Marshall Valvestate: Attaching MOSFET to Heat Sink

[guitarkitbuilder]

I'm replacing a damaged 2SK1058 MOSFET on a Marshall Valvestate 8200 but I'm not sure of what type of heatsink coating they use to keep the MOSFET electrically isolated.  Does anyone know?  It's a yellowish-tan in color but I can't tell for sure if it was a film of some kind, or what.  The area behind the removed MOSFET is damaged a little so I'm leery of reinstalling without also fixing the surface of the heatsink.

[joecool85]

I'm replacing a damaged 2SK1058 MOSFET on a Marshall Valvestate 8200 but I'm not sure of what type of heatsink coating they use to keep the MOSFET electrically isolated.  Does anyone know?  It's a yellowish-tan in color but I can't tell for sure if it was a film of some kind, or what.  The area behind the removed MOSFET is damaged a little so I'm leery of reinstalling without also fixing the surface of the heatsink.

You don't have to use whatever they do.  Just clean up the heatsink (carefully use a razor blade like they use in auto shops) and then use regular thermal paste along with mica/ceramic insulator kit.

[guitarkitbuilder]

Thanks Joe, yes I agree that the mica insulator with thermal compound will work, but I do this as a hobby and like to restore everything to OEM condition whenever I can.  I just got my answer this morning.  I spoke with Marshall's Spares Department and found out that they used a rubberized thermal conductor/ electrical insulator strip, which after years of use may appear paper-like when peeled back from the heat sink. The replacement part number is THRM-00006. They also confirmed that a mica insulator with thermal compound would be a suitable replacement.

[bry melvin]

few amps use mica+paste anymore. Mouser sells insulators for all types of transistor/ic packages that would replace the plastic type ones.

The one DON't on thermal paste is NEVER use some of the silver paste that computer fanboys are fond of. Most of these are actually electrically conductive and a real pain to clean off. I've fixed a few amps after someone has done THAT upgrade, literally having to scrub the circuit board and heat sink with solvents repeatedly.

Also sith SS amps no one is seiously concerned about keeping everything "original" in fact most of the transistors in older amps aren't even available in original form as everything is moving to lead free versions.

[guitarkitbuilder]

Thanks Bry, that's right, the Kapton type of insulator has replaced Mica in many applications but I haven't seen it on Mouser or elsewhere for the TO-3P package used by the MOSFETS I'm replacing.  

As for keeping it OEM, I may be out of step with the crowd, but when I restore vintage tube amps I go for authenticity and I don't see why I'd treat my solid state ones any differently.  As a past circuit designer myself, I also know that if I deviate from the manufacturer's build method, I may unknowingly change something, and while it may work initially, it may not be as robust as the original design.  I'm particularly careful with thermal issues on power devices.  I don't want the amp to work on the bench, but then fail in a gig under load.

[J M Fahey]

Good news then, Mica+Paste *is* better than Silpads (rubber) or Kapton+paste.
Big manufacturers switched to a knowingly inferior products to avoid a tight bottleneck in otherwise robotized assembly.
Rubber has the added disadvantage that it has a definite lifespan (around 10 years) while grease+zinc oxide has a much longer one, and Mica is, well, a rock millions of years old.
As an example, space suits used in Moon expeditions (only 40 years ago)  are absolutely rotten at the rubber joints, although they used *real* "Nasa grade" rubber (who else would?) and are kept under controlled atmospheric conditions at the Smithsonian, no less, so ....

[guitarkitbuilder]

OK I think I'm detecting some sarcasm.  But riddle me this Batman - since Mica + thermal compound pre-dates the rubber solutions, why did Marshall choose to use rubber on this amplifier when both solutions were available?  It's hard to get cheaper than the Mica+ solution so what would explain it?  I'd be happy if you do know, and would enlighten me, but if I still don't know, then I think it's foolish to make a change out of ignorance, or the assumption that manufacturers switch to a knowingly inferior product.  On top of that, these Valvestates have a reputation for failure, and for never being reliable once repaired.  Is that because people are not using the OEM parts/methods?  I don't know, but why introduce another unknown into the equation when I don't have to?

That said, thanks for an entertaining response.

[J M Fahey]

But riddle me this Batman - since Mica + thermal compound pre-dates the rubber solutions, why did Marshall choose to use rubber on this amplifier when both solutions were available?

Simple: a rubber is "applied" (just placed there) in 5 seconds by hand, probably 1 or 2 seconds by machine, you just tighten the screw/nut , presto !!
To boot, it's "clean".
A mica must be greased evenly on both sides, after which it becomes a nightmare to handle (relatively speaking), you just can't leave it anywhere for a split second , must be applied either to the transistor or heatsink at once.
How long does it take? ... 15 / 30 seconds? Per transistor?
That's a deal breaker in an automated factory.
Did I mention it's messy?
Parts cost is about the same; workforce cost (and time) is not, absolutely.
PS: a mica is reusable after a repair, its dimensions don't change; a Silpad is forbidden to do so "because it compresses unevenly and has "memory" of that " (just read the Silpad Datasheet)
To talk less and show more, let's refer to original data, in this case by SilPad maker and Motorola/On Semiconductor, today called Freescale Semiconductor (joys of Globalization).
1) There are many SilPad "rubbers", many of them very good, in fact as good as mica+grease.
Anyway they are very expensive; 99% of makers use Silpad Grey (Silpad 400).
I have *never* seen anybody using gold, pink, etc. (although maybe NASA does)


And, how well does it transmit heat?
In this case, *smaller* ºC/W means *better* transmission.
Not surprisingly, Silpad Grey (400), being the cheapest, is also worst.
In a TO220 size, it is around 5ºC/W.
There are others much better, close to 1.5ºC/W, although "nobody" uses them.
The *best* (can't even imagine the price) reaches 1.23ºC/W , although that rating is impossible in the normal, gray one.


And what rating does old, messy, humble mica+grease reach?:

1.6 ºC/W.
So, the unqualified statement "Silpad beats Mica/Grease" must be qualified with: "*some* (very expensive/not available from regular suppliers) Silpad beats cheap mica/grease; *regular* (grey) Silpad does not by a wide margin (3:1 in favor of Mica/grease)"

Back to Marshall: insulator is not the only block in the thermal path, you hav also the chip internal resistance, the heat sink one, plus any extra losses for improper position.
In fact, it has been measured that a Silpad insulated transistor sits at 80ºC while the same with mica/grease sits at 70ºC; the reduced life expectancy may be compensated by the much lower production cost.
If it makes it beyond the first (factory guarantee) year, for Marshall it's the same whether the amp lives 3 or 6 years.

[guitarkitbuilder]

JM, thanks for taking the time and trouble to post this information, it is really very helpful to me.  It's great to have real data.  Do you know for sure that this material used by Marshall is Sil-Pad? 

From a manufacturing perspective, I would guess that using a rubberized product vs mica/lube was as much ease and convenience as cost.  From reading the Berguist site, that appears to have been the motivation for inventing the product.  In my case however it's moot since I haven't found a source for the original material.  Marshall referred me to US Music Corp. for parts, but they say that they're transitioning as the US parts supplier from Korg, and the transition is happening over the next few months.

Also, the MOSFETS I'm replacing use a TO-3P package, which looks like a TO-220 but is a bit larger.  It looks nothing like a TO-3.  I ordered an assortment of mica insulators yesterday from Mouser, but they didn't carry one for a TO-3P.  I could either go with a TO-220 mica and have some overhang, or use another size that may be large enough but have a cutout or two in the wrong place.  Any suggestions on that front?

[J M Fahey]

What was originally referred to as "TO3P" was an early (misguided) marketing plot to "sell" the plastic case idea to a world used to metallic TO3.
For compatibility, the earlier ones had a rounded top to allow use of then standard "oval" TO3 micas.
The TO3 part comes from the fact (check it) that if you bend down (at 90º from the body) base and emitter legs and cut the collector leg short, you *can* mount it on a TO3 (metallic) heat sink or board.
I have been making amplifiers for 40 years now, and used TO3 2N3055 for about 8000 amplifiers.
When they come for repair today I fit plastic ones in the same heatsink; obviously with only 1 screw.
The modern TO3P cases are rectangular and carry their own designation (at last)
You must order micas for TO218 or TO247.
The size is the same, but TO218 requires an extra insulating nipple for the screw (order a few just in case) ; TO247 has it "built in"; mica for both is the same.
Order a few screws and nuts at the same time, to avoid wasting time with a local hardware store clerk who does not know what you're talking about.