SS amps - Discrete transistor?

[raphaeldc]

Hi,

Do you know wich companies are using discrete transistors in its amps (preamp and power section), instead of integrated circuits?

Cheers,

Raphael

[seamonkey]

New ones?

Or are you looking for an old Solid State Amp?

FET or Bipolar?

How good is your hearing?

[bongomania]

While different components can sound different, there is absolutely no reason to think that a transistor will sound better or perform more reliably just because it's "discreet". In fact most large-scale manufacturers will have chosen the cheapest transistors they can find, discreet or not.

I guess I could see the question making sense if you were planning on replacing all the components yourself, as a hobby project; though really it's not much easier than replacing an IC.

[raphaeldc]

Originally Posted by bongomania

While different components can sound different, there is absolutely no reason to think that a transistor will sound better or perform more reliably just because it's "discreet". In fact most large-scale manufacturers will have chosen the cheapest transistors they can find, discreet or not.

I guess I could see the question making sense if you were planning on replacing all the components yourself, as a hobby project; though really it's not much easier than replacing an IC.

I get your point... the thing is that I had an experience in the past that led me to think about that. I studied electronics during high school, but since it was a long time ago, there're a lot of thing I don't remember anymore.

But like I said, I fixed and old combo I had, but it was really hard to find the same IC to replace. I know that I could have the same problem with a transistor, but it seems to me that discreet components make repairing easier, because you can identify the problem in a easier way, IMO... and get it fixed!

But I see that there's no huge difference in sound or reliability.

Anyway, I've researching about that and I see some manufacturers telling that they're using discreet components for tone issues, like Genz Benz and GK. Why do you think they're going that way? And I've also heard that FET transistors behave quite like a tube, regarding to tone matters. Do you agree?

[fdeck]

Many of the discrete transistors in convenient packages for hand wiring are disappearing.

FETs and triodes have similar characteristics, so it is possible that one could achieve some resemblance in the tone of circuits made from those devices.

Here you go, boo-teek point to point wiring, and all discretes:

[billfitzmaurice]

Originally Posted by raphaeldc

Do you know wich companies are using discrete transistors in its amps (preamp and power section), instead of integrated circuits?

I doubt if any major manufacturers do.

[bongomania]

Originally Posted by raphaeldc

...I see some manufacturers telling that they're using discreet components for tone issues, like Genz Benz and GK. Why do you think they're going that way?

It's mostly marketing, because musicians are superstitious and they like to believe that older ways are always better.

Originally Posted by raphaeldc

I've also heard that FET transistors behave quite like a tube, regarding to tone matters. Do you agree?

They certainly can be used that way--it depends on how the circuit overall is designed. The Mesa Walkabout is a classic example of FETs being used to emulate tubes. But not all FET circuits are "tubelike" (which itself means a hundred different things, to different people); and FETs are not inherently discrete.

[grendle]

All about cost savings. Reliability is better if they can figure out how to solder lol. That's actually the limiting factor , the physical connection between devices.

[warnergt]

Originally Posted by raphaeldc

...I fixed and old combo I had, but it was really hard to find the same IC to replace...

Between the Internet, Google and mouser.com, I think it is
easier than ever to find replacement ICs. Amps with ICs are
most definitely serviceable even for an old school technician.

They can pack a lot in an IC. And the science has reached
a point where they can do it with incredibly low noise. Going
with discrete transistors just doesn't seem worth the effort
for a new design.

[Jerrold Tiers]

Originally Posted by bongomania

While different components can sound different, there is absolutely no reason to think that a transistor will sound better or perform more reliably just because it's "discreet".

Actually, I do NOT 100% agree with this..... and I have reasons.

if you have a "discrete" part, used as an amplifying stage in a preamp, for instance, the performance of the part tends to be due to its "inherent characteristics", and how they interact with the circuit.

Faults, non-linearities, behavior when overdriven, those all are based on the inherent part performance. What it is, it is, good or bad. Overdriven, it tends to produce increasing distortion as level increases towards overdrive, and finally becomes quite distorted as it clips. But even that is due to inherent character, good or bad.

With an IC, the basic part when used as an amplifier is very linear, rapidly becoming non-linear when it "clips". Any attempt to make it perform "like a discrete part (usually a tube) performs" is essentially always by some sort of "simulation", using even more parts to alter the basically linear performance. That new performance characteristic is not inherent to the part, and the 'simulation" only works within certain limits (usually below "clipping"). Outside of those limits, in clipping, for instance, the IC often just acts like an IC again.

The discrete part, if it is in what is basically a good circuit, has no limits beyond which it changes character..... the "simulation limits" cannot be exceeded because there IS no simulation.

A bad circuit is a bad circuit. I have heard horrible all-tube amplifiers..... But a good discrete part circuit can be relied upon to stay "in character" far better than a "simulation".

[raphaeldc]

Originally Posted by Jerrold Tiers

Actually, I do NOT 100% agree with this..... and I have reasons.

if you have a "discrete" part, used as an amplifying stage in a preamp, for instance, the performance of the part tends to be due to its "inherent characteristics", and how they interact with the circuit.

Faults, non-linearities, behavior when overdriven, those all are based on the inherent part performance. What it is, it is, good or bad. Overdriven, it tends to produce increasing distortion as level increases towards overdrive, and finally becomes quite distorted as it clips. But even that is due to inherent character, good or bad.

With an IC, the basic part when used as an amplifier is very linear, rapidly becoming non-linear when it "clips". Any attempt to make it perform "like a discrete part (usually a tube) performs" is essentially always by some sort of "simulation", using even more parts to alter the basically linear performance. That new performance characteristic is not inherent to the part, and the 'simulation" only works within certain limits (usually below "clipping"). Outside of those limits, in clipping, for instance, the IC often just acts like an IC again.

The discrete part, if it is in what is basically a good circuit, has no limits beyond which it changes character..... the "simulation limits" cannot be exceeded because there IS no simulation.

A bad circuit is a bad circuit. I have heard horrible all-tube amplifiers..... But a good discrete part circuit can be relied upon to stay "in character" far better than a "simulation".

I agree with... well, from what I've learned in the past, that's what I'd expect from a discrete component circuit. From what I recall, regarding to transistor, they usually have their operation influenced by various conditions, and I don't think that ICs are subjected to all these conditions... that can be good or bad, but also can be set in way to be good, i think...

I also agree that cutting costs and new technologies are making everything smaller, but every major manufacturer is still using tubes in their designs. So why not "go back" and use transistors in their preamps and poweramps? Don't they last longer than tubes?

[Rick Auricchio]

To the OP: Many modern amps use carefully-matched output transistors, so replacing one often means you need to replace them all. And you often won't find matched parts available; manufacturers match the parts in-house.

[agedhorse]

Originally Posted by Rick Auricchio

To the OP: Many modern amps use carefully-matched output transistors, so replacing one often means you need to replace them all. And you often won't find matched parts available; manufacturers match the parts in-house.

IMO, not properly designed ones.

There are design techniques that we can use to minimize the effects of "tolerance spread" to the point that close matching is un-necessary. I have been doing this for "decades", as have many other engineers.

There are indeed designs that depend on close tolerance matching, specifically current gain but then the problem becomes at what current should the match be done at, since the static gain varies with current and the curves will not match.

IME, power amps that require matched transistors to function properly tend to suffer from reliability problems especially as devices age and parameters shift from their "new" conditions. The shift might be quite small but for some amps this may be enough to cause problems. For amps that do not require matching, a little gain shift isn't going to be an issue.

REgarding replacement IC's, generally if the repair tech is familiar with the availabilities of op amps that are (or have been) used in audio work, suitable replacements are easy (and cheap) to secure. Same thing applies to most transistors. The exception to transistors are some older power MOS-FETS that have been discontinued due to process changes.

[bongomania]

Originally Posted by Jerrold Tiers

...if you have a "discrete" part, used as an amplifying stage in a preamp, for instance, the performance of the part tends to be due to its "inherent characteristics", and how they interact with the circuit.

You are suggesting that the transistors in an IC have no inherent characteristics????

Originally Posted by Jerrold Tiers

With an IC, the basic part when used as an amplifier is very linear, rapidly becoming non-linear when it "clips". Any attempt to make it perform "like a discrete part (usually a tube) performs" is essentially always by some sort of "simulation", using even more parts to alter the basically linear performance.

You are taking the stance that because solid state components are sometimes used to emulate tube components, that therefore there is some difference between integrated and discreet? That doesn't make sense. First off, what about integrated FETs used in a tubelike circuit? Does the fact that they are in a chip make them "not FETs"? And what about the fact that the large majority of transistors are designed to be as linear as possible--whether they are discreet or integrated? Seriously now, you can't be suggesting that a discreet transistor, designed for linearity, is somehow "more like a tube" just because it is not in a chip form?

[brendanbassist]

Originally Posted by bongomania

You are suggesting that the transistors in an IC have no inherent characteristics????

You are taking the stance that because solid state components are sometimes used to emulate tube components, that therefore there is some difference between integrated and discreet? That doesn't make sense. First off, what about integrated FETs used in a tubelike circuit? Does the fact that they are in a chip make them "not FETs"? And what about the fact that the large majority of transistors are designed to be as linear as possible--whether they are discreet or integrated? Seriously now, you can't be suggesting that a discreet transistor, designed for linearity, is somehow "more like a tube" just because it is not in a chip form?

+1

[UncleFluffy]

Originally Posted by bongomania

You are suggesting that the transistors in an IC have no inherent characteristics????

They do, but because they're all fabricated together there's a strong correlation between the characteristics of neighbouring transistors. You may not know what A is, but you do know that B is going to be 2xA. With discretes, you don't have that guarantee.

Once you've got a good handle on the ratios, you can arrange things so that variation in one part of the circuit is compensated for in another part.

[agedhorse]

Also, in an IC form, you can use topologies that are not possible (practically) in discrete. Because of inherent thermal and materials property matching, and very, very short signal paths on the die, very high gain-bandwidth product designs are possible which opens up a whole slew of potentially performance enhancing circuiut applications.

The point is to understand how each tool in the toolbox or each ingredient in the kitchen works and then to use these ingredients to the designer's advantage while avoiding the inherent pitfalls of each method.

[Jerrold Tiers]

I think you are "putting words in my keyboard"......... and missing several points.....

The main one is that IC amplifiers, typically opamps, have the "character" of the parts which must be used "around them" as negative feedback. Their makers try very hard to make that true.

By contrast, discrete transistors (or tubes), used as voltage gain stages having one or two transistors per stage, have a character due to their inherent structure and mode of operation, which the very limited amount of feedback possible in a one or two device circuit does not eliminate. There may be NO feedback at all, in which case it is obvious that "what you see is what you get".

To make the IC opamp act like the discrete device, you must add extra parts to the opamp to "simulate" or "imitate" the action of the device you want to "imitate".

Originally Posted by bongomania

You are suggesting that the transistors in an IC have no inherent characteristics????

In a way, YES......

of course they DO have characteristics....... but the way they are used in, for instance, an "opamp", is a way that is calculated to ELIMINATE THE EFFECT OF THOSE CHARACTERISTICS as much as it is possible to do. This is done with negative feedback to cancel distortions. The manufacturer tries to make the opamp as "ideal" as possible..... to "just have gain", to be as independent of the device type used inside as possible, to erase the "character" of those parts as much as possible.

The feedback used in any opamp circuit takes what is a fairly linear but imperfect circuit and suppresses the non-linearities by a factor of anywhere from 100 to 100,000. The gain of the opamp with feedback resistors may be 10, but the gain with no feedback may be between 1000 and 1 million (it varies with frequency, and is less at high frequencies.)

Meanwhile, a single transistor or tube is commonly used in MI applications with NO feedback, or possibly with a relatively small amount (SVT mid circuit). What it does, it does, honestly and consistently as a result of it's inherent nature. No imitations, no simulations.

Originally Posted by bongomania

You are taking the stance that because solid state components are sometimes used to emulate tube components, that therefore there is some difference between integrated and discreet? That doesn't make sense.

That is an odd argument...... nothing much to do with what I wrote, but I'll comment on it anyway....

There surely is a huge difference in the way they are used, which makes a big difference in the performance. See above about opamps.

As to whether that is "because" ICs are used to simulate or imitate tubes, not really.....

Opamps with various amounts of extra parts "wrapped around them" have indeed been used to "imitate" (simulate) various behaviors of tubes (several patents in the 1980s and 1990s, including one of mine, and several involving fairly huge numbers of parts, by Pritchard and others).

But those are simulations, "imitations" and they lose character outside of certain limits..... while the tube they "simulate" has its inherent set of characteristics no matter HOW you use it...

A tube IS a tube, not an imitation.

Discrete transistors have a character also, the very thing the IC maker tries hard to eliminate...... It is just about as hard to "imitate" a discrete transistor with ICs as it is to imitate a tube.

Not sure how that "doesn't make sense", but you said that, not I, so I will leave it to you to explain what you meant.

Originally Posted by bongomania

First off, what about integrated FETs used in a tubelike circuit? Does the fact that they are in a chip make them "not FETs"?

You will first have to FIND what you describe...... I have no clue what you are talking about.

What is an "integrated FET" as you put it? What do you mean by "used in a tubelike circuit"?

If you are referring to opamps that happen to use FETs as the amplfying devices.......then you need to read over what I wrote about opamps..... they are made and used in ways that ELIMINATE the character of the devices as much as it is possible to do. So in that way yes, they become "not FETs".... the IC maker tries hard to do that.

Whatever character you want the IC opamp (or bipolar transistor opamp) to have you will have to supply by parts added in the feedback. Inherently it is very linear with a very abrupt onset of clipping. About as unlike a discrete device as it is possible for the manufacturer to make it.

Originally Posted by bongomania

And what about the fact that the large majority of transistors are designed to be as linear as possible--whether they are discreet or integrated? Seriously now, you can't be suggesting that a discreet transistor, designed for linearity, is somehow "more like a tube" just because it is not in a chip form?

First, a transistor isn't really "designed for linearity"..... as far as voltage gain, which is what we typically care about in a preamp.

Second, it has nothing to do with "being like a tube", and a lot more to do with "being like itself ".

Transistors (bipolar types) are current-operated devices, not really voltage-operated. They are often designed for consistent current gain over a relatively large range of current. But a bipolar transistor is inherently rather non-linear as far as actual voltage gain. The base can accept only a very small range of voltages relative to the emitter before very serious non-linearities occur.

A tube is, in many cases, designed for linear voltage gain. Both tubes and FETs are voltage operated devices. JFETs are the most like a tube of any solid state device, and if you want to take THAT side path..... I have seen a JFET of a special type that was an exact double for a 12AX7. Even tubes have a limited range of input voltages, although that is typically larger than for a bipolar transistor.

All the above types of discrete device use techniques of local feedback (which is quite different from opamp feedback) to increase the linear input range.

YES.... a discrete transistor is much more of a 'transistor" in its characteristics than a transistor which is used in an opamp.......

Remember, the transistor IS a transistor, not a simulation, while the manufacturer of the opamp tried very hard to make the opamp "ideal" and eliminate the effect of the actual devices used inside..... to REMOVE the effect of there being actual transistors inside as much as possible. And they have gotten VERY good at doing that.

[raphaeldc]

I totally agree with everything you said. That's what I've learned in school!

So everything end up with each one's choice, from the manufacturer an the user, right? The real thing or the "emulation"?

If I want to have an amp with transistors, instead of ICs, I need to look for old amps?

Originally Posted by Jerrold Tiers

I think you are "putting words in my keyboard"......... and missing several points.....

The main one is that IC amplifiers, typically opamps, have the "character" of the parts which must be used "around them" as negative feedback. Their makers try very hard to make that true.

By contrast, discrete transistors (or tubes), used as voltage gain stages having one or two transistors per stage, have a character due to their inherent structure and mode of operation, which the very limited amount of feedback possible in a one or two device circuit does not eliminate. There may be NO feedback at all, in which case it is obvious that "what you see is what you get".

To make the IC opamp act like the discrete device, you must add extra parts to the opamp to "simulate" or "imitate" the action of the device you want to "imitate".

In a way, YES......

of course they DO have characteristics....... but the way they are used in, for instance, an "opamp", is a way that is calculated to ELIMINATE THE EFFECT OF THOSE CHARACTERISTICS as much as it is possible to do. This is done with negative feedback to cancel distortions. The manufacturer tries to make the opamp as "ideal" as possible..... to "just have gain", to be as independent of the device type used inside as possible, to erase the "character" of those parts as much as possible.

The feedback used in any opamp circuit takes what is a fairly linear but imperfect circuit and suppresses the non-linearities by a factor of anywhere from 100 to 100,000. The gain of the opamp with feedback resistors may be 10, but the gain with no feedback may be between 1000 and 1 million (it varies with frequency, and is less at high frequencies.)

Meanwhile, a single transistor or tube is commonly used in MI applications with NO feedback, or possibly with a relatively small amount (SVT mid circuit). What it does, it does, honestly and consistently as a result of it's inherent nature. No imitations, no simulations.

That is an odd argument...... nothing much to do with what I wrote, but I'll comment on it anyway....

There surely is a huge difference in the way they are used, which makes a big difference in the performance. See above about opamps.

As to whether that is "because" ICs are used to simulate or imitate tubes, not really.....

Opamps with various amounts of extra parts "wrapped around them" have indeed been used to "imitate" (simulate) various behaviors of tubes (several patents in the 1980s and 1990s, including one of mine, and several involving fairly huge numbers of parts, by Pritchard and others).

But those are simulations, "imitations" and they lose character outside of certain limits..... while the tube they "simulate" has its inherent set of characteristics no matter HOW you use it...

A tube IS a tube, not an imitation.

Discrete transistors have a character also, the very thing the IC maker tries hard to eliminate...... It is just about as hard to "imitate" a discrete transistor with ICs as it is to imitate a tube.

Not sure how that "doesn't make sense", but you said that, not I, so I will leave it to you to explain what you meant.

You will first have to FIND what you describe...... I have no clue what you are talking about.

What is an "integrated FET" as you put it? What do you mean by "used in a tubelike circuit"?

If you are referring to opamps that happen to use FETs as the amplfying devices.......then you need to read over what I wrote about opamps..... they are made and used in ways that ELIMINATE the character of the devices as much as it is possible to do. So in that way yes, they become "not FETs".... the IC maker tries hard to do that.

Whatever character you want the IC opamp (or bipolar transistor opamp) to have you will have to supply by parts added in the feedback. Inherently it is very linear with a very abrupt onset of clipping. About as unlike a discrete device as it is possible for the manufacturer to make it.

First, a transistor isn't really "designed for linearity"..... as far as voltage gain, which is what we typically care about in a preamp.

Second, it has nothing to do with "being like a tube", and a lot more to do with "being like itself ".

Transistors (bipolar types) are current-operated devices, not really voltage-operated. They are often designed for consistent current gain over a relatively large range of current. But a bipolar transistor is inherently rather non-linear as far as actual voltage gain. The base can accept only a very small range of voltages relative to the emitter before very serious non-linearities occur.

A tube is, in many cases, designed for linear voltage gain. Both tubes and FETs are voltage operated devices. JFETs are the most like a tube of any solid state device, and if you want to take THAT side path..... I have seen a JFET of a special type that was an exact double for a 12AX7. Even tubes have a limited range of input voltages, although that is typically larger than for a bipolar transistor.

All the above types of discrete device use techniques of local feedback (which is quite different from opamp feedback) to increase the linear input range.

YES.... a discrete transistor is much more of a 'transistor" in its characteristics than a transistor which is used in an opamp.......

Remember, the transistor IS a transistor, not a simulation, while the manufacturer of the opamp tried very hard to make the opamp "ideal" and eliminate the effect of the actual devices used inside..... to REMOVE the effect of there being actual transistors inside as much as possible. And they have gotten VERY good at doing that.

[teemuk]

YES.... a discrete transistor is much more of a 'transistor" in its characteristics than a transistor which is used in an opamp.......

That just doesn't make any sense. A discrete transistor is a transistor, an integrated circuit is a circuit. It doesn't really make much sense to compare the two.

If you - on the other hand - compare the transistors within the IC to discrete transistors, or discrete circuit to a similar circuit in integrated form then they are pretty much alike. You might have some particular differences like IC's providing closer tolerance of components within, and better thermal coupling of components within - characteristics gained with costs like, for example, lesser die area to dissipate the heat. But overall those go to very finer design details.

As for the OP's question; the manufacturers use both discrete and IC, as well as "hybrids" such as driver ICs for discrete output devices. It's really more an issue if there's a suitable IC that can be used in the particular design with lesser costs and greater convenience than a discrete design. For example: output power of less than 60W is easier and cheaper to realize with some IC-based solution but due to power dissipation limits of ICs higher power designs may require discrete solutions... or aforementioned hybrid circuitry.

To use a discrete or IC is really just a question of design goals and routines. A discrete circuit can be designed to perform like an integrated one and an integrated one like a discrete one. It's really just what the designer aims to and is convenient with to use. In the end, we talk about circuits not about devices. Device to device components don't make any sense in real applications. An amplifier is never just a transistor, or just a tube or just a FET.