First: The MOSFET body diode is NOT a zener! it's just a normal silicon PN junction diode inherent in ALL MOSFETs. It is wise to parallel a smaller faster diode directly at MOSFET source-drain.
Second: A more powerful MOSFET with higher current rating and lower Rds-on is not an upgrade. Unless u plan only to drive very low impedance loads but it will likely give you more problems than expected and certainly worse sound unless the mod is followed thru elsewhere (driver, filter, driver power supply, wierd spikes, etc).
Before concidering an upgrade, or modification, you must know the switching frequency, and the capabilities of the MOSFET driver. Can the driver provide 1A, 2A, or whatever.
The high-side boost circuit of the driver may have been designed for a given MOSFET (the boost cap value is very important). The dead-time as mentioned is another issue.
Select MOSFET with the lowest possible gate-charge and just low enough Rds-on for decent power dissipation for max output current, do not over-rate it (if you need 10A max, dont select a 100A FET). Remember if 10A is peak output for rated power, the amplifier will almost never reach that, it will probably spend 99% of it's time at 1-10% of max power.
Second: A more powerful MOSFET with higher current rating and lower Rds-on is not an upgrade. Unless u plan only to drive very low impedance loads but it will likely give you more problems than expected and certainly worse sound unless the mod is followed thru elsewhere (driver, filter, driver power supply, wierd spikes, etc).
Before concidering an upgrade, or modification, you must know the switching frequency, and the capabilities of the MOSFET driver. Can the driver provide 1A, 2A, or whatever.
The high-side boost circuit of the driver may have been designed for a given MOSFET (the boost cap value is very important). The dead-time as mentioned is another issue.
Select MOSFET with the lowest possible gate-charge and just low enough Rds-on for decent power dissipation for max output current, do not over-rate it (if you need 10A max, dont select a 100A FET). Remember if 10A is peak output for rated power, the amplifier will almost never reach that, it will probably spend 99% of it's time at 1-10% of max power.
By testing it. I've done it many times. It is 1.1 to 1.3 times Vdsmax.
Temp co is basically the same as regular reference diodes.
And many MOSFETs are avalanche rated. This is a very important behaviour for all MOSFETs. Excellent or at least good z-diode behaviour for body diode of a Si MOSFET is compulsory.
OK, it is not zener, since only diodes under 6 V can be called zener, at higher voltage avalanche breakdown occures, and Zener (RIP!) explicitely asked not to call those diodes zener, but z-diodes (by the shape of the characteristics).
We don't avalache MOSFETs in amplifiers, so it aint relevant. The 'body diode' is the intrinsic PN junction diode. It is just like a typical silicon diode with forward voltage ca 0.7V.
Putting in semiconductors that are that different is nonsense. Larger input capacitance might need a higher current driver, or can even cause instability, your bias will need to change, in some cases the thermal tracking can become unstable ( more often with bjt's ). A properly designed amp considers all the important parameters of the output transistors, you change those parameters all bets are off. So very much like putting a truck motor into a sports car.
We don't want avalanche, but it happens many times no matter how we dont want it. And improperly designed MOSFETs fail sometimes because of this. And it is relevant because the opening post asked about it.
Is there such thing as "typical silicon diode", and if yes, then is it different from z-diodes? Is forward voltage relevant in determining if a diode was zener or not?
Possibly. But you don't know. So it is not nonsense, just needs further informations.
We are in ClassD forum. And not my bias.
How do you know it was properly designed? How do you know it cannot be improved with a 20 years younger FET?
A truck motor will not fit into a sports car. And it will react too slowly. Etc... The problem with this idea is not the unknown behaviour. It is known to be bad. In contrast that MOSFET is not known to be bad in the amp. It is unknown.
Well if it avalanches the MOSFET has too low Vds rating. Pick one wth higher Vds breakdown rating.
What I mean by 'typical ' diode is a silicon pn diode with forward voltage ca .7volts.
All zeners are simply silicon diodes with controlled (tested) reverse voltage levels. All zeners can be used as typical diodes with forward voltage of ca 0.7 volts if turned the other way round. Just remember it will conduct both ways if zener threshold is passed. All diodes will zener at some point beyond it's reverse voltage rating, but the voltage will vary and is unknown untill it happens.
I am just putting out cautions too look for. You keep saying I don't know, do you know it will definitly be an improvement? No you don't. This is the real nonsense. I make educated assumptions that you can't deny but you call it nonsense. It is known the amp was not designed for the new fet, it was designed for the old one. And a fet with more capacitance will also react slower. So my analogy stands. Your nit picking looks like guesses, and it sounds like you need to learn more about semiconductors and amp design. You may improve the amp but if there are any major differences in parameters probably not. If knowledge about semiconductors is nonsense, but willy nilly parts swaping is not, suit your self but don't tell others it's a good idea with out proof.
Sorry, but that's plain wrong. There are no 'bad MOSFETs', just badly used ones. Unless you know the amp and how the amp works, it's Schrödingers MOSFET, it's bad and good at the same time until you look at it (closer). Even if the possibility it's an improvement is very slim. He didn't write anything about the amp, so you can't know if it's bad or not.
I already wrote this. No news.
That we dont know about an unknown amp?
Read it again! I called your false analogy nonsense. And every relevant assumption were made after this. Its impossible to call them nonsense before they were written, dont you think? You know causality.
Or not designed at all. Ive seen too many amps built without real designing.
You don't know if "slower reaction" causes significant problem there. Its easy to design an amp that is quite optimised for IRF540 but works better (runs cooler, and not higher distortion) with this AO... OK, gate drive dissipation will be about 400 mW higher for 1 cannel, who cares?
You dont learn! Ive already warned you to the danger of surface similarities!
Learn reading! I never wrote it was a good idea. Just not nonsense like your unneeded car example.
You can change any FET with any FET as long as you are willing to accept the consequences 😱
That depends on how close "pretty close" is 😉
Seriously though, if you can no longer find a source for the original FET the best option is to try and find a substitute FET (obviously in the same package) with as close to the same specifications as possible - not just Vds, Rdson, and Id but also Vgs, Qg, Qgs, Qgd, td(on and off), tr, tf, and body diode characteristics. Watch out when comparing specifications because not all data sheets use the same test conditions (even from the same manufacturer). The body diode characteristics also matter, but as long as the current specs, Qrr, and trr are the same or better it's probably ok.
Ofcourse I support these statements but the analogy with the car is wrong. You don't know what's the goal. If it's going to pull a huge trailer, it's not 'bad'.
Zener breakdown and Avalanche are different physical phenomena to actual semi designers, serious device physics students - or even just designers who have read up on this stuff for decades
Last edited:
'The zener effect' does not dominate at any voltage, there aren't any more zeners at 5V than at 50. The voltage were the diode 'zeners' is where it's reverse break down is. Someone figured out how they can manufacture the diodes with different breakdown voltages and then sell them as zener diodes for various voltages. Their tempco is well understood and simply the nature of PN junctions.
If you expect the MOSFETs to see 80V peak, then pick a 100V rated MOSFET. If it will see 100V you must pick one with a rating of 120V, etc,etc. Nobody designs an amp with the assumption the MOSFET's avalache capability will handle the higher voltages than waht the FET is rated for.
That avalache diode is nothing anybody designing amps should even bother thinking teeny weeny bits about.
OK, nufff said about this off-topic, at least fer me.
- Status
- Not open for further replies.