OK, I get it. But I really don't need more than 100W, so I will try with some other MOSFETs.
And just to clarify - I'm building NMOS 200. Quasi said that Ciss parameter is very important, I was planning to use MOSFET with Ciss less than 1nF, but I see that Malta says even MOSFETs with more than 3nF are applicable. What is the biggest acceptable value for Ciss?
If you build class D amplifiers, Ciss / RDSon / Qg / rise and fall time is very very important,
for class AB audio from 5 to 20 khz music playback forget all that, completely uninteresting Hi-FI fairy tales,
every mosfet can do 5 to 20 khz, that's absolutely no challenge, the mosfet laughs about it
even the worst slow 500Volt SMPS mosfet can do the job very good in Class AB with great clarity sound, but will fail 100% in Class D from the first time
The only important thing when choosing a mosfet for class AB is that the mosfet can withstand a very high power loss, so for example 1 pair like FDA69n25 or similar power dissipation if you want to listen to loud music with class AB.
IRFP250 is ok for living room volume, that's enough for the neighbors to knock on the door for safe operation.
Believe me or not, also built the NMOS 200 a long time ago, never fail !
If you don't need more than 100W choose 1 pair FDA69n25 or similar power dissipation, the amp is very safe and the neighbour will love you
Last edited:
It is not Class D, it is AB I think (NMOS 200). But Quasi (designer of the amp) said Ciss is important in this case. I simulated it in ltspice and not much change regarding frequency response with different MOSFETs, but maybe there's something more about it that I don't know.
Thanks for response!
I've read somewhere in this thread that the max. output power is total MOSFET max. dissipation power divided by four, so If I need 100W, I can put two IRFP250 (I can buy them locally), they have total of 428W dissipation. Right?
Hi RankotIt is not Class D, it is AB I think (NMOS 200). But Quasi (designer of the amp) said Ciss is important in this case. I simulated it in ltspice and not much change regarding frequency response with different MOSFETs, but maybe there's something more about it that I don't know.
Thanks for response!
I've read somewhere in this thread that the max. output power is total MOSFET max. dissipation power divided by four, so If I need 100W, I can put two IRFP250 (I can buy them locally), they have total of 428W dissipation. Right?
Ciss is important because it affects the load seen by the driver transistors as you approach 20Khz. In a 100 watt application where you might use only 1 or 2 pairs Ciss is unlikely to be a problem. It might be of concern where you use 6 pairs as I have in the past. Remember the gate has to be discharged as well to avoid too much cross-conduction.
Malta is correct that Ciss is more important in Class D amplifiers that are switching hundreds of Khz. Whilst MOSFETs are voltage driven devices the charge and discharge currents on the gate are significant and require a robust driver (same for switching power supplies). Also if the MOSFET gate is not charged fast enough it can get hot during the transistion between off and fully on and this can cause it to fail. Higher Ciss = slower switching time.
Malta is also correct that even a "slow" MOSFET is plenty fast enough for audio and generally faster than most output transistors. This is why they are prone to oscillation if not designed well with good layouts.
My designs have all centred around highly available and inexpensive MOSFETs. To date these are IRFP240, IRFP250, IRFP450, IRFP460 and IRFP840 (in pairs). I'm not promoting International Rectifier at all, it's just that I first published this design nearly 17 years ago when there were less MOSFETs around.
I wouldn't get too hung up on seeking the perfect device. Anything with specs similar to the above will be fine - even perfect for your application.
BTW, Plenty of these amps have been built ranging in powers from 100 to over 900 watts.
Cheers
Q
Last edited:
It is not Class D, it is AB I think (NMOS 200). But Quasi (designer of the amp) said Ciss is important in this case. I simulated it in ltspice and not much change regarding frequency response with different MOSFETs, but maybe there's something more about it that I don't know.
Thanks for response!
I've read somewhere in this thread that the max. output power is total MOSFET max. dissipation power divided by four, so If I need 100W, I can put two IRFP250 (I can buy them locally), they have total of 428W dissipation. Right?
Mosfets are not BJT, Mosfets paralleling in Class AB is to avoid without snubbers otherwise Oscillation and the amp will kill itself
In this case dont use two pair IRFP 250
MOSFET max. dissipation power divided by three and BJT by five
Use 480 Watt one single pair FDA69n25 Mosfet or with similar power dissipation and you dont need snubbers, paralleling Mosfets, source resistors and enjoy
You can believe me, since I make Class D how much hi-fi junk that supposedly matters has been written and believed about class AB amps.
For security reasons, transistors are matched in class AB for
PA power amplifiers from 500 watts to the kilowatt range, the transistors had the same beta. All this is no longer necessary since PA amplifiers work in Class H / Class TD / EEEngine mode. Nobody buys heavy class AB high output power PA amplifiers anymore
when Quasi built these great amps many, years ago there were no 480 watt mosfets on the market like FDA69n25
I have build mor then 10 years ago NMOS and ACTRK, great sounding amps, have using only 1 single pair IRFP460
for Hi-Fi to listen loud Rave and Techno music with big speakers, often pronlem with neighbours !!!
Yes and Ciss is important when you use 6 pairs but not for 1 pair Mosfets in Class AB
and for 6 pairs in Class AB its better to add Totem Pole driver same in Class D to drive the mosfets
have fun
I just wanted to point out, that this power rating and SOA of these mosfets usually applies to switching applications.
For a linear application (such as audio amp), it's very unrealistic. Check SOA charts in the datasheet.
In real life, in audio amp, you can't dissipate more then 60W from plastic mosfet on the constant basis.
Also, even for 60W per mosfet you can't get away from active cooling of some kind.
So all these "480W per mosfet" assumptions, are fairy tales as far as audio amps go.
I would estimate that in audio amp, from one pair of mosfets (even FDA69n25) it would be very difficult to get more than 75W-90W,
assuming very good cooling.
All in all, it's much easier to use multiple pairs rather than one pair with liquid hydrogen for cooling
Source resistors are increasing THD. With one pair of mosfets we can skip them.
For multiple pairs, they have to be used, but the smaller the better.
Which means - it's better to match mosfets, and minimize values of source resistors.
In multi-pair amps, widely unmatched mosfets WILL oscillate, with or without snubbers.
Last edited:
OK, I have just found few IRFP460 in my drawer, so I can try to use one pair? I have already built TO-220 version of the PCB, but I will extend leads to MOSFETs and mount them on the heatsink. I have installed source resistors, too, and I will leave them (2 x 1 Ohm in parallel).
Which matching procedure shall I use?
Shall I use this approach for low current https://www.diyaudio.com/community/threads/nelson-pass-easy-peasy-mosfet-vgs-measurement.299546/, or Nelson's high current matching?
Don't see why not. They seem OK.
If you are planning to connect mosfets by wires (not on PCB) - gate stoppers (resistors) should be soldered as close to the the gate
as possible - which means NOT on pcb, but on the mosfet. Since chance of oscillations is higher in this case, snubbers mentioned by Malta will also be useful.
Check out "Alpha Nirvana" amp thread - they talk about wires/snubbers there. There is a little snubber PCB designed for that purpose, all together with gate stopper on it.
Any method of matching mosfets will be ok as long as they are matched at higher current, not just couple mA..
If you are planning to use only one pair per channel, there is no need to match anything.
I have redesign 2022 Quasi ACTRK, have replace discrete with OPamp Input stage, Crowbar DC Protection, Clip Limiter
Only 1 Mosfet Pair FDA69N25
waiting PCB for Trial
Hi Rankot.
DC speaker protect is a must. A relay based one can be found here; https://sites.google.com/site/quasisdiyaudiosite/dc-detect-speaker-protection
I do like however the solid state relay designs. One is here https://www.diyaudio.com/community/threads/speaker-protection-board.377896/, but also do a search.
Cheers
Q
DC speaker protect is a must. A relay based one can be found here; https://sites.google.com/site/quasisdiyaudiosite/dc-detect-speaker-protection
I do like however the solid state relay designs. One is here https://www.diyaudio.com/community/threads/speaker-protection-board.377896/, but also do a search.
Cheers
Q
This one from your site use normally on or normally off relay? It also needs external power, so it is OK if placed inside an amp, but I'd like to have one which can be placed inside the speaker.
But in case I decide to put it into amp, I need something for mono amp, so I'll probably go with your discrete schematics and add separate power transformer for protection circuit.
I really like this one with MOSFET protection, but those optocouplers are not available in my contry, and I can't wait for two months for parts right now
Hi Quasi,
It's like my first post in here, so don't shoot me, if this has been discussed before...
I noticed that you have this and the NPN-transistor version of this design. There are many powerful IGBT's available. Is it possible to use IGBTs in audio buffer stage? Or in these designs? Just curious.
-Tommi
It's like my first post in here, so don't shoot me, if this has been discussed before...
I noticed that you have this and the NPN-transistor version of this design. There are many powerful IGBT's available. Is it possible to use IGBTs in audio buffer stage? Or in these designs? Just curious.
-Tommi
Hi Tommi
With some component changes my designs could use IGBT's as the output devices. I haven't played with IGBTs before, however here is an example of a design by Elektor magazine. https://www.diyaudio.com/community/threads/giesberts-igbt-amplifier-elektor-1995.156966/post-2017939. Note this is a complementary output stage using N & P channel devices.
Cheers
Q
With some component changes my designs could use IGBT's as the output devices. I haven't played with IGBTs before, however here is an example of a design by Elektor magazine. https://www.diyaudio.com/community/threads/giesberts-igbt-amplifier-elektor-1995.156966/post-2017939. Note this is a complementary output stage using N & P channel devices.
Cheers
Q
I've spent whole day yesterday reading various articles about that, and here is my solution, added to another thread, to stop hijacking this one: https://www.diyaudio.com/community/threads/speaker-protection-board.377896/page-5#post-6927637
Thanks to everyone for sharing knowledge, experience and a good will!