, shot noise, etc of the teeny tiny little parts in there.

8.  Yes, the opamps use their own gain all the time, but we don't adjust them for maximum gain signal.  We usually don't run them at higher signal gains that can be had from discreets.

9. SWEET!  i have been wanting to do it for a while and just have never had the time.

10. Yes, that is all true you are absolutely right.  I was referring to the rather unscrupulous marketing behavior of many companies over the years that base their posted measurements on the dominant harmonics, or weight the scale of the upper harmonics so that they can post a more impressive spec... sorry about the confusion.

[brett]

I was referring to the rather unscrupulous marketing behavior of many companies over the years that base their posted measurements on the dominant harmonics, or weight the scale of the upper harmonics so that they can post a more impressive spec...

I feel that it is harsh to refer to these "many companies" as unscrupulous in trying to make their products look good, especially when you've just used the same terminology to try to make a more impressive post look good.

Concerning:

which means that those upper harmonics are just hanging out.  They are usually inaudible in the linear region, but when you clip it...

What, in layman's terms, do "hanging out" mean?  And what are the implications when they are both "Hanging out" and "inaudible" at the same time?

thanks

[RaceDriver205]

which means that those upper harmonics are just hanging out.  They are usually inaudible in the linear region, but when you clip it...

Argh! Audio super-jargon!

[Steben]

Germanium bipolar, Silicon, FET's... all respond different.

Agreed, with the caveat that you end that sentence with 'when overdriven'. I would defy you to be able to pick out each of those with a full spectrum response and a gain of 5, blind.

No, I couldn't do that, but it doesn't harm my quote. I didn't generalize that they always "sound" different, they "respond" different, meaning the functioning of transistors differ a lot and will explain a lot. FET's are square law function (as tubes), BJT's are logarithmic.

If not used in complicated circuitry, they all add a fairly large amount of distortion.

The complication of the circuitry surrounding them has nothing to do with anything. The distortion, when operating within the headroom limits, is pretty low for a single Class A stage.

It's all in figures. Maybe I was too much speaking in HiFi terms. Yet a class A, un-feeded-back ;-), has a gentler THD slope, meaning it doesn't go from clean linear to harsh clipping. As I said earlier, nothing will ever be completely linear (THD=0). The more feedback, the less distortion in the "linear" region, the harsher switching between linear and clipping.

In simple "discrete" amplifiers, if you put input voltage next to output, you will never get a straight line (= no distortion at all), but a curved one.

In any situation with any kind of gain be it voltage or current you are NEVER going to get a straight line as you call it. There is no 'straight wire with gain' EVERYTHING imparts it's own distortion onto the signal, this is true of opamps and yes, tubes as well.

true, I ment to say: class A without any feedback, nor CCS, will be the least linear. And I repeat triode tubes have the most linear response without feedback/CCS "help".

As we speak of clipping on the other hand, bipolars get often nasty, especially the ones ment for "audio". FET's are more gentle.

This is a vast generalization. Look at Gus Smalley's NPN Boost for a bipolar that gets great distortion and it is simply clipping, then try a JFET with a CCS load and I would reverse your generalizations. It all depends on the way the stage is set up if you are comparing a JFET stage with a gain of 20 and a BJT stage with a gain of 200, yeah, sure the JFET is going to seem more gentle, but you are asking something completely different.

Let's not speak of feedback and CCS. I was talking about the functioning of bipolars vs FET's. FET's are resistor units that are driven with a reference voltage like a tube. IF there is no headroom left, they clip, and they return from clipping as soon as the headroom comes back. A Bipolar is not a resistor, at least not a reliable one. It is a current driven unit that gets "saturated", it doesn't just clip. And that's a different dog. Depending on the specific qualities of your BJT, a large or small (switching transistor) amount of electrons will be collected during saturation and when the input signal goes down again and you would expect the output signal to come out of "clipping", it stays a while there at the rail, because of the gathered electrons. Latch-up...(google away i would say) This gives a very unnatural distortion. Some opamps have this too, some are especially designed not to have it.

They all add a very small amount of distortion, most very little noise.

Here I have to wholeheartedly disagree with you. Opamps are much more noisier than single discrete circuits. More active devices in an opamp = More noise. And the small amount of distortion comes from feedback, which isn't unique to opamps, you can improve the linear response of any circuit by adding negative feedback.

I'm not sure of that. the (old) 741 is a very noisy bitch. But I defy you to produce a better simple discrete circuit that beats a NE5534 opamp stage with a couple of resistors. And try to do it with germanium...

Opamps are almost linear, and for guitar use they are perfect linear

This has nothing to do with the opamp, but the negative feedback used, see above. And if they are 'perfectly linear' does that make them better than your statement that: "Most tubes offer in fact the most linear response"?

I agree. Yet opamps are inherently used with feedback, discrete (class A fe) not. I speak of opamps with feedback as a natural fact. triode tubes offer most linearity "without" feedback.

As we speak of clipping, most opamps are worse than bipolars. There are some, however, that clip sharp, yet not chaotic and are used in distortion boxes.

Again, a vast generalization. And can you please point me to the opamps that are used for their clipping in distortion boxes? And I mean the clipping, not the exploitation of certain adjustments available to the opamps used (the slew rate distortion of the 308 in the Rat, or the bias trickery that the Hotcake uses?)

Some circuits act a little as opamps (stress on external component to determine gain etc...), yet offer better musical qualities. Discrete feedback designs, CMOS inverters,... All are a bit used as opamps, but get nice clipping.

No, some circuits have the external components determine the gain but offer better musical qualities (in your opinion). Having the external components determine the gain does not an opamp make.

You mix gain with harshness. Having the external components determine the gain does not a harshness make.

[brett]

Hi

But I defy you to produce a better simple discrete circuit that beats a NE5534 opamp stage with a couple of resistors.

If op-amps can be made very quiet, could you work some magic for me on this cicuit for me, please?  It's a noisey oyster.

I'd use the NE5532, but the input impedance is too low.  I guess the BJT input stage is quiet, but causes other problems.

[Steben]

Hi

But I defy you to produce a better simple discrete circuit that beats a NE5534 opamp stage with a couple of resistors.

If op-amps can be made very quiet, could you work some magic for me on this cicuit for me, please?  It's a noisey oyster.

I'd use the NE5532, but the input impedance is too low.  I guess the BJT input stage is quiet, but causes other problems.

I'm a bit confused about the input of A2. I understand you want to bias it with the output of A1, right? Never forget that DC over a pot crackles like hell, regardless the opamp choice.. On the other hand the gain is (10+2.2)/2.2 = 5.5 times (1000+2.2)/2.2 = 455.5, which means a stunning total of over 2505. I see you reduce the first gain back with at max settings a resulting gain of 1.25, giving a total of around 580. Yet you bring back a first gain (yet added noise from opamp) and you do it with noise-inducing resistors, in combination with a following high gain. If that doesn't give noise, nothing does. Don't forget that a guitar's own noise is audible at those gain levels, it is even giving the main source of noise. If you bring that 1M resistor back to 100k to 150k, the noise will be gone without a lot of difference.

Awkward cicruit, by the way. It's almost like you raise the distortion by lowering the gain pot. With this gain, when de diodes start conducting you get the original signal, but with a shift of 0.7V. The smaller the original signal, the flatter the resulting top, giving more distortion.

The tubescreamer style is simple and attractive, yet treacherous. There is always the danger of stepping into crossover distortion if the signal gets big in comparison to the treshold of the diodes. That's why germaniums sound so sloppy in a TS.

[alextheian-alex]

It seems that some people took my post as ann affront of some kind... I am just offering another opinion.   The thread is transistors vs opamps... and wandered off to include tubes as well, so i threw in my $.02.  We are all just trying to help each other here.

I was referring to the rather unscrupulous marketing behavior of many companies over the years that base their posted measurements on the dominant harmonics, or weight the scale of the upper harmonics so that they can post a more impressive spec...

I feel that it is harsh to refer to these "many companies" as unscrupulous in trying to make their products look good, especially when you've just used the same terminology to try to make a more impressive post look good.

OUCH!  I am not trying to ruffle anyones feathers here.  I am just a little bitter and jaded about the marketing side of the audio world.   Sometimes it seems like it is totally up to the amateur audio crowd to test these things and blog the truth since it seems impossable to get the straight truth from manufacturers.

Concerning:

which means that those upper harmonics are just hanging out.  They are usually inaudible in the linear region, but when you clip it...

What, in layman's terms, do "hanging out" mean?  And what are the implications when they are both "Hanging out" and "inaudible" at the same time?

thanks
[/quote]

HEH... just trying to keep things lighthearted by using non-techie lingo.  By "hanging out" i meant that they are definately there, but they are very low, and become more prominant with clipping.

[puretube]

commonly known opamps consist of at least 2 transistors on a dye, & tubes are valves...   

[Caferacernoc]

"Again, a vast generalization. And can you please point me to the opamps that are used for their clipping in distortion boxes? And I mean the clipping, not the exploitation of certain adjustments available to the opamps used (the slew rate distortion of the 308 in the Rat, or the bias trickery that the Hotcake uses?)"

Nearly every diode clipper stompbox has plenty of distortion even when you remove the clipping diodes. So the opamp IS being used for it's clipping. MXR dist+, Rat, Tubescreamer, DOD OD250, Boss DS1, OD1...... Ross, Sansamp. I'm not saying they CHOSE the opamp for it's clipping properties, but those clipping properties are being used. A lot.
Why else would people care about the JRC chip in the TS808? Or care that the reissue MXR and DOD OD250 don't use the 741?