What are the pros and cons of each in output stage in car audio amps, why almost all modern amps use FET's?
I just fixed one more PG Titanium amp with Darlington's outputs, and need to admit, that I haven't heard any amp with FET's outputs (highs-mids) that sounds as beautiful for me as these series. I know, that's probably personal preference, but somehow Darlington's sounds much better then FET's for me 😕 They also seems to be far less efficient - runs hotter at same load
I just fixed one more PG Titanium amp with Darlington's outputs, and need to admit, that I haven't heard any amp with FET's outputs (highs-mids) that sounds as beautiful for me as these series. I know, that's probably personal preference, but somehow Darlington's sounds much better then FET's for me 😕 They also seems to be far less efficient - runs hotter at same load
It interesting that you can tell the difference between them. I talked to my EE professor the other day and he told me that even if there is a difference non of human beings on Earth can sense it.
He also argued that solid state and tube amps sound indistinguishable to a human ear
He also argued that solid state and tube amps sound indistinguishable to a human ear
I generally hate to weigh in on things that are subjective because it usually ends in pushing and shoving and even name-calling but here's my 2 cents.
In general, most solid state amplifiers built for the masses have a significant amount of negative feedback. This insures that the output is simply an amplified version of the input signal. Some of these amps may have less feedback on the high frequency end (for stability against oscillation with highly reactive loads) which may make them sound a 'tiny' bit different than others but the difference is typically inaudible. In my opinion, an amplifier should amplify and not change the sound (not even a little). It's an amplifier not a signal processor.
Even if the amplifier section does nothing but amplify, the crossovers, equalizers and other built-in processors can change the way the amp sounds. This could account for the difference some people hear.
Tube amps are significantly different. Feedback is used much less. In those that have little or no feedback, the tubes and the output transformer can make a huge difference in the way an amplifier sounds. To complicate matters even more, the variations in the speaker's impedance curve can change the frequency response of the amp (due to low damping factors). Depending on the amp/speaker combination, one amp may sound great with speaker A and lousy with speaker B even though speaker B sounds great on another amplifier. Tube amps can 'sound' wonderful but I've never had much interest in them because I am bit of a purist.
Let the arguing begin. Hopefully it won't get too bloody.
In general, most solid state amplifiers built for the masses have a significant amount of negative feedback. This insures that the output is simply an amplified version of the input signal. Some of these amps may have less feedback on the high frequency end (for stability against oscillation with highly reactive loads) which may make them sound a 'tiny' bit different than others but the difference is typically inaudible. In my opinion, an amplifier should amplify and not change the sound (not even a little). It's an amplifier not a signal processor.
Even if the amplifier section does nothing but amplify, the crossovers, equalizers and other built-in processors can change the way the amp sounds. This could account for the difference some people hear.
Tube amps are significantly different. Feedback is used much less. In those that have little or no feedback, the tubes and the output transformer can make a huge difference in the way an amplifier sounds. To complicate matters even more, the variations in the speaker's impedance curve can change the frequency response of the amp (due to low damping factors). Depending on the amp/speaker combination, one amp may sound great with speaker A and lousy with speaker B even though speaker B sounds great on another amplifier. Tube amps can 'sound' wonderful but I've never had much interest in them because I am bit of a purist.
Let the arguing begin. Hopefully it won't get too bloody.
The professor was arguing that it is impossible to tell the difference because as soon as you connect an amplifier to a REAL speaker, which will make additional 0.5-1% THD (don’t quote me, I forgot the exact numbers) there is no way you can tell the difference. 
For a typical distortion test where you'd drive a pure 1kHz test tone into the amp, the THD may increase very slightly (i.e. 0.045% to 0.050%) when you go from no load to 4 ohms but it shouldn't increase much more than that. Of course, the actual increase in distortion depends on the overall design of the amplifier. An increase to over 1% without driving the amp into clipping would be a sign of a poor design (for an amp using significant negative feedback).
You should also realize that the distortion changes with frequency. Typically, distortion will increase at a higher rate at the extremes of the audio spectrum.
You should also realize that the distortion changes with frequency. Typically, distortion will increase at a higher rate at the extremes of the audio spectrum.
I think the professor said that If you measure the distortion of an amplifier alone you have the IDEAL picture that CAN show a slight advantage of one amplifier over the other. But REAL speakers create a lot of distortion (due to non-linear rubber resistance, momentum of the cone, ect..) so the distortion of very good amplifier circuitry becomes negligibly small and inaudible comparing to the distortion produced by the speaker itself.
It is like comparing two rifles. Of source, you can run a few test shots in vacuum to determine that one rifle is 1% more precise, but if you go outside, where there is REAL wind resistance that creates 30% more uncertainty in the bullet's trajectory, both rifles will perform indistinguishably poor.
His point was that it is possible to make solid state amplifiers with VERY small THD but there is no point in pursuing this goal because you can't make very good speaker. 🙂
It is like comparing two rifles. Of source, you can run a few test shots in vacuum to determine that one rifle is 1% more precise, but if you go outside, where there is REAL wind resistance that creates 30% more uncertainty in the bullet's trajectory, both rifles will perform indistinguishably poor.
His point was that it is possible to make solid state amplifiers with VERY small THD but there is no point in pursuing this goal because you can't make very good speaker. 🙂
He's correct. Speakers are most definitely the weak link in any audio system. Even the best speakers would be hard pressed to achieve less than 1% THD across the entire spectrum.
I absolutely agree, that probably major audible difference in sound coming from built-in sound pre-processing pre-amp circuitry. The combination of amp and speakers is also extremely critical. But still, why the most expensive amps today uses bipolar outputs instead FET's? Marketing gimmick? More reliable? More control over output stage? Higher quality of available bipolar parts? They looks good through transparent plexiglass 😀 ?
As for the Darlington vs MosFet battle, there are lots of reasons why builder A uses BJT's and Builder B uses MosFets.
1: they have opposite thermal charcteristics. BJT's exhibit thermal runaway when heated and if left uncontrolled they will self destruct so it requires extra circuitry to control bias thermally so as to prevent this. Where as MosFets act just the opposite by not conducting as much hot as they do cold. So they tend to 'backoff' so to speak thus preventing thermal failure via runaway. So a builder can save time and money and parts count by taking this into account and using it to his designs advantage. (I.E. Fets are cheaper to build and drive correctly).
2: Darlingtons have been used in amplifiers for years as a solution to a problem also. Since they are actually two BJT's in the same package they thermal track very well and offer hugh gains and high current capabilities in the final stage. So here again they are easy to drive except for the thermal character of heat induced runaway. More modern Darlingtons and standard BJT's offer a built in diode on the same die so as to offer the very best possible thermal tracking possible (please see ON Semi's transistor lineup for more details).
Darlingtons offer a lower parts count and very sturdy BJT for output usage. PG uses a triple darlington design. Thus simplifying design and parts count while allowing for a complex design to be achieved. All darlington stage offer high current gains and high thermal stability due to the common packaging of the dual transistor. This can affect the overall use of feedback by lowering its need to get the amp to work properly. Also high frequency devices also aid in this design, by High I mean 25 mega-hertz or higher rated semi devices just to produce 20 to 20 kilo-hertz signals accurately to the speaker load. Like PG uses.
There are lots of folks out there that believe that negative feedback causes distortion to the signal. Thats why there are low and no feedback amplifier designs out in the world. feedback is for a lack of a better word a band-aid that is used to counter poor operation characteristics of many solid state amplifiers. Darlingtons can be used to lower the use of feedback, and thus improve the sound character of a amplifier. (Please goggle low TIM amplifier for some of the free info about Transient Intermodulation Distortion which is increasd by large negative feedback in Solid state amplifiers). This should take you to Georgia Tech's Dr. leach and his very old but much loved and copied amp design ). You can also search this forum about this subject also, as it is a goldmine of info on this subject.
As for the sonic differences in amplifiers:
Now this may be due to to many reasons from design of the amp to the way the amp loads into the speaker. But one thing for sure there are thousands of different amplifiers for sale on the market and most of then boast of sounding better that the other via design differences and construction methods.
I chose not to counter the mass's about this sonic part of the business. It would run head long against a billion dollar industry of amplifier design.🙂
1: they have opposite thermal charcteristics. BJT's exhibit thermal runaway when heated and if left uncontrolled they will self destruct so it requires extra circuitry to control bias thermally so as to prevent this. Where as MosFets act just the opposite by not conducting as much hot as they do cold. So they tend to 'backoff' so to speak thus preventing thermal failure via runaway. So a builder can save time and money and parts count by taking this into account and using it to his designs advantage. (I.E. Fets are cheaper to build and drive correctly).
2: Darlingtons have been used in amplifiers for years as a solution to a problem also. Since they are actually two BJT's in the same package they thermal track very well and offer hugh gains and high current capabilities in the final stage. So here again they are easy to drive except for the thermal character of heat induced runaway. More modern Darlingtons and standard BJT's offer a built in diode on the same die so as to offer the very best possible thermal tracking possible (please see ON Semi's transistor lineup for more details).
Darlingtons offer a lower parts count and very sturdy BJT for output usage. PG uses a triple darlington design. Thus simplifying design and parts count while allowing for a complex design to be achieved. All darlington stage offer high current gains and high thermal stability due to the common packaging of the dual transistor. This can affect the overall use of feedback by lowering its need to get the amp to work properly. Also high frequency devices also aid in this design, by High I mean 25 mega-hertz or higher rated semi devices just to produce 20 to 20 kilo-hertz signals accurately to the speaker load. Like PG uses.
There are lots of folks out there that believe that negative feedback causes distortion to the signal. Thats why there are low and no feedback amplifier designs out in the world. feedback is for a lack of a better word a band-aid that is used to counter poor operation characteristics of many solid state amplifiers. Darlingtons can be used to lower the use of feedback, and thus improve the sound character of a amplifier. (Please goggle low TIM amplifier for some of the free info about Transient Intermodulation Distortion which is increasd by large negative feedback in Solid state amplifiers). This should take you to Georgia Tech's Dr. leach and his very old but much loved and copied amp design ). You can also search this forum about this subject also, as it is a goldmine of info on this subject.
As for the sonic differences in amplifiers:
Now this may be due to to many reasons from design of the amp to the way the amp loads into the speaker. But one thing for sure there are thousands of different amplifiers for sale on the market and most of then boast of sounding better that the other via design differences and construction methods.
I chose not to counter the mass's about this sonic part of the business. It would run head long against a billion dollar industry of amplifier design.🙂
No matter if an amp uses FETs or BJTs, if it's a class AB amplifier, it WILL have to have bias compensation.
It's been stated (wrongly) many times that MOSFETs will not go into thermal avalanche. If the amp is a class AB where there is a small bias current flowing at idle, the amp can and will go into thermal avalanche without compensation.
To make sure that I was not wrong, I checked this on a Rockford 360.2. I set the voltage across the source resistors to 0.003 volts. The sink temp was ~70F. The Vgs was 3.031 volts. I ran the amp until it was 120F. I rechecked the voltage across the source resistor and it was 0.003/0.004 (essentially no difference in idle current). I remeasured the Vgs and it was at 2.872. The bias compensation compensated for the difference in the threshold voltage vs the temp.
I also tried it with a simple circuit. I mounted an IRFZ48 to a small sink. I connected a resistor in series between the drain and the power supply +B. I set the bias current to 0.006 volts across a 0.25 ohm resistor. I heated the sink with a heat gun and the idle current went to 0.088 volts. Both of these example show that the FETs don't 'back off' at high temps.
It is true that the FETs will exhibit more RDSon at high temps but this only applies when the FET is in saturation and the only time this applies in a class AB amp is when the amp is clipping.
If you see an amp that operates in linear mode that doesn't use bias compensation, it's very likely a class B amp.
Can you give model numbers of the PG amps that use actual darlington transistors? I've seen amps that use 2SC3281s and 2SC5200s with MJF15030s and 2SC5248s for drivers but none of these are darlingtons in a single package. They use darlington circuits but, as far as I know, they don't widely use darlington transistors. Ask the guys on the PG forum if they can provide this information.
Feedback is not a band-aid. All electronic components have loss/inefficiency. To build a low distortion amp with no feedback takes a massive amount of components. The amp will be expensive and very large. This is particularly true for amps that have to drive low impedance loads. Global feedback is almost entirely responsible for low output impedance on most all commercially availble audio amplifiers. The Leach amp uses global negative feedback.
As for BJT vs FET for outputs, global feeback will correct for any differences. Global feedback is so good that you could insert a 1 ohm resistor in series with one transistor of a complementary pair and the output would remain symmetrical and essentially distortion free until you ran the amp to a level where it ran out of rail voltage.
This post is not here to start an argument. If you believe I'm wrong please speak up but also provide specifics. I hate misinformation and if I'm wrong I want to be corrected but I don't want third hand information or hearsay.
It's been stated (wrongly) many times that MOSFETs will not go into thermal avalanche. If the amp is a class AB where there is a small bias current flowing at idle, the amp can and will go into thermal avalanche without compensation.
To make sure that I was not wrong, I checked this on a Rockford 360.2. I set the voltage across the source resistors to 0.003 volts. The sink temp was ~70F. The Vgs was 3.031 volts. I ran the amp until it was 120F. I rechecked the voltage across the source resistor and it was 0.003/0.004 (essentially no difference in idle current). I remeasured the Vgs and it was at 2.872. The bias compensation compensated for the difference in the threshold voltage vs the temp.
I also tried it with a simple circuit. I mounted an IRFZ48 to a small sink. I connected a resistor in series between the drain and the power supply +B. I set the bias current to 0.006 volts across a 0.25 ohm resistor. I heated the sink with a heat gun and the idle current went to 0.088 volts. Both of these example show that the FETs don't 'back off' at high temps.
It is true that the FETs will exhibit more RDSon at high temps but this only applies when the FET is in saturation and the only time this applies in a class AB amp is when the amp is clipping.
If you see an amp that operates in linear mode that doesn't use bias compensation, it's very likely a class B amp.
Can you give model numbers of the PG amps that use actual darlington transistors? I've seen amps that use 2SC3281s and 2SC5200s with MJF15030s and 2SC5248s for drivers but none of these are darlingtons in a single package. They use darlington circuits but, as far as I know, they don't widely use darlington transistors. Ask the guys on the PG forum if they can provide this information.
Feedback is not a band-aid. All electronic components have loss/inefficiency. To build a low distortion amp with no feedback takes a massive amount of components. The amp will be expensive and very large. This is particularly true for amps that have to drive low impedance loads. Global feedback is almost entirely responsible for low output impedance on most all commercially availble audio amplifiers. The Leach amp uses global negative feedback.
As for BJT vs FET for outputs, global feeback will correct for any differences. Global feedback is so good that you could insert a 1 ohm resistor in series with one transistor of a complementary pair and the output would remain symmetrical and essentially distortion free until you ran the amp to a level where it ran out of rail voltage.
This post is not here to start an argument. If you believe I'm wrong please speak up but also provide specifics. I hate misinformation and if I'm wrong I want to be corrected but I don't want third hand information or hearsay.
There is a lot of good info in this thread 
I've seen uncontrolled thermal avalanche of MOSFETS several times due to faulty compensation circuit, with 4-ch MTX and JL 500/5, last one still don't want to compensate properly, when I heat it with hairdryer 🙂 bias just skyrockets...
Yes, Perry, I remeber 2SC5200's under heat sink, and it seems to be after googling for a while they are "Triple Diffuse Type", and not really darlingtons, as I incorrectly assumed. My assumption was made on fact, that I heard they used darlingtons in their design (probably it was earlier, then these Ti-series).
I've seen uncontrolled thermal avalanche of MOSFETS several times due to faulty compensation circuit, with 4-ch MTX and JL 500/5, last one still don't want to compensate properly, when I heat it with hairdryer 🙂 bias just skyrockets...
Yes, Perry, I remeber 2SC5200's under heat sink, and it seems to be after googling for a while they are "Triple Diffuse Type", and not really darlingtons, as I incorrectly assumed. My assumption was made on fact, that I heard they used darlingtons in their design (probably it was earlier, then these Ti-series).
Hi Perry.
Well I don't specifically remember ever stating that a Mosfet would not die from heat. I do remember trying to explain the differences in the overall design concepts that must be taken into account of if your trying to decide which type of amplifier to build. I.E. positive vs negative temp coefficients and the resulting circuitry and its complexity, and cost.
And because I don't know the level of the person asking the question, perhaps I may have been much too gereral for a past master like yourself.
PG states they use a "Triple Darlington output" in their older MS, MPS, M, ZPA and Xenon series amps. I own these amps. I also have several of the Zx and Ti amps. And they use the 1943 and 5200 type outputs, Please see the attached link to all the PG amplifier Pdf files:
ftp://208.187.38.55/Phoenix_Gold/Manuals/Amplifiers/
Any further questions should be answered by their product pdf files. I think their products stand for themselves, and I hope this link will help our poster to answer any of his questions.
I do appreciate your attempt to clarify anything you did not feel was clear and easily understood in my post. I always appreciate help from any source. I hope my re-post has cleared up any misinterpretations of my statements and their basic at best attempt to explain some of the more serious issues that must be dealt with in designing a amplifier of either BJT or MosFet designs.
My attempt was to shead light on a few of the many things that weight in on a engineers mind when trying to decide which design direction in which to go.
About the only thing I felt that I left out was the cost difference in the driver stages. Mosfets have drastically lower drive requirements so they cost way less to drive circuitry wise in the previous stages leading up to the output devices, where as BJT's require higher cost of parts to drive the output stage.
As for valterdaw's comments about his BJT amp sounding better than mosfet amps, Well he's not alone in those feelings. But these personal feelings are always a source of tension in the audio world.
And even though they are both Solid State amplifiers, there is a world of difference inside, and this can and does have impact on SQ IMO.
Well I don't specifically remember ever stating that a Mosfet would not die from heat. I do remember trying to explain the differences in the overall design concepts that must be taken into account of if your trying to decide which type of amplifier to build. I.E. positive vs negative temp coefficients and the resulting circuitry and its complexity, and cost.
And because I don't know the level of the person asking the question, perhaps I may have been much too gereral for a past master like yourself.
PG states they use a "Triple Darlington output" in their older MS, MPS, M, ZPA and Xenon series amps. I own these amps. I also have several of the Zx and Ti amps. And they use the 1943 and 5200 type outputs, Please see the attached link to all the PG amplifier Pdf files:
ftp://208.187.38.55/Phoenix_Gold/Manuals/Amplifiers/
Any further questions should be answered by their product pdf files. I think their products stand for themselves, and I hope this link will help our poster to answer any of his questions.
I do appreciate your attempt to clarify anything you did not feel was clear and easily understood in my post. I always appreciate help from any source. I hope my re-post has cleared up any misinterpretations of my statements and their basic at best attempt to explain some of the more serious issues that must be dealt with in designing a amplifier of either BJT or MosFet designs.
My attempt was to shead light on a few of the many things that weight in on a engineers mind when trying to decide which design direction in which to go.
About the only thing I felt that I left out was the cost difference in the driver stages. Mosfets have drastically lower drive requirements so they cost way less to drive circuitry wise in the previous stages leading up to the output devices, where as BJT's require higher cost of parts to drive the output stage.
As for valterdaw's comments about his BJT amp sounding better than mosfet amps, Well he's not alone in those feelings. But these personal feelings are always a source of tension in the audio world.
And even though they are both Solid State amplifiers, there is a world of difference inside, and this can and does have impact on SQ IMO.
There is a great, but rather lengthly thread on this issue in the Solid State forum that may best explain the differences, pros and cons, just may take some time to read.🙂
http://www.diyaudio.com/forums/showthread.php?threadid=101745
http://www.diyaudio.com/forums/showthread.php?threadid=101745
- Status
- Not open for further replies.