transistor ?

You need to take a close look at both the IR-HEXFET and the V-MOSFET temperature ratings. I'm not very familiar with the Japaneese MOSFETS but the IRF's have very high chip temerature ratings. This may be an advantage with the IR HEXFETS. These amps run extremely hot so using the FET with the better temoperature ratings will make for a more reliable amplifier. It may also be that Nelson likes everything to be made in the good ole USA.
Mark Gulbrandsen
Salt Lake City, UT

Nelson Pass

The one and only
Paid Member
2001-03-29 12:38 am
It's just as likely that Nelson has a warehouse
full of them :)

Actually, the IR types are pretty much a standard,
and are available from many other manufacturers.
International Rectifier is the dominant supplier of
power MOSFETS. I have tried IR parts from other
companies, but have found that IR has the most
consistent product.
Aaark! Mini Nelson


“Hi-fi audio tester” hey? Does that mean spending your life lounging back in a comfy chair, glass of red in one hand (well even if it is Californian red :p), remote in the other, while you keep your PA busy changing CDs for you? I bet it’s even one of those magical glasses, the ones that are always kept full of company expense wine too … I have the other type, the one that always seems to be empty!

What a great job :)

[Edited by Pete Fleming on 10-27-2001 at 05:51 AM]
Choosing Power MOSFETS

I have used multiple Toshiba 2SK1530 and 2SJ201 for my output stages operating class “A” for quite a number of years. These Toshiba devices can be very difficult to find at times but are very well suited for audio use.

I have on hand a number of Magnatec N and P Channel devices in TO247 cases that I will be testing in several months. These lateral MOSFET's are reportedly designed specifically for audio use and should provide performance equal to or exceeding that of the Toshiba parts.

A link to the Magnatec site is:

The Magnatec MOSFETS have a none standard pin out, which makes them not compatible with most existing circuit boards. They can also be difficult to obtain due to limited distribution channels.

I have used, and do use IRF devices in some stages with excellent results. But as I remember they had some what higher distortion figures when compared to the Toshiba parts when I used them in output stages, Thus in my opinion they may not always be the best choice.

I shall be testing my old MOSFETS from Toshiba and some from IRF, along with the ones I have from Magnatec in the output stage my new amplifier in the next two or three months. This new amplifier does not use a source follower output stage so the differences in device performance may be quite noticeable.

It is only logical to conclude that devices that are most readily available are selected more often for new design or construction and are thus used the most. This is likely to be true even though such devices may not be the best choice for the particular application.

John Fassotte
Alaskan Audio

[Edited by alaskanaudio on 10-27-2001 at 07:49 PM]

Nelson Pass

The one and only
Paid Member
2001-03-29 12:38 am
I don't want to get off on a rant here, but ....

Aside from cost and availability, a lot of the
decision making about what to use comes down to
some simple parameters.

Let's take an example of three popular transistors
from IR: the IRF230, 240, and 250.

Device, Watts, Amps, Capacitance, Transconductance

230, 75, 9, 800 pf, 5 S

240, 125, 18, 1300 pf, 10 S

250, 150, 30, 2600 pf, 20 S

If you look at the chips, you will see that the
surface area is somewhat proportional to these
parameters, except for the wattage of the 250 which
shows the limits of package dissipation.

As a practical matter, you can make a 240 out of
two 230's, or a 250 out of two 240's.

I have had good experience with all three parts, and
used singly, each has a distinctive character. The
230's have a little more accurate but drier top end,
and the 250's are better on the bottom end, but not
as accurate on the top. This is pretty much what you
would expect, and I have found that the happy medium
is the 240. The 040 and 140 are similar parts but
are doped to lower voltages and have higher

The earlier Japanese parts such as the Hitachi's
2SK175's (for example) had transconductance figures
on the order of 1 S, and capacitance figures of
around 800 pF. From this we can generalize that
modern power Mosfets give you more current and
tranconductance for a given amount of capacitance,
which is a reasonable basis for preferring them.

In actual practice, the IR type parts give me most
of what I want, which is better sound with a simpler

You might prefer the sound of the older, lower
transconductance parts, but that hasn't been my
experience. Of course with something like a Zen
amp, you can substitute them right in and see for
yourself. :)
Choosing the right active parts

I agree with Nelson Pass that there is more to choosing a transistor than cost and availability. This is especially true when you buy these by the thousands instead of the small quantities that the average DIY might purchase.

When I evaluate a transistor or any such active device I start with a very basic circuit to see how it reacts in the circuit using a controlled stimulus. The best device will in general show a proper relationship between the measured levels of the harmonics produced at a reference output level. Device gain in is going to be less important than the harmonic spectrum produced.

After the best device is found by using the simple circuit method the circuit is then redesigned and made more complex by adding parts to enhance the performance. This enhancement may include parts used to provide controlled feedback loops and thermal stabilization.

There is of course a limit as to how much a particular circuit design can be enhanced by making it more complex. Each designer has to establish his or her own definition of what good performance is.

John Fassotte
Alaskan Audio

[Edited by alaskanaudio on 10-29-2001 at 10:04 PM]

Nelson Pass

The one and only
Paid Member
2001-03-29 12:38 am
The capacitance of the Mosfet is an issue for
two reasons, first because (of course) it will
limit the high frequency bandwidth given a
particular source impedance. This is not a big
problem, but must be considered.

More importantly, the capacitance of the device
varies with voltage, and this creates non-linearity
at high frequencies.

The capacitance of the Mosfet is the reason for using
fairly low impedances in the input networks of the
original Zen amp.

You might think that paralleling devices exacerbates
the capacitance effect, but this is compensated by
the increased transconductance, and is not such a
big problem until you parallel LOTS of devices.