FET's?

I personally like the 2SK1529/2SJ200 for outputs. These are Lateral types, But the negative TempoCo doesn't kick in till 6 Amps or such, so they still need a Vbe multiplier.

I've Used standard Laterals (SK1058/SJ352 and the higher voltage Equilivant) but find them to be a bit touchy. I've blow through too many of them while experimenting with circuits, so I use the above Mosfets as they are very robust and I've lost very few of them. Also the Above Fets are touted as being made specifically for audio.

Seems the other popular candidates are the I.R. devices (IRFP240/140/040 Etc.). I've used the IRF540/9540 and they seem alright. However the Vgs on is a tad higher than the above devices. I still use the IR device for power supplies.

This is just my experience/opinion, hope it helps.

-D.
 

ilimzn

Member
2005-02-11 1:25 pm
Zagreb
A word of warning about vertical MOSFETs (including T-FETs, HEXFETs etc):
Types claimed by the manufacturer to be complementary, are not so electrically, but merely by the fact that N and P doping is reversed (oversimplification but close enough). This means that the P parts have higher Rdson and lower current capacity and GM than their N counterparts, and this is simply a matter of physics, born out of lower carrier mobility in P channel parts.
In audio design, complementary means the closest you can get in major electrical parameters, for amplifiers in particular, you want to see matching current and power handling, GM, treshold voltage and gate/drain and gate/source capacitances. Maximum voltage handling is secondary and needs to be higher than rail-to-rail voltage plus a security factor - higher will not make much of a difference. In order to get good audio from IRF/IRFP and similar parts, keep in mind that IRF/IRFPxxx and IRF/IRFP9xxx are not the complements you need - a bit wider search in the databooks will reveal a couple of better combinations, which will come VERY close or even be indistinguishable from LMOS, at a small fraction of the price. Hint: P parts will have lower Vdsmax than N parts...
 

AR1

Member
2005-03-14 7:46 pm
New York
OPS devices

In a related matter I am repairing an old Carvin FET 1000 Amplifier, I am using IRFPXX and IRFP9XX parts to replace the output stage devices which are all fried. I did not have the original manufactures part number as the OPS was apparently already replaced once and the tech obviously replaced them with IRFPXX devices.

I did notice that the Ron Spec was notably different between the IRFPXX and the IRFP9XX. Is this going to be an issue in performance. Sounds like their might be some crossover distortion issues, correct? Perhaps this is why it fried in the first place?
 
P channel devices will have higher Ron than the equivalent N channel devices (law of physics problem!).

Ron is not a very important parameter in linear class A or B amps as the output devices are operating in the linear region. Ron is normally only important if the FETs are being used in a fully-enhanced (saturated) mode, such as in Class-D amps. (and is very important in keeping losses low in high-current switch-mode regulators).
 

ilimzn

Member
2005-02-11 1:25 pm
Zagreb
Ouroboros said:

P channel devices will have higher Ron than the equivalent N channel devices (law of physics problem!).
Ron is not a very important parameter in linear class A or B amps as the output devices are operating in the linear region...

True, but the same mechanism responsible for higher Ron in P parts is also responsible for lower maximum current and lower GM, and for linear operation both certainly are a problem.
Due to this issue, not only does eg. IRFP240 have 21A Idmax compared to 12A for the IRFP9240, it also has nearly 50% higher GM. I know I would not call that complementary, yet these are invariably used as complements. Something to do with Lemmings, I suppose ;)
Lower GM on the P part not only introduces crossover distortion, but also asymetrical current limit, with resultant distortion under heavy load being the least problem - short term power dissipation issues and even a particulairly nasty form of thermal runaway (if the current limit is set close to the max for the part) can occur.
 
Dozuki said:
I personally like the 2SK1529/2SJ200 for outputs. These are Lateral types, But the negative TempoCo doesn't kick in till 6 Amps or such, so they still need a Vbe multiplier.

If it walks like a duck and quacks like a duck, chances are that it is a duck. I think you will find that the Toshiba output parts are vertical MOSFETs, not lateral.
 

Nelson Pass

The one and only
Paid Member
2001-03-29 12:38 am
AR1 said:
Sounds like I am going to have some issues if I use the IRFP240 and the IRF9240. I already bought the devices so I might as well use them and see what happens.

This is mostly dependent on the type of use. If you are using
them as Source Followers (Common-Drain) then you won't see
significant effects. If you use them as Common-Source, having
voltage as well as current gain, then you'll see some more of
any differences.

I use the 240 and 9240 as followers all the time, and they give
me no trouble at all.
 

AR1

Member
2005-03-14 7:46 pm
New York
Thanks

Nelson Pass said:


This is mostly dependent on the type of use. If you are using
them as Source Followers (Common-Drain) then you won't see
significant effects. If you use them as Common-Source, having
voltage as well as current gain, then you'll see some more of
any differences.

I use the 240 and 9240 as followers all the time, and they give
me no trouble at all.

You know, I don't even rember what the OPS topology is. I do not have the schematic in front of me at the moment but I will check it when I get home, Thanks for the tip.
 
I too have no problem AT ALL with 240/9240 in either inverting with gain or EF configuration. My low supply loss inverting design saves about 5V on -ve clipping and 4V on +ve clipping over the simple EF config, permitting lower supply voltages and C rating and lower power losses.

I wouldn't bother changing to a P340 for more loss! As far as HD2 - mine are nested in 2 feedback loops so you just don't see it.
 

AR1

Member
2005-03-14 7:46 pm
New York
Schematic

amplifierguru said:
I too have no problem AT ALL with 240/9240 in either inverting with gain or EF configuration. My low supply loss inverting design saves about 5V on -ve clipping and 4V on +ve clipping over the simple EF config, permitting lower supply voltages and C rating and lower power losses.

I bought this amp off EBay for $80. The OPS in both channels were fried. They had IRFP9240 and W33N20 as complements. I called Carvin and they said the original parts were14D20 and 20N20. So apparently someone had already repaired the OPS with little luck. The Guy who sold me the amp said it was working fine and one day he tuned it on and it just shut off. He did mention that he also bought it used. I have several of these amps; since Carvin no longer makes them I figured I cold make use out of a spare for parts if not anything else. Would be nice to get it working.

I have posted the schematic. The drivers on channel one are shot as well I have enough replacements to change them all, and I will.
 

Attachments

  • fet1000new.gif
    fet1000new.gif
    21.8 KB · Views: 578
I cannot expand your thumbnail to any magnification to see detail, but it looks like a simplified version of an Electronics Now design minus a predriver cascode and using a chip front end instead of the discrete. The original design used 240/9240 and I would hazard a guess that they would work better than the 'industrial' FETs specified. The design can achieve reasonable results depending on the GBW/quality of the front end chip.
 
amplifierguru said:
I too have no problem AT ALL with 240/9240 in either inverting with gain or EF configuration. My low supply loss inverting design saves about 5V on -ve clipping and 4V on +ve clipping over the simple EF config, permitting lower supply voltages and C rating and lower power losses.
I wouldn't bother changing to a P340 for more loss! As far as HD2 - mine are nested in 2 feedback loops so you just don't see it.

I don't think there would be outright problems with P240/P9240 - not the kind that would prevent an amp from working.

I should also mention that I tend to use one pair of FETs whenever possible, and no source resistors. And I like the idea of high linearity before feedback, and I don't like the idea of asymetric current limit - the fact of the matter is, the lower Idmax part (Pch) is the limiting one anyway.
It seems to me that using a better matching Nch part resulting in somewhat better linearity and equal current limit, is little trade-off for equal rail loss on both ends, even if it is 1V higher for the Nch part that is normally specified.

On the other hand, it would be impractical to run and buy P340s to match with P9240s, if you already have a bunch of P240 in stock - at least until you use them all up ;)