Best audio power MOSFET

[ervinl]

Hi,

I want to know what is the selection of high power mosfet (N channel) for audio amplifier purpose. Currently I use 2sk 3911 (Toshiba), seems it has high spec (600V, 20A), but very high input capacitance (around 4nF).

Considering another kind of MOSFET to get more clear, nice sound of my DIY power amp, or just to parallel the same mosfet to get more damping factor?

Some mosfet available e.g. IRFP 240 (popular), 2sk 1058 (also popular) etc. The original design is using IRFP 350 hexfet. There are some other mosfets available in my local market (not remember).

Thanks,

Ervin L

[CBS240]

Any schematic available?

2SK1058 is a lateral mosfet. It does not require temperature compensation as do the vertical types like IRFPXXX ect in Class AB bias. Look at the transfer characteristics of these two devices and you will notice the zero temperature coefficients is not the same. The vertical types require some type of temp compensation, as do BJT's, at class AB bias levels even if it is Quasi.

Damping factor has much more to do with the drive circuit than it does the overall Gm. Paralleling many together won't help you much with this variable.

[abraxalito]

Does your application actually need a 600V device? If not, I'd go for a lower voltage one as these generally have superior gm.

[ervinl]

Which one is better for MOSFET Power Audio Amplifier, higher voltage or higher current (e.g. to achieve linearity, slew rate, good damping factor, low THD etc)?

I ask this basically to enhance current DIY Power Follower (Pavel Macura based).

If there should be thermal adjustment/compensation, how to do so with this simple SE Class A? I also notice when it runs hot, the bias setting will be different than when it is cold (different voltage measured at source).
Or does it mean using lateral MOSFET is better for this schematic (instead the original was IRFP 350, an HEXFET, and some other schema is IRFP 240 and 150).

[jackinnj]

The lateral mosfets have their thermal advantages, but also lower transconductance. I think that they sound great, but so do the IRF and Fairchild MOSFETs I use in my F5 amps.

Troy Huebner wrote application note for National Semi in which he discussed the merits -- I think that the best measured results came from the 2SJ201/2SK1530 -- but these do need some thermal compensation. The Ap Note is, I believe, AN-1645. In his case, the driver they were using (a Nat Semi chip) couldn't provide sufficient bias to the IRF devices, and he notes that this is a shortcoming of the analysis.

[nigelwright7557]

Originally Posted by ervinl

Hi,

Some mosfet available e.g. IRFP 240 (popular), 2sk 1058 (also popular) etc. The original design is using IRFP 350 hexfet. There are some other mosfets available in my local market (not remember).

Thanks,

Ervin L

I have had good results with IRFP240. Its input capacitance is quite low compared to other IRFP series. They are also very cheap compared to some of the more exotic MOSFET's.

[umut1001]

IRFP240 is good.I used fqa19n20c and 2sk1530/2sj201.The best result in my F5 amp is with Toshiba mosfets 2sk1530/2sj201.They sound better than the others(but unfortunately very expensive)

[ilimzn]

Vertical power MOSFETs (and similar technologies, such as PI and trench types like Toshiba J200/K1530) tend to have a 'sweet spot' where the capacitance, gm and current are sensible enough tp get good performance in audio - usually this happens in MOSFETs with Vdsmax between 100 and 300V - usually on the lower end of this range for P channel MOSFETs, and on the higher end for N channel MOSFETs fro, the IRF(P) and similar MOSFETs. For the same geometry, P and N channel MOSFETs get to have different characteristics so it's not trivial to find complementary pairs, unless the MOSFETs were not made to be pairs, like the Toshiba parts - if you opened up the case, you would find completely different chips inside, size wise as well as what is diffused on the silicon.
What i am trying to say is, there are a LOT of MOSFETs that will produce great results in amps IF you know how to use them properly - just like any ather part. But there are hidden pitfalls - like finding correct complementaries in the case of IRF(P) types, for instance. Also, the output stages can topologically be quite different in order to drive MOSFETs correctly - it's not just a case of sticking different output devices into the amp and biassing them properly. One thing to look out for and handle, is the nonlinearity of input capacitance, as well as gm at low drain currents. All MOSFETs have this problem, however, progress in the way MOSFETs are normally used (power electronics) tend to provide ever better characteristics for N channel MOSFETs which points to use in single ended class A or quasicomp circuits of multiple kinds. That, just like declaring a single type of MOSFET is 'best' just means you happen to be satisfied with solving the problems inherent in ANY MOSFET by the brute force method at hand, and declare the trade-off (and there is always one or more of those) as acceptable - when in fact, it's all the same underlying physics and set of problems, just in different amounts, producing different 'flavors' of the same basic behavior.

[nigelwright7557]

Originally Posted by ilimzn

That, just like declaring a single type of MOSFET is 'best' just means you happen to be satisfied with solving the problems inherent in ANY MOSFET by the brute force method at hand, and declare the trade-off (and there is always one or more of those) as acceptable - when in fact, it's all the same underlying physics and set of problems, just in different amounts, producing different 'flavors' of the same basic behavior.

The IRFP240/9240 pair are by far not compliemntary yet you can still design a decent amplifier using them.

A lof of differences are reduced by using lots of feedback.

[ilimzn]

True, but as you know, the best way not to have a problem is not to correct it when it's there, but not to have it in the first place. This sort of thing tends to result in abuse of NPV. One other possibility would be to put a source resistor on the IRFP240, but not on the 9240. Or, simply, use IRFP340 instead of IRFP240. My point is, know thy MOSFETs - there are many methods to make them sing.