amplifierguru said:
hi guru ,
The complementary symmetry is well achieved in bipolars and lateral mosfets , but when you talk about switching vertical mosfets they have no complement with their p-channel counterparts, in their case the p-channel switching mosfet exhibit triple the gate capacitance++ 4 times the Rds ++ slower turn ON & turn OFF as well.
Thats why pure n-channel Vertical mosfets serves even better than best complementary bipolars, because in bipolars the there is a difference in collector and emmiter current [Ie=Ic+Ib]whereas there is no difference in drain and source current of mosfets which in conduction mode[Id=Is].
Secondly n-channel mosfets are available in high voltage++high power density++high current version ++ no second breakdown voltage ++ extremely large SOA for free.
hope you understand it.
regards,
Kanwar
I would agree in general... assuming you use a phase splitter drive, the capacitance seen by the driver is the same for both the top and bottom FET, which is certainly a positive thing. You do get different operating temperatures for the driver transistors (unless the bottom end driver is cascoded to the output, though you lose one Ib that way on it's end), but assuming proper heatsinking, and driver choice (miller effect - one side sees it, the other not) it is not a problem in practice.
As for complementary V-MOS - in theory, the situation is similar to complementary L-MOS. The real difference is not in the ability to produce a reasonable complement, but in the will to do so. P and N ch L-MOSFETs are internally very different as intrinsic differences in carrier movability demand it, just the same as they do for V-MOS. That being said, it is true that the transistors LABELED as complements (eg. IRFP240 and IRFP9240, i.e. XXX and 9XXX) are NOT complements. Not even close. Gm on P parts can be as low as half that of the N parts, and the distribution of capacitances can be very different. HOWEVER - if one looks carefully through the data, it is possible to find very reasonable complements, with major parameter differences between P and N parts lower than the standard spread of parameters amongst parts of the same kind. So far I've managed to match only 6 pairs. Within those, some parameters are however radically different (for instance Vds max) and it is the P type that is usually the lower specced. This is a small price to pay for more important parameters such as maximum current, gm, capacitance, and power dissipation being close matches.
As for complementary V-MOS - in theory, the situation is similar to complementary L-MOS. The real difference is not in the ability to produce a reasonable complement, but in the will to do so. P and N ch L-MOSFETs are internally very different as intrinsic differences in carrier movability demand it, just the same as they do for V-MOS. That being said, it is true that the transistors LABELED as complements (eg. IRFP240 and IRFP9240, i.e. XXX and 9XXX) are NOT complements. Not even close. Gm on P parts can be as low as half that of the N parts, and the distribution of capacitances can be very different. HOWEVER - if one looks carefully through the data, it is possible to find very reasonable complements, with major parameter differences between P and N parts lower than the standard spread of parameters amongst parts of the same kind. So far I've managed to match only 6 pairs. Within those, some parameters are however radically different (for instance Vds max) and it is the P type that is usually the lower specced. This is a small price to pay for more important parameters such as maximum current, gm, capacitance, and power dissipation being close matches.
ilimzn said:
The main thing is the price to pay for p-channel is usually 2-3 times more than its sbstitute n-channel and secondly you cannot again guarantee what will be the out come after the implementation in the circuit, while n- channel ones are easier to use and intelligent to manner in brilliant designs so far..........
UPUPA,
I think your senses were ruled by complementary bipolars only, try to get your senses to be ruled by high definition N-channel power amplifiers. Then only you could get a perspective vision of NVMOS sound. As long as price is concerned we have implement N-channel mosfets worth 15dollars each in our amps which donot have any p-channel complements. paying the huge price for p-channel only lets you to get poor performance not a good performance.
Workhorse said:
For amps up to 45V rails I like to use IRFP240 and IRFP9140 as complementaries (and they actually track very well). There is practiclally no price difference at between them.
But you are right that price and availability (and even existence) of parts becomes a real problem once you get over about 90V rails if you want to go the complementary way. After that it is N-ch only.
Upupa, designs like the standard quasi-complementary arrangement, where one side is a follower and the other some form of CFP will indeed generate more even order distortion because these two have very different transfer characteristics. There are however different driving arrangements that largely circumvent that and I immagine they would sound jut fine.
Hi Ilimzn,
Can only agree with you on testing clipping recovery, etc.. For me it goes without saying.
When I suggested you put Re's into your sim supplies, I meant that to 'pick up' the spray of HD from the output stage commutation currents to test the PSRR as a THD measure. Like a real amp not a sim one with stiff supplies. A generator attached to the supplies is not quite the same - unless it can do a 1V 100Hz sawtooth plus the HD spray above - I'd like one of those.
I recently simmed an amp with 0.000006% THD 400Wrms/4 ohm/1KHz that went to 0.035% with Re=0R22 in the supplies!
Guess the composition. Not pleasant.
Workhorse,
I can understand your reasons for N channel as you are concerned with sheer POWER - nothing else and, of course P ch is not so easy to drive and may be more expensive and more lossy. It's not really a problem at 100W - 200W amplifier powers and people would be more concerned at the quality of delivery of that power, i.e. 10% THD would be unacceptable ( especially if it's 2HD,3HD......nHD ALL). Perhaps not important for Workhorse PA market, after all it's common that a party doesn't liven up until the music is distorting (clipping). A 1000W clean amp will do it and annoy the whole neighbourhood while a mini 3 in 1 can hit party mode at not much more than a couple of watts.
Given that 400W of Pch MOSFETS are not THAT difficult to drive, plenty of nice push pull complementary at low power stages are available to drive them and with low OL distortion so a credible audiophile amp can be produced quite simply.
I'll happily watch while you sim your contorted Nch creations that almost...
Perhaps you could try some Re in the PS lines as well just for an extra perturbation.
Can only agree with you on testing clipping recovery, etc.. For me it goes without saying.
When I suggested you put Re's into your sim supplies, I meant that to 'pick up' the spray of HD from the output stage commutation currents to test the PSRR as a THD measure. Like a real amp not a sim one with stiff supplies. A generator attached to the supplies is not quite the same - unless it can do a 1V 100Hz sawtooth plus the HD spray above - I'd like one of those.
I recently simmed an amp with 0.000006% THD 400Wrms/4 ohm/1KHz that went to 0.035% with Re=0R22 in the supplies!
Guess the composition. Not pleasant.
Workhorse,
I can understand your reasons for N channel as you are concerned with sheer POWER - nothing else and, of course P ch is not so easy to drive and may be more expensive and more lossy. It's not really a problem at 100W - 200W amplifier powers and people would be more concerned at the quality of delivery of that power, i.e. 10% THD would be unacceptable ( especially if it's 2HD,3HD......nHD ALL). Perhaps not important for Workhorse PA market, after all it's common that a party doesn't liven up until the music is distorting (clipping). A 1000W clean amp will do it and annoy the whole neighbourhood while a mini 3 in 1 can hit party mode at not much more than a couple of watts.
Given that 400W of Pch MOSFETS are not THAT difficult to drive, plenty of nice push pull complementary at low power stages are available to drive them and with low OL distortion so a credible audiophile amp can be produced quite simply.
I'll happily watch while you sim your contorted Nch creations that almost...
Perhaps you could try some Re in the PS lines as well just for an extra perturbation.
jacco vermeulen said:
Hehe, obviously I have to give away all my secrets
Upupa Epops said:
Yes, complementary as far as the same cell technology, just N changed into P, but electrically, not even close save for the same Vdsmax.
darkfenriz said:
IRFP340 / IRFP9240 are a reasonable match (within 5-8%), and could probably get very close with limited selection. Not as good as IRFP240/9140 but about 7x better than 240/9240
I have a ton of models to add to my simulator to see if I can find more - this is what I use as a first approximation as it is really simple to compare transfer characteristics that way. Once I find a match, testing them real world is in order...
amplifierguru said:
Ah, I thought you meant Rs for the MOSFETs (I never put them there unless I have to parallel MOSFETs).
I have a couple of things that I normally do when testing.
I put a series connected voltage source in the power lines that I sim for AC analysis in order to get the PSRR curves. If there is a problem, it will be quite evident there, especially as you can see where most PS hash is injected (though of course for some topologies it hardly takes a genius to figure it out without a simulator).
The other popular one is putting resistors into the power lines and backdriving the output with a current source. I do the second part anyway to figure out OL gain, distortion, impedance by leaving only DC feedback.
If nothing else, this should be attempted to simulate a first approximation of fuses in the power lines (which is generally not the best place to put them and the ARE nonlinear by definition!).
I sim for THD with R in the power lines only after I do the two above, no point really in seeing there is a problem (and there always is) unless you can see where it happens
To be honest I'd like to see the amp as well as the simulator and models used - I'm not sure I would trust anything more than -120..-140dB in the harmonic analysis, if even that. Just recently I've seen all sorts of numeric problems on one design but then, that is implementation and platform speciffic to a degree. One of the best examples was an 3-stage amp that had a -200dB second harmonic, 82dB OLG and 26dB CLG. I'm not saying it is IMPOSSIBLE, just that I'm not that lucky
I know this was for workhorse originally. If you have read my answer regarding that simple R-to-R amp I posted in the other thread, the N-only output was a consequence of the driving stage chosen because it was simple and could drive it R-to-R. If I have complementary pairs, I use them
amplifierguru said:
Guru, I think your judgement lacks an absolute view of professional amplifiers. professional amplifiers are for POWER+LOWER DISTORTION not a junk of soundcrap material or just a low grade high power clipper machines.
Regarding quality, our professional high power amps have THD less than 0.01% at full power[++1200W] in 20KHZ .
Secondly, your experience in designing the BJT and Lateral mosfets amp is quite high[due to you past expertise], but our experience in designing the N-channel Vertical Mosfet maybe greater then yours.
I dont know yet that you have tried nmos amps or not. People dont just buy a pro-amp for just power purposes, they also look at the specs and listen to there ears to compare the best available amp in the market .Amps ranging from 100 to 500W are very moderate powered ones and are easy to design than the amps representing kilowatt power levels and presently only the N-channel mosfets & IGBTs are the devices in the market that are best for high power levels upto 5KW in Class-AB.
A good professional will not annoy the party, unless it is operated under its limits and the operator has the brains to operate it. Clipping can be heard from any amp whether its a home hifi or pro amp.
When someone wants 2400Wrms of Audiophile quality , then only N-mos amplifiers excels more than any other bipolar design.
Since you are a senior designer, you must have the acquisition of the perspective things related to present day pro-market trends.
Things usually dont happen in that way in which one wants them to be,one has to do a continuous ..............
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.