Magura said:Well, to some extend you can tell that you'll have problems with liniarity due to a whopping input capacitance compared to for instance IRFP240.
Magura
The Gate capacitance is Higher and The Total gate charge is somewhat less than IRFP240.....The use of complementary push pull drivers is mandatory for this type of mosfet for proper charging and discharging of gate capacitances....
G.Kleinschmidt said:
Thanks for this data, Glenn. Yes, there are some BJT's out there with very good SOA and also good ft.
Some of the data in the earlier discussion had already convinced me that, just because MOSFETs do not suffer from secondary breakdown, they don't necessarily have a clear advantage. Individual parts really have to be looked at.
Based on destructive measurements I did earlier in the discussion, it also appeared that the IRF SOA spec of constant 300 watts was an overly-conservative hand-waive on their part that did not necessarily reflect real testing. I think my data showed 10 ms transient capabilty of about 7A, but I don't have it in front of me. Of course, in fairness, I didn't do the same destructive testing on bipolars, so they might have a lot more margin than the spec as well.
Also, just because the mechanism in MOSFETs is purely thermal, they can still go pretty fast when they go.
Bob
john curl said:
I think the earlier discussion ended up pretty much as a standoff in regard to MOSFETs vs BJT in the SOA department. My view is that the decision to use one or the other turns on design choices and preferences other than SOA.
The one thing that I have not done, which I would still like to do is to destructively test some of the ring-emitter devices to see if the tested 100V 10 ms SOA exceeds the published spec by as much as the HEXFETs I tested.
Bob
Amazed On Your Opinion
This is not a TRENCH FET.......and nor its a HEXFET!
Its their propietry fabrication process known as POWER MOSV, they claim this process about 2 step above than the HEXFET from IRF....
I am very well aware of the drawbacks associated with TRENCH fets in their Linear working domain in comparision to hexfets....
I have just confirmed this from Microsemi......its not a TRENCH FET[APT is now taken over by Microsemi]
Have you ever tried it for real.....NO...you haven't
Then how did it came in your mind its not "SUITABLE" for audio...just because of its huge gate input capacitance of 10200pF along with absence of its P-channel complement device....just because it doesn't suits your error correction circuitry...which requires and based on complementary devices....
or just because you all here are FANS of complementary devices and have a hard time to digest a NON-complementary amp which could also perform even better in sonics.....
K a n w a r
Bob Cordell said:
This is not a TRENCH FET.......and nor its a HEXFET!
Its their propietry fabrication process known as POWER MOSV, they claim this process about 2 step above than the HEXFET from IRF....
I am very well aware of the drawbacks associated with TRENCH fets in their Linear working domain in comparision to hexfets....
I have just confirmed this from Microsemi......its not a TRENCH FET[APT is now taken over by Microsemi]
Have you ever tried it for real.....NO...you haven't
Then how did it came in your mind its not "SUITABLE" for audio...just because of its huge gate input capacitance of 10200pF along with absence of its P-channel complement device....just because it doesn't suits your error correction circuitry...which requires and based on complementary devices....
or just because you all here are FANS of complementary devices and have a hard time to digest a NON-complementary amp which could also perform even better in sonics.....
K a n w a r
Re: Amazed On Your Opinion
Calm down, Anwar. That was just my opinion based on the fact that it looked like a trench FET. I said absolutely nothing about its lack of a p-type complement. You should keep in mind that the device may have the traits of a trench FET even though they gave it some proprietary name; manufacturers do that all the time.
Please look at its spec sheet and tell us at what current the temperature coefficient of Ids with gate voltage is zero, and for that part tell us the power dissipation and Rds on. That will likely give a clue as to whether it fits in the trench FET category, whether or not they call it a trench FET.
You are right about one thing: I am a fan of complementary class AB output stages. I personally would never try to build an all-N class AB output stage. I stopped doing that in 1970. But, to each his own, the devil is in the details, and the proof is in the pudding. Apparently you have built an amplifier with these devices, so tell us about it.
Cheers,
Bob
Workhorse said:
Calm down, Anwar. That was just my opinion based on the fact that it looked like a trench FET. I said absolutely nothing about its lack of a p-type complement. You should keep in mind that the device may have the traits of a trench FET even though they gave it some proprietary name; manufacturers do that all the time.
Please look at its spec sheet and tell us at what current the temperature coefficient of Ids with gate voltage is zero, and for that part tell us the power dissipation and Rds on. That will likely give a clue as to whether it fits in the trench FET category, whether or not they call it a trench FET.
You are right about one thing: I am a fan of complementary class AB output stages. I personally would never try to build an all-N class AB output stage. I stopped doing that in 1970. But, to each his own, the devil is in the details, and the proof is in the pudding. Apparently you have built an amplifier with these devices, so tell us about it.
Cheers,
Bob
I agree with Bob. My contribution to this thread was that I only wanted bipolar transistors to get a 'fair shake' when it came to consideration in building power amps.
Bob, for example, uses bipolar transistors where I have already designed them out, so he must not be overly critical of bipolars in principle. I have replaced bipolar with mosfets, when I could throughout my amp, however when it comes to a very powerful output stage, bipolar transistors can work pretty well.
Personally, I prefer 'idealized' fets. However, IR does not make 'idealized' fets, and neither does Hitachi. Each process suffers from one thing or another.
However, if I were to make a 50W power amp, I would use only fets as output devices.
Twenty years ago, I tried to make a 100W power amp with paralleled IRF140's and found that I still needed fast short circuit protection, to my dismay.
I ultimately found that bipolar power devices had improved significantly, both in speed and in safe area, so it was possible to make 100-400W (8ohms) power amps by just paralleling more devices in the output stage. I use a complementary power mosfet driver composed of the J201 and its complement, so adding more output pairs does not load the VAS stage in any significant way.
Kanwar's industrial fet could be made to work, but would it really work better than what Bob and I have already done? With a clever circuit, perhaps, but I don't see any real advantage over paralleled output devices, at the moment, and an N channel will not easily retrofit into either Bob's or my designs.
Bob, for example, uses bipolar transistors where I have already designed them out, so he must not be overly critical of bipolars in principle. I have replaced bipolar with mosfets, when I could throughout my amp, however when it comes to a very powerful output stage, bipolar transistors can work pretty well.
Personally, I prefer 'idealized' fets. However, IR does not make 'idealized' fets, and neither does Hitachi. Each process suffers from one thing or another.
However, if I were to make a 50W power amp, I would use only fets as output devices.
Twenty years ago, I tried to make a 100W power amp with paralleled IRF140's and found that I still needed fast short circuit protection, to my dismay.
I ultimately found that bipolar power devices had improved significantly, both in speed and in safe area, so it was possible to make 100-400W (8ohms) power amps by just paralleling more devices in the output stage. I use a complementary power mosfet driver composed of the J201 and its complement, so adding more output pairs does not load the VAS stage in any significant way.
Kanwar's industrial fet could be made to work, but would it really work better than what Bob and I have already done? With a clever circuit, perhaps, but I don't see any real advantage over paralleled output devices, at the moment, and an N channel will not easily retrofit into either Bob's or my designs.
Magura:
I think the Sanken 2SA1216/2SC2922 are popular ring emitter
parts these days. (2SA1215/2SC2921 for lower VCE rating)
mlloyd1
who is looking forward to Bob's test results on ring emitter bipolar outputs
I think the Sanken 2SA1216/2SC2922 are popular ring emitter
parts these days. (2SA1215/2SC2921 for lower VCE rating)
mlloyd1
who is looking forward to Bob's test results on ring emitter bipolar outputs
Magura said:
Ring-emitter devices?? Which part numbers would that be?
Cheers
Magura
john curl said:
John, VERY good point. There are multiple good reasons for paralleling output devices in higher power designs, and this is definitely one of them. In particular, this mitigates the thermal bias stability problem that is brought on by the fact that the junction temperature is usually well above, and not tracking well, the temperature of the heat sink.
I also agree that MOSFET designs really should have some form of fast short-circuit protection in the output stage, since they will conduct HUGE amounts of current under certain conditions (there is no input drive current issue or beta droop issue to protect themselves from this) - look what happens to a screwdriver when it is used to short a reservoir capacitor. And my experience is that fuses are not fast enough unless you have quite a few MOSFETs in parallel. On the other hand, I have found that I can employ very non-invasive fast electronic short circuit protection that kills the gate drive under short-circuit conditions.
Bob
mlloyd1 said:
Me, though i know some of the outcome from (intimate) audio designers who have been using them for a long time and have tested peak output current/power.
Morten,
they're also referred to as RET, or multi-emitters.
Do a search for these terms and my name for more info, brands, type numbers, and pictures of Sanken, Toshiba and Hitachi ring-emitter output devices.
I do hope Mr Curl meant a 50-watter that's nice, and especially Hot.
Re: Re: Amazed On Your Opinion
18 Times at least as far as i could remember Jacco
Bob, my name is Kanwar not Anwar mind it!
APT20M22LVR Zero Temp Coeff is at Id 70A with Vgs=5.25V Pd=520W RDS 0.022Ohms TO-264 Package
IRFP260N [HEXFET] Zero Temp Coeff is at Id 70A with Vgs=6.5V Pd=300W RDS 0.04Ohms TO-247 Package
Now is it Trench or not....acc to you...Tell us...Bob
Heat Dissipation is not a issue for me, because I use Class-H, Class-D , Class-TD for my amps which use these mosfets...Heat dissipation is very less compared to Class-AB..also this Mosfet comes in TO264 Package bigger than TO-247 type used in Hexfet...
Mosfets are directly mounted on heatsink, with no insulator in between them...
Mosfet do require fast short circuit protection but also another important aspect involved is the TYPE of output inductor [outer diameter must be large as well it should be edge wound], which greatly influences the reliability of mosfet during the SC situation..Secondly the clamping of Gate to source voltage during SC near to ZERO volts isnt sufficient, it must be clamped to Neagtive Vgs of about -5V atleast...to ensure fast Turn-Off of mosfet....
Cheers,
Kanwar
jacco vermeulen said:
18 Times at least as far as i could remember Jacco
Bob Cordell said:
Bob, my name is Kanwar not Anwar mind it!
APT20M22LVR Zero Temp Coeff is at Id 70A with Vgs=5.25V Pd=520W RDS 0.022Ohms TO-264 Package
IRFP260N [HEXFET] Zero Temp Coeff is at Id 70A with Vgs=6.5V Pd=300W RDS 0.04Ohms TO-247 Package
Now is it Trench or not....acc to you...Tell us...Bob
Bob Cordell said:
Heat Dissipation is not a issue for me, because I use Class-H, Class-D , Class-TD for my amps which use these mosfets...Heat dissipation is very less compared to Class-AB..also this Mosfet comes in TO264 Package bigger than TO-247 type used in Hexfet...
Mosfets are directly mounted on heatsink, with no insulator in between them...
Mosfet do require fast short circuit protection but also another important aspect involved is the TYPE of output inductor [outer diameter must be large as well it should be edge wound], which greatly influences the reliability of mosfet during the SC situation..Secondly the clamping of Gate to source voltage during SC near to ZERO volts isnt sufficient, it must be clamped to Neagtive Vgs of about -5V atleast...to ensure fast Turn-Off of mosfet....
Cheers,
Kanwar
Another aspect of consideration when paralleling FETs is Current Sharing...
Source resistors donot help them in current sharing at all..
Either use a Single large Die mosfet in High efficiency design
Or drive each individual mosfet in parallel configuration with Voltage Controlled Current Source Driver, which would guarantee proper current sharing [Costly but Robust]
Or Match the mosfets with+-5% tolerance w.r.t. Vgs at a given Id
Kanwar
Source resistors donot help them in current sharing at all..
Either use a Single large Die mosfet in High efficiency design
Or drive each individual mosfet in parallel configuration with Voltage Controlled Current Source Driver, which would guarantee proper current sharing [Costly but Robust]
Or Match the mosfets with+-5% tolerance w.r.t. Vgs at a given Id
Kanwar
Kanwar, your approach is interesting, but is not what is specifically discussed on this thread. It seems to me that it would take 6 of your devices to match the 18 that I use per channel, in my JC-1 power amp design. Because my 18 devices are spread out in a line, it is MUCH EASIER to couple to enough fin area that I don't need a fan, even with several hundred watts of dissipation It is the thermal interface to the heatsink that really makes this part problematic. There is just not enough surface area to get the heat out properly in a standard output stage.
John, Your design approach is somewhat different, I think you want to rely on convection cooling, while i prefer fan cooling...Your amp is in Class-AB or Class-A mode...
While I prefer Tracking supply much better as the heat dissipation is very less...For 1000W continuous I would implement only 1 pair of these mosfets in Class-TD...
While I prefer Tracking supply much better as the heat dissipation is very less...For 1000W continuous I would implement only 1 pair of these mosfets in Class-TD...