Lateral Mosfets

[hylaphone]

Hi folks --
I have a Marshall 3540 here with a shorted 2SK134 --- Looking at BUZ901/BUZ906 as replacements, is it ok to just replace the complimentary pair or do I need a whole new set of 4 Mosfets?
Here is the schematic:

[MarkusBass]

They are different types and you are trying to save $20 at a risk of burning the amp again. I suggest that you replace all 4 transistors and adjust the quiescent current if required.

Mark

[hylaphone]

No, I would simply like to understand the protocol as I've not worked on a Mosfet amp before.
Would an original Hitachi 2SK134 drop in without replacing the others?

[J M Fahey]

Which model do you have?
Schematic (and probably same basic PCB) applies to 4 combinations: mono or stereo 200W/ch , mono or stereo 100W/ch .

200W/ch has 4 transistors, 100W/ch has 2 , notice an output pair and associated components are "not used/NF (not fitted)" in the 200W/ch version.

IF you have a 100W board (2 Mos total, an N and a P type) , you can replace just one, because they are in series and bias current is the same for both by definition, besides they are both very similar MosFets.

Now if you have the 4 MosFet version, one dead means replacing at least the one in parallel with it because the real problem with power devices is equal current sharing.

Probably not back in the day, but modern BUZ___ , 2SK/J___ tend to be the same, for the very good reason that:
1) Hitachi stopped making them long ago
2) it seems to be the main manufacturer today (if not the only) is EXICON/PLC/Profusion/whatever , and they plain make "dies" only characterized by parameters (such as 200V 10A or whatever) , lesser factories buy the chips, encapsulate and label them.
They have their own house brand, of course.
In fact, I guess even if you bought a modern 2SK134, I wouldn't be surprised to find it's the same (inside) as an equivalent BUZ

That said, if you can get all 4, all together from the same batch, it's added peace of mind and the best , safest way to solve this.
I'd pull the still working original transistors and keep them as spares, you never know when an urgent job or a battered old amp comes for repair at an inconvenient time or in need of being repaired at a minimum cost.

[hylaphone]

Thanks JM for the excellent info
This is the 200W/ch version, each side has the two pairs.

The shorted 2SK134 took out that channel's rectifier
R34 source resistor was cooked on *both* channels - but diode checking only reveals the one shorted MOSFET - I'm not sure what that suggests..
I've replaced those .33ohm source resistors, and both rectifiers for good measure.

[J M Fahey]

Mmmmmhhhh, if it toasted so much, then I'd replace all 4 .

When amps short and the proper rated fuse blows instantly, there is damage, of course, but is not widespread , while when user puts a larger one until *something else* inside the amp does its job by blowing itself, parts suffer, specially semiconductors.

When I feel particularly pissed off at some moron who does such things, I feel tempted to wire one of these in series with his fuse, typically need around 2A for triggering so in 220V Countries fit a lot of amplifiers:

the real stuff, of course

[oc disorder]

Here's some info re

EXICON/PLC/Profusion/whatever

ECF10N20 $9.57USD Nch 200V 8A 125W TO3
ECF10P20 Pch -200V -8A 125W TO3

ECF20N20 $17.70USD Nch 200V 16A 250W TO3
ECF20P20 Pch -200V -16A 250W TO3


EXICON

[gbono]

Mmmmmhhhh, if it toasted so much, then I'd replace all 4 .

When amps short and the proper rated fuse blows instantly, there is damage, of course, but is not widespread , while when user puts a larger one until *something else* inside the amp does its job by blowing itself, parts suffer, specially semiconductors.

When I feel particularly pissed off at some moron who does such things, I feel tempted to wire one of these in series with his fuse, typically need around 2A for triggering so in 220V Countries fit a lot of amplifiers:

the real stuff, of course

I suggest a shock tube - can be rigged so the safety pin is removed when they try to pull the chassis from the enclosure and the st reliably initiates C-4

[Enzo]

Well, OK, as long as you use the Deluxe DEtonator and not the ordinary one. The REAL crimp instead of a fake crimp sells it for me.

[gbono]

Well, OK, as long as you use the Deluxe DEtonator and not the ordinary one. The REAL crimp instead of a fake crimp sells it for me.

also... "All detonators are completely inert and do not contain any hazardous materials." Inert LLC

[J M Fahey]

I'd LOVE to see what an un-real crimp looks like then

[hylaphone]

Ok, leaning towards BUZ900/905 as replacements, they appear closer in spec - keeping the two channels similar.
I'll replace the two pairs.

Never having done this - and without a bias pot - how would one adjust idle current in this amp?

Lastly - I mentioned R34 being cooked in *both* channels. Does that suggest another failure besides the one shorted MOSFET?
Admittedly I am much more comfortable with tube amps

[Enzo]

OK then, imagine you are working on two Fender twin reverb amps. And on each, the screen resistor was burnt on the second power tube from the left. Does that suggest anything other than the same position 6L6 coincidently failed in each amp?

[hylaphone]

OK then, imagine you are working on two Fender twin reverb amps. And on each, the screen resistor was burnt on the second power tube from the left. Does that suggest anything other than the same position 6L6 coincidently failed in each amp?

Yikes. Other than someone running them both hard into mismatched loads, taking out the 6L6, taking out the screen resistors... I'm drawing a blank.

[Enzo]

The point was, if you are familiar with tube systems, you probably would not think that there was something about Twin reverbs that cause the second tube to fail consistently. It wasn't about why a tube failed, it was about why THAT tube in each, it could be any one of them. There is no reason to think any differently about the transistor amp. it could just as easily been one of the other three transistors that failed in each case. So R34 is just a coincidence.