IRF540 Amplifier

[Ermac]

Can I use IRF540/9540 in this amp if I keep the PS voltage below 50V?

[CBS240]

Short answer, no.

2SK1058 and 2SJ162 are lateral type fets and IRF540/9540 are vertical types. The bias will not hold steady with temperature unless you use compensations, Vgs multiplier or something. The other option is to bias at their zero temp co. point. Being at several amps, this is impractical. The zero temp co. operating point for the 2SK1058/2SJ162 is around 100mA.

[AndrewT]

Ermap.
ESP will object to you posting his schematic on the Forum.
Try asking him next time.

His usual request is that you post a link to his site/project/schematic.

[Ermac]

I just read this about Hexfet...
ESP Article

Thank you!

[jaycee]

If IR540 is the device you want to use, build the TO-220 version of Quasi's NMOS200 amp.

http://users.adam.com.au/cgpap/QuasiWeb/index.htm

[Ermac]

Originally posted by jaycee
If IR540 is the device you want to use, build the TO-220 version of Quasi's NMOS200 amp.

http://users.adam.com.au/cgpap/QuasiWeb/index.htm

Thank you! What nice amps!

[CBS240]

Originally posted by Ermac
I just read this about Hexfet...
ESP Article

Thank you!

Hi Ermac

IMO, hexfets are not suitable for good fidelity, even regarding this article, mostly because at some point in time, each device will be operating in that region below 1A and at cut-off. The transconductance becomes very non-linear in this operating range. However, one can vastly improve the performance of a class AB hexfet OPS by means of error correction. Bob Cordell has a great paper on this based on Dr. Hawksford's error correction for Darlington outputs here. Even with EC the bias for these devices should be higher than for BJT's or crossover non-linearity will show up. BJT transconductance is much more linear in this operation range.

For example, in this photo, the top waveform is showing the output of an EC hexfet amp (30KHz @ 2Ap, zero degrees), and the error signal on bottom, the error being the difference between the output and the gate drive with the output hexfets being bias in class B or cut-off. Notice the cross-over is obvious in the output waveform because the error does not have enough slope and cannot keep up with the high frequencies needed to correct for this amount of crossover distortion.

This photo shows the same setup but with 300mA of bias. The cross-over is still there as indicated by the change in slope of the error signal around the crossover point. Without EC this error would show up as distortion at the output and if GNF is employed, would be mixed with the input and create IMD products. A higher bias only helps a hexfet become a bit more linear in crossing through this region but does not correct the problem, the distortions are still there, worse than BJT's for certain. Thus, IMHO, hexfets by themselves are inherently not very good for linear audio.

BTW, my best sounding amp is an EC hexfet design.

[Ermac]

Hi CBS240,

I really refused to build a hexfet amp as I noticed they don't have any output protection beside of their instability.
BTW which powerAmp do you suggest me to build?

[CBS240]

Yes, instability can be a problem with hexfets. Layout is important and they typically require gate Zobel filters and gate stopper resistors in order to really be stable. OTHO, they are very rugged and quite cheap.

As for which amp to build, it depends on your experience level, but I would first go for a fairly simple circuit like the one in the link above, but IMO, you should build a version with a single output pair and suitable rail voltages to start with. When you get that working properly, then move up. It is a lot cheaper and a lot less discouraging if you only blow up 1 pair of OPT due to a mistake than several at a time.

Good luck

[Ermac]

I have made several power amplifiers till now. one pair and two pair output stage with 2n3055, TIP35, 2n3773.

But I meant by the question that which amplifier do you suggest between LYNX and Melody?
I like to build the LYNX, but I do suspect if powerAmps having opamp in their first stage are as good as those entirely with transistors. Please guide me regard this.

Regards!
Ermac