oops, give or take one decadeBob Cordell said:
sorry.
38hours without sleep seems to have serious impact on my brain activity. Time to go bed, over and out.
Mzzj, I wish to thank you for "showing that the emperor has no clothes" It is important that the intrinsic problems of mosfets be brought out, that are not necessarily easily notable on the data sheet. The challenge of paralleling mosfets is an important factor, for example, and thermal hot spots is another.
The Gm factor shows why lateral fets have always been found to be more rugged than vertical fets in linear audio applications.
Personally, I have tried and failed to make a 100W all FET power amp about 20 years ago without any active protection circuitry. However, I could and did make 100W power amps with bipolar transistors even earlier, with just passive output protection. I believed the IR and Motorola data sheets! I destroyed many devices in the process.
Charles Hansen has experience with this as well. What about you, Nelson?
The Gm factor shows why lateral fets have always been found to be more rugged than vertical fets in linear audio applications.
Personally, I have tried and failed to make a 100W all FET power amp about 20 years ago without any active protection circuitry. However, I could and did make 100W power amps with bipolar transistors even earlier, with just passive output protection. I believed the IR and Motorola data sheets! I destroyed many devices in the process.
Charles Hansen has experience with this as well. What about you, Nelson?
john curl said:
Hi John,
The emperor is quite well clothed, thank you. Mzzj makes some pretty good points, though, and the references he cites are instructive. However, those references must be read very carefully and put into proper context. I do agree that it is valuable to bring out limitations of whatever devices we use, whether they are BJTs or MOSFETs. This is engineering, and you pick your poison.
One thing that I had not adequately kept up with is some of the developments in newer switchmode MOSFET technology, particularly in regard to so-called Trench FETs. These are quite different from the IRFP240 type planar power MOSFETs we have usually used, and indeed appear to be less suitable for use in linear circuits. That should NOT be mis-interpreted to mean that the planar types like the IRFP240 should not be used in linear circuits. The so-called hot spots are a particular SOA-limiting feature of the Trench FETs, and it is the SOA curves of the Trench FETs that they are warning about in those articles.
Among all these power MOSFETs, one of the things one needs to understand is that they ALL start out with a positive temperature coefficient of drain current at a given gate voltage. BUT, they all eventually have that temperature coefficient eventually go through zero and then negative. The main difference is at what value of drain current they do so. I pointed all of this out in my MOSFET amplifier paper 20+ years ago.
The lower the value of this TC crossover point, the better, and that is what is nice about the laterals; the correlated price you tend to pay is lower gm and higher RdsON. The IRF240 devices have this crossover in the low ampere region, which means you do have to pay more attention to thermal bias stability than with laterals, but still not nearly so much as with bipolars (read my paper for the comparative bias stability analysis). The propensity to hot spots also tends to be connected with higher TC crossover point, but it appears that for the IRF240 it is not a controlling influence on available SOA; the data that I showed in my earlier posts shows clearly that the IR specified SOA limit is quite conservative.
BUT, if you get into these newer trench FETs specifically designed for switching, then the TC crossover current gets into the tens of amperes and becomes a significant issue in limiting SOA at higher voltages. Although it is probably not the same physical phenomenon, it sure begins to look like and behave like secondary breakdown in bipolars. So I certainly do not advocate using these Trench FETs for power amplifier output stages.
I'm still waiting for you to tell us what kind of protection you put in your amplifiers.
The ball is in your court on doing some destructive SOA testing on your favorite ring emitter transistor for comparison to my results. If you want me to do it, just send me a couple of those NPN transistors and I'll be happy to blow them up with 10 ms pulses and report the results. Just send me an email at Bob@cordellaudio.com and I'll supply you with my address. I sincerely think that the results of such a comparative experiment would be valuable to everyone here. I seriously doubt that they will show as much SOA as the IRFP240's that I blew up, which withstood 5 A for 10 ms at 120V with a case temperature of 60C - but I could be wrong - I just care about finding out what the answer is.
C'mon, John. The fact that you blew up a bunch of MOSFETs 20 years ago doesn't prove anything except that you may have been inexperienced in their use. I blew up plenty of them when I first started out also. I must say that I much prefer to deliberately blow them up, however.
I do agree on at least one point: Vertical MOSFET amplifiers in most cases must have active short-circuit protection - at least that has been my experience - and lateral MOSFET amplifiers seem to be able often to get away without it. Although I haven't tried it, I would guess that a substantially over-built vertical MOSFET output stage with numerous devices paralleled might get away with simple passive short circuit protection.
Bob
Bob, the circuitry that we use to protect Parasound amps is the property of Parasound and I won't publish it. However, it works pretty well, since we have many thousand amps out there in the real world. For the record, I generally use fets EVERYWHERE, except the final output. Now, I might even try to do that again, but I know from experience that it can be difficult without the protection that Parasound uses, so what is the point? Is there something really wrong with a powerful (135A) assembly of bipolar transistors that are driven by an open loop source of about 15ohms?
You state that your full power output at 20KHz is very low. How low is it at high frequencies between 1W-10W into a 4 ohm load, 2ohm ? That's where it counts.
I know that my output parts work. Do you know if your 240's really hold up in a more powerful amp? For the record, the very smallest power amp that I am associated with is nominally 100W/channel. It is not the same situation, as what your experience with lower power amps makes you so sure of yourself. Ask Charles Hansen or Nelson Pass. They have even more experience than either you or me in using output fets.
You state that your full power output at 20KHz is very low. How low is it at high frequencies between 1W-10W into a 4 ohm load, 2ohm ? That's where it counts.
I know that my output parts work. Do you know if your 240's really hold up in a more powerful amp? For the record, the very smallest power amp that I am associated with is nominally 100W/channel. It is not the same situation, as what your experience with lower power amps makes you so sure of yourself. Ask Charles Hansen or Nelson Pass. They have even more experience than either you or me in using output fets.
john curl said:
John, I didn't ask you to publish your Parasound protection circuit. I completely respect proprietary matters. I only asked what the nature of it was, as I made clear the first time I posted the question. Is it a fuse? Is it a relay? Is it a crowbar? Is it some kind of VI limiting? That's all.
The MOSFET amp I published in 1984 was just a 50W amp so as to demonstrate the point. Its THD did not rise at lower power levels. Its THD did rise into lower impedance loads, roughly in proprtion to the inverse of the load resistance; so the THD-20 of 0.0006% into 50 watts rose to about 0.0012 at a tad less than 100 watts into 4 0hms. I don't recall testing THD into 2 ohms, but the amp was stable into 1 ohm, and you did see that nice 20 kHz tone burst at 22V peak into 1 ohm.
In 1984 I built a 200 watt per channel version that worked fine.
There is nothing wrong with a 135 Amp assembly of bipolar transistors driven by 15 ohms. You are the one who asserted that MOSFETs don't hold up at high power levels, and I'm just presenting real data (rather than a bunch of BS) that proves you wrong. Those ring emitter transistors are great transistors, but there is no evidence that they hold up better at high power than MOSFETs. Send me some so we can compare real data.
Bob
G’day.
An interesting thread here. Haven’t got much time to delve into this topic much at the moment, but I’d like to quickly introduce myself, without sounding too pompous I hope, by saying that I build amplifiers with power outputs that make even the biggest “concert hall” designs around here look like small fry.
To avoid petty argumentation, before I state my preference, I'd just like clearly state that both types of devices are perfectly suited to the construction of audio power amplifier output stages.
My preference is for bipolar transistors principally for efficiency and cost. Mosfets are intrinsically faster at switching than bipolar devices and with them small gains can be had in the reduction of crossover distortion, particularly higher up in the audio frequency range. However, there are significant drawbacks also. The much lower transconductance of mosfet devices results in significant (for want of a better term) compression distortion, which offsets the gains made in the crossover distortion department to some degree. Then there is the issue of bias current. Mosfet devices can only compete with bipolar transistors in crossover distortion figures when biased at much greater idle currents, typically 5-10 times as much.
Most people don’t realise just how significantly this particular trait can reduce the efficiency and increase the cost of a high power audio amplifier – particularly in a unit with 20 to 30 paralleled pairs of output devices. Contrasted to peak output current into the specified load, the bias current can seem insignificant, but that is not so.
The average current delivered to the load is much less than the peak current when delivering a full amplitude sine wave, and audio power amplifiers intended for amplifying music are seldom called upon to deliver continuous sine waves. When amplifying music, even that with a low dynamic range, the maximum average load current is a great deal smaller, and the bias current of the output devices can quite easily contribute to 50% or more of the amplifiers total power dissipation in a poorly conceived design.
This translates directly to a requirement for greater heat sinking and $$$.
Another problem with Mosfet devices is that they typically cost anywhere from 20-80% more than perforated emitter bipolar devices with comparable dissipation and average current ratings. Having drawn more reactive load lines for high power out put stages than I’ve had hot dinners and having fit them into the temperature derated SOA curves of dozens of Mosfet and Bipolar devices, I can give the assurance that the benefits of freedom from secondary breakdown with Mosfets seldom makes up for the few extra pairs of bipolar devices typically required, when the actual prices of the devices are tallied up.
Bob, you present the IRF240 Mosfet as a rugged audio power transistor and request a similar performing bipolar.
I give you the MJL21193. At the transition into secondary breakdown, it can handle 2.25A at 80V for a period of one second, according to the data sheet. From experience, I can say that that is a conservative specification. The current and voltage figures you quote for the destruction of the IRF240 may be higher, but they’re applied for a duration 100 times shorter than the 1s quoted for the MJL21193. The MJL21193 is also a lot cheaper than the IRF240 and its maximum power dissipation is rated at 200W as opposed to 125W for the IRF240.
Cheers.
Glen.
An interesting thread here. Haven’t got much time to delve into this topic much at the moment, but I’d like to quickly introduce myself, without sounding too pompous I hope, by saying that I build amplifiers with power outputs that make even the biggest “concert hall” designs around here look like small fry.
To avoid petty argumentation, before I state my preference, I'd just like clearly state that both types of devices are perfectly suited to the construction of audio power amplifier output stages.
My preference is for bipolar transistors principally for efficiency and cost. Mosfets are intrinsically faster at switching than bipolar devices and with them small gains can be had in the reduction of crossover distortion, particularly higher up in the audio frequency range. However, there are significant drawbacks also. The much lower transconductance of mosfet devices results in significant (for want of a better term) compression distortion, which offsets the gains made in the crossover distortion department to some degree. Then there is the issue of bias current. Mosfet devices can only compete with bipolar transistors in crossover distortion figures when biased at much greater idle currents, typically 5-10 times as much.
Most people don’t realise just how significantly this particular trait can reduce the efficiency and increase the cost of a high power audio amplifier – particularly in a unit with 20 to 30 paralleled pairs of output devices. Contrasted to peak output current into the specified load, the bias current can seem insignificant, but that is not so.
The average current delivered to the load is much less than the peak current when delivering a full amplitude sine wave, and audio power amplifiers intended for amplifying music are seldom called upon to deliver continuous sine waves. When amplifying music, even that with a low dynamic range, the maximum average load current is a great deal smaller, and the bias current of the output devices can quite easily contribute to 50% or more of the amplifiers total power dissipation in a poorly conceived design.
This translates directly to a requirement for greater heat sinking and $$$.
Another problem with Mosfet devices is that they typically cost anywhere from 20-80% more than perforated emitter bipolar devices with comparable dissipation and average current ratings. Having drawn more reactive load lines for high power out put stages than I’ve had hot dinners and having fit them into the temperature derated SOA curves of dozens of Mosfet and Bipolar devices, I can give the assurance that the benefits of freedom from secondary breakdown with Mosfets seldom makes up for the few extra pairs of bipolar devices typically required, when the actual prices of the devices are tallied up.
Bob, you present the IRF240 Mosfet as a rugged audio power transistor and request a similar performing bipolar.
I give you the MJL21193. At the transition into secondary breakdown, it can handle 2.25A at 80V for a period of one second, according to the data sheet. From experience, I can say that that is a conservative specification. The current and voltage figures you quote for the destruction of the IRF240 may be higher, but they’re applied for a duration 100 times shorter than the 1s quoted for the MJL21193. The MJL21193 is also a lot cheaper than the IRF240 and its maximum power dissipation is rated at 200W as opposed to 125W for the IRF240.
Cheers.
Glen.
G.Kleinschmidt said:
Glen,
I don't have much time right now, but have to say that I think I agree with just about everything you have said, at least within the context you've pointed it out. I especially agree with your point about the nominal bias current for MOSFETs vs bipolar, but will reserve the right to put forth a few caveats that may apply in huge PA amplifiers vs audiophile amplifiers. Two very different markets, of course. I am a bit surprized about your cost comparison.
Best regards,
Bob
Can we hear a difference in sound reproduction when the output are bipolars or mosfets? Assume both are well designed. Do bipolars/mosfets has their own unique sound characteristic?
If they don't, we can go straight to choose which one is more rugged.
If they don't, we can go straight to choose which one is more rugged.
G.Kleinschmidt said:
Glen, let me get back to you on some of your points. Thanks for your detailed post.
First, tell us more about your amplifier. What is its rated power into what load? Is it Class AB, G, or H, or something else? What is its rated THD-1 and THD-20 at full power? Is it specified to deliver 1/3 or 1/8 power continuously into its rated load without overheating? I am familiar with high-power amplifier design, having been involved in the design of the Crest 8001 (bipolar, Class-H).
Your point about the bias current for MOSFETs vs bipolar is well-taken, one of the reasons I designed my MOSFET amplifier with error correction back in 1982. The EC almost eliminates the problem of MOSFET transconductance droop in the crossover region. You can definitely get away with 20-50 mA of bias on a BJT if you're willing to accept only fair sonics. At 150 mA bias, my MOSFET amplifier did 0.02% THD-20 into 8 ohms without EC.
The Public Address, Pro Audio and Audiophile markets are very different. In the latter, it is quite common for high-end amplifiers to operate in Class AAB, with output devices somewhat over-biased, for best sonics. There is less concern about idle dissipation in the audiophile arena. At the same time, there is very much more concern for high-frequency linearity.
Please define better what you mean by compression distortion. I don't believe my MOSFET amplifier suffers from it, be it from reduced transconductance or not. At high currents, gm of MOSFETs is plenty.
The MJL21193 transistors you mention are very slow devices by today's standards, with an ft of only 4 MHz. That is adequate for PA applications, but certainly not for high-end audio.
They are also not less expensive than IRFP240s. I don't know where you got your numbers, but at DigiKey, the MJL21193 is $3.78 in QTY 100, while the IRFP240 is only $1.71.
It's unfortunate that the MJL21193 is not rated for SOA at 10 ms, which is what is the more important number for audio.
Cheers,
Bob
Hi, Bob Cordell,
Bob, why in Crest 8002, the CCS that powering the input differential is replaced by transistor CCS (in 8001 it is only R powered from 27V, if I'm not mistaken)?
I don't measure anything, but simple R seems to sound better?
So, you're the one who design this amp? It is one of industry's standard.
Bob, why in Crest 8002, the CCS that powering the input differential is replaced by transistor CCS (in 8001 it is only R powered from 27V, if I'm not mistaken)?
I don't measure anything, but simple R seems to sound better?
First, tell us more about your amplifier. What is its rated power into what load? Is it Class AB, G, or H, or something else? What is its rated THD-1 and THD-20 at full power? Is it specified to deliver 1/3 or 1/8 power continuously into its rated load without overheating? I am familiar with high-power amplifier design, having been involved in the design of the Crest 8001 (bipolar, Class-H).
I’ve designed and built various bipolar A, AB designs from 500-5000W with THD levels well below perceptible limits. These are not mass produced units, but mostly one-off designs either for my own use or for specific applications.
Your point about the bias current for MOSFETs vs bipolar is well-taken, one of the reasons I designed my MOSFET amplifier with error correction back in 1982. The EC almost eliminates the problem of MOSFET transconductance droop in the crossover region. You can definitely get away with 20-50 mA of bias on a BJT if you're willing to accept only fair sonics. At 150 mA bias, my MOSFET amplifier did 0.02% THD-20 into 8 ohms without EC.
I’m sorry, but I don’t agree with your assertion that a BJT amplifier biased to 20-50mA per device will only deliver “fair sonics”. That’s just plain absurd. The THD figure you quote isn’t particularly difficult to achieve in non-EC amplifiers with bipolar output stages and ultra low THD figures at 20kHz are superfluous anyway. If you can hear the second or third harmonic of 20kHz then you’re doing better than me.
The Public Address, Pro Audio and Audiophile markets are very different. In the latter, it is quite common for high-end amplifiers to operate in Class AAB, with output devices somewhat over-biased, for best sonics. There is less concern about idle dissipation in the audiophile arena. At the same time, there is very much more concern for high-frequency linearity.
OK, but I don’t design for the esoteric audiophile market. I’m interested in producing efficient cost effective designs. Heat dissipation must be kept to a minimum and in that regard BJT’s with their much lower bias current requirements and ability to run at high temperatures without throttling back rule supreme.
Please define better what you mean by compression distortion. I don't believe my MOSFET amplifier suffers from it, be it from reduced transconductance or not. At high currents, gm of MOSFETs is plenty.
I didn’t say that the gm of Mosfets wasn’t plenty for linear amplification, I was making a comparison to BJT’s which have a much higher transconductance and therefore suffer from a particular distortion mechanism to a lesser degree than Mosfets devices. In source follower configurations the instantaneous output voltage is equal to the voltage at the gate minus the gate to source voltage, which for some devices can become significantly non linear at higher currents.
The MJL21193 transistors you mention are very slow devices by today's standards, with an ft of only 4 MHz. That is adequate for PA applications, but certainly not for high-end audio.
I didn’t cite the MJL21193 as the best audio transistor out there, but as a unit with comparable robustness, as that is what you had asked for.
They are also not less expensive than IRFP240s. I don't know where you got your numbers, but at DigiKey, the MJL21193 is $3.78 in QTY 100, while the IRFP240 is only $1.71.
Things are obviously better where you are. I was comparing it to the TO3 IRF240, which from an Australian supplier costs approximately 50% more than both the MJ21193 and MJL21193. I didn’t get a price on the IRFP240.
It's unfortunate that the MJL21193 is not rated for SOA at 10 ms, which is what is the more important number for audio.
The data sheets only provide limited information. That does not mean the MJL21193 will not withstand much high dissipation levels for brief 10mS bursts. I haven’t subjected one to destructive testing, but going by it’s robust 1 second rating I doubt that it would be much less able, if at all, than the IRF240 device.
Cheers,
Glen
I’ve designed and built various bipolar A, AB designs from 500-5000W with THD levels well below perceptible limits. These are not mass produced units, but mostly one-off designs either for my own use or for specific applications.
Your point about the bias current for MOSFETs vs bipolar is well-taken, one of the reasons I designed my MOSFET amplifier with error correction back in 1982. The EC almost eliminates the problem of MOSFET transconductance droop in the crossover region. You can definitely get away with 20-50 mA of bias on a BJT if you're willing to accept only fair sonics. At 150 mA bias, my MOSFET amplifier did 0.02% THD-20 into 8 ohms without EC.
I’m sorry, but I don’t agree with your assertion that a BJT amplifier biased to 20-50mA per device will only deliver “fair sonics”. That’s just plain absurd. The THD figure you quote isn’t particularly difficult to achieve in non-EC amplifiers with bipolar output stages and ultra low THD figures at 20kHz are superfluous anyway. If you can hear the second or third harmonic of 20kHz then you’re doing better than me.
The Public Address, Pro Audio and Audiophile markets are very different. In the latter, it is quite common for high-end amplifiers to operate in Class AAB, with output devices somewhat over-biased, for best sonics. There is less concern about idle dissipation in the audiophile arena. At the same time, there is very much more concern for high-frequency linearity.
OK, but I don’t design for the esoteric audiophile market. I’m interested in producing efficient cost effective designs. Heat dissipation must be kept to a minimum and in that regard BJT’s with their much lower bias current requirements and ability to run at high temperatures without throttling back rule supreme.
Please define better what you mean by compression distortion. I don't believe my MOSFET amplifier suffers from it, be it from reduced transconductance or not. At high currents, gm of MOSFETs is plenty.
I didn’t say that the gm of Mosfets wasn’t plenty for linear amplification, I was making a comparison to BJT’s which have a much higher transconductance and therefore suffer from a particular distortion mechanism to a lesser degree than Mosfets devices. In source follower configurations the instantaneous output voltage is equal to the voltage at the gate minus the gate to source voltage, which for some devices can become significantly non linear at higher currents.
The MJL21193 transistors you mention are very slow devices by today's standards, with an ft of only 4 MHz. That is adequate for PA applications, but certainly not for high-end audio.
I didn’t cite the MJL21193 as the best audio transistor out there, but as a unit with comparable robustness, as that is what you had asked for.
They are also not less expensive than IRFP240s. I don't know where you got your numbers, but at DigiKey, the MJL21193 is $3.78 in QTY 100, while the IRFP240 is only $1.71.
Things are obviously better where you are. I was comparing it to the TO3 IRF240, which from an Australian supplier costs approximately 50% more than both the MJ21193 and MJL21193. I didn’t get a price on the IRFP240.
It's unfortunate that the MJL21193 is not rated for SOA at 10 ms, which is what is the more important number for audio.
The data sheets only provide limited information. That does not mean the MJL21193 will not withstand much high dissipation levels for brief 10mS bursts. I haven’t subjected one to destructive testing, but going by it’s robust 1 second rating I doubt that it would be much less able, if at all, than the IRF240 device.
Cheers,
Glen
I really hate it if a control loop switches direction somewhere in the middle..
regards, Gerhard
At 150 mA bias, my MOSFET amplifier did 0.02% THD-20 into 8 ohms without EC.
Bob, Just to further elaborate on the results that can be achieved with bipolar output stages, I provide a link to a rather basic, albeit refined Douglas Self design that does not use error correction and is of comparable power specification to your MOSFET design:
http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm
It produces less than 0.02% THD at 40kHz delivering 40W into 8 ohms.
I have never published my designs, but I have achieved similar results using paralleled MJL21193/MJL21194 complementary pairs at 15 times this power figure - devices which you assert are not suitable for high-end audio.
Cheers,
Glen
Bob, Just to further elaborate on the results that can be achieved with bipolar output stages, I provide a link to a rather basic, albeit refined Douglas Self design that does not use error correction and is of comparable power specification to your MOSFET design:
http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm
It produces less than 0.02% THD at 40kHz delivering 40W into 8 ohms.
I have never published my designs, but I have achieved similar results using paralleled MJL21193/MJL21194 complementary pairs at 15 times this power figure - devices which you assert are not suitable for high-end audio.
Cheers,
Glen
Wavebourn:
I can hear intermodulation between 20 KHz and 20100Hz well.
OK, but how well can you hear the intermodulation products between the 2nd harmonic of 20kHz and 20100Hz? 🙂
I can hear intermodulation between 20 KHz and 20100Hz well.
OK, but how well can you hear the intermodulation products between the 2nd harmonic of 20kHz and 20100Hz? 🙂
G.Kleinschmidt said:
Do you mean how I can hear products with 100 and 200 Hz frequencies? 😀
I don't know. 😉
Also, I can distinguish natural sibiliant consonants from distorted.
Wavebourn:
Also, I can distinguish natural sibiliant consonants from distorted.
Man, I don't even know what sibiliant consonants are, but that sounds impressive. I can identify between the 400Hz and 1kHz test tones of my RF amp test rig without looking at the frequency switch.....usually.
Am I similarly gifted?
😕
Also, I can distinguish natural sibiliant consonants from distorted.
Man, I don't even know what sibiliant consonants are, but that sounds impressive. I can identify between the 400Hz and 1kHz test tones of my RF amp test rig without looking at the frequency switch.....usually.
Am I similarly gifted?
😕G.Kleinschmidt said:
Another thing which is worth noting is that in this design D.Self used the now rather old hat Bipolar MJ802/MJ4502 complementary pair - transistors which only have an fT of 2MHz - half that of the MJL21193/MJL21194 complementary pairs 4MHz fT. The MJ802/MJ4502 transistors also have quite inferior hfe Vs collector current linearity.
In my designs using the MJL21193/MJL21194 pairs I run larger emitter resistors and a bias current of 40-50mA.
Cheers,
Glen
Hi, Bob Cordell,
What is your opinion about quasi complementary output stage (that uses all NPN devices)?
It can be seen that the positive part is common collector, while the negative part is common emitor usage mode of output transistors, so it is not balanced between positive and negative half.
What is the merit and drawback of this quasi complementary output stage?
What is your opinion about quasi complementary output stage (that uses all NPN devices)?
It can be seen that the positive part is common collector, while the negative part is common emitor usage mode of output transistors, so it is not balanced between positive and negative half.
What is the merit and drawback of this quasi complementary output stage?