Toshiba 2sk370 and 2sj108

[Zen Mod]

probably jinxed

discus hernia ; massive- 3 pcs

mostly living without pain killers

[Charles Hansen]

Originally posted by korben69
Nice to read your input Mr HANSEN, many thank's for it.
You're right we were in trouble for nothing.
BTW can your remark about 2SJ108/2SJ74 and 2SK370/2SK170 be applied to 2SJ109-BL and 2SK389-BL ?

Originally posted by jacco vermeulen
Korben,

look up the datasheets of the 2SK170/2SK370/2SK389.
Parameters of the 2SK170 and 2SK370 are identical, except dissipation.
Nearly all parameters of the 2SK389 are different=different device

Actually the 2SJ109 is almost exactly the same as a dual monolithic version of the 2SJ74, and the 2SJ389 is almost exactly the same as a dual monolithic version of the 2SK170. The differences noted on the datasheets are very minor and can be ignored for the most part. For example, I think the P-channel dual monolithic is rated at 30 volts instead of 25 volts for the single part.

In the real world, there are some things to watch out for.

The most obvious thing is that the discrete parts will need to be matched by hand. This is not a problem for us as a manufacturer, as we purchase thousands at a time.

But the weird thing are the dual monolithic parts. There is a 7th pin connected to the substrate. If you try to ohm out the part and deduce the internal construction, it makes no sense. I have even asked FET designers, and they can't make sense of it either. The practical implication of the strange substrate construction and/or connection is that the dual monolithic parts seem to be very sensitive to static discharge. (Normally JFET's are very resistant to damage from static discharge.)

If you use them in the input circuit where plugging in a cable can introduce a static shock, they can fail. I have tried floating the substrate, tying it to ground, and tying it to the power supply rail. Nothing seems to help. So now I only use the dual monolithic parts in the "interior" parts of a circuit where they are not subjected to static discharges. (When the end-of-life was announced, we purchased something like 30,000 of each polarity! I figure we have a 15 or 20 year supply. Hopefully someone will make a replacement before then. If not, we can always just match the K170/J74 parts.)

[Charles Hansen]

Originally posted by anatech
It sounds like the Hyperbaric Oxygen Therapy would have helped. Oh well.

You may want to investigate further. There are many documented cases of brain damage (for example) being at least partially repaired even several *decades* after the injury.

[Charles Hansen]

Originally posted by Charles Hansen
I escaped that fate, but my posts must be approved by a moderator before they are published.

I have just been informed that my posts are no longer being moderated. Thank you, Variac!

A nice side benefit is that the e-mail notification of replies to a thread have been restored.

[Charles Hansen]

Originally posted by jacco vermeulen
The new thermaltrack design is overwhelming, btw.

If you are talking about our new amplifier, the MX-R, that uses the ThermalTrak parts, thank you for the kind comments. We are very proud of that design.

If you are talking about the parts themselves, it turns out that they are very tricky to use. The example circuit in the app note is of little use. If you got it to work on one unit (by sheer luck), it wouldn't work on multiple samples due to tolerances in the parts.

Several people have tried to use them. My understanding is that McIntosh is using them, but in a rather odd way (at least to my way of thinking). They use the diode as a temperature sensor, but this signal is sent to a microprocessor that calculates the desired bias and applies it via a complex circuit. I don't like adding microprocessors, as the clock adds RFI to the circuit. We came up with a different approach, but it took several months of experimentation.

[anatech]

Hi Charles,

They use the diode as a temperature sensor, but this signal is sent to a microprocessor that calculates the desired bias and applies it via a complex circuit.

I used to do warranty service for McIntosh in the 90's, and back then they seemed to stick to well designed audio circuits. It seems like Clarion's influence is being felt. Of course they may be searching for mis-adjustment proof products as the average quality of technician continues to spiral downwards.

I do agree with you that microprocessors have no place in an amplifier circuit. M.L. take note.

For On Semi devices, I've been very impressed by the MJW0281A and MJW0302A parts. The ThermalTrak diode may be best used in a standard VBE multiplier.

I must apologize, I'm not familiar with your newer products - yet.

-Chris

[Charles Hansen]

Originally posted by anatech
For On Semi devices, I've been very impressed by the MJW0281A and MJW0302A parts. The ThermalTrak diode may be best used in a standard VBE multiplier.

Those parts are the same die in a smaller physical package (TO-247) instead of the bigger MJL3281 and MJL1302 (TO-264). Unless you have radical space limitations, just use the bigger parts for a larger SOA.

Change the leading "M" to an "N" and you have the ThermalTrak parts. You can try a Vbe mulitplier. I try to avoid them when I can just on general principles, as they contain a (very short!) feedback loop.

There are lots of ways to skin this cat, but the important thing is to realize there are *many* sources of thermal drift, and each one must be compensated properly. A Vbe mulitplier tends to throw a blanket approach at two or three or four different problems, each with different time constants and thermal coeffiecients. Not the best, but usually workable....

[anatech]

Hi Charles,
I did test some samples and found those to have tighter matching of all specs. I know they are they same die. Could they have been selected?

I have some ThermalTrak parts which I've tested but have yet to use any in a circuit. Their pinout is one reason, can't just swap them in and out of working prototype boards.

As far as SOA is concerned, I'm not worried. About the highest power amplifier I'm likely to build would be 200W or less. Mostly 100W or less really. If I'm that close to the SOA, I'd add a couple devices. I understand where you are coming from though. You design production amplifiers. I experiment.

A Vbe mulitplier tends to throw a blanket approach at two or three or four different problems, each with different time constants and thermal coeffiecients. Not the best, but usually workable....

Yes. You are correct. As I spend more time around here I tend to pick up little tricks. So possibly I'll learn sometime how to finesse this problem. For now I'm happy to have good sounding circuits without craters in the middle of the parts. You would have my full and undivided attention when talking about these things.

-Chris

[Charles Hansen]

Originally posted by anatech
I did test some samples and found those to have tighter matching of all specs. I know they are they same die. Could they have been selected?

The guy that designed the ThermalTrak parts (Mark Busier) improved the process for the power transistor dies with the result of much tighter matching. This improved process showed up first with the smaller parts you tried, and then the ThermalTrak parts. But all of the parts are now built with the improved process, for at least the last year or so. So you should see the tighter matching on all of these OnSemi parts.

[luvdunhill]

Mr. Hansen:

Glad to see you around here! I don't mean to "put you back to work" so soon, but do you by any chance know of a suitable replacement for the RFG60P06 and IRCP054 devices? I am under the impression that such devices are not really available these days and these were items of a bygone era, but when I say that you popped up again, figured I'd ask