AB amplifier heatsink on -V rail by design? Marshall MG250DFX

[sixstringartist]

Here is a shot of the heatsink and the relevant schematics: https://imgur.com/a/grp2K

The quick synopsis is that -V is connected to pin 8 on the TDA7293 DMOS amplifier IC, which is also connected to the back plane. That is then screwed into a heatsink which is screwed into the case. The only thing preventing the heatsink from shorting out to the case is a set of small plastic insulators around the screws w/ plastic caps to hide the screw heads.

I didnt notice this initially and ended up shorting the heatsink to ground by not having the back panel screw insulators tightly in place. How common is this? I was totally surprised and blown away that a large heatsink in the case could be energized and additionally the screw heads!

[jpc]

Haven't touched an amp for ages but it used to be very common, especially if you had split power rails.

[sixstringartist]

What was the reasoning? Seems more likely to get shorted on things.

[b_force]

Well, the proper way to do this, is to use a thermal pad to isolate this.
I guess they found that a little bit to expensive and rather safe a few cents.

[Armadillo]

the case is a set of small plastic insulators around the screws w/ plastic caps to hide the screw heads.

Relax!, its part of the journey. You learnt and the next time you see that thermal insulation plastic sleeve, you learnt to respect it and will take a meter to check for no continuity or shorts AND run through your own list of Pre turn-on Check-routine before turning it on.

The plastic sleeve is with a flange end and the screw is supposed to compressed onto the flange end. Maybe you should picture how it "Hide the screw heads"?

[capt bullshot]

I didnt notice this initially and ended up shorting the heatsink to ground by not having the back panel screw insulators tightly in place. How common is this? I was totally surprised and blown away that a large heatsink in the case could be energized and additionally the screw heads!

That style was quite common in the days then. Todays style would be to isolate the Amplifier from the heatsink, making the build process of the unit easier. Back then, electrically isolating the IC from the heatsink was a rather cumbersome process involving mica sheet and thermal grease, using isolated screws was considered more elegant. Some of these designers never arrived in todays world and might still go this way.

[langwadt]

I didnt notice this initially and ended up shorting the heatsink to ground by not having the back panel screw insulators tightly in place. How common is this? I was totally surprised and blown away that a large heatsink in the case could be energized and additionally the screw heads!

That style was quite common in the days then. Todays style would be to isolate the Amplifier from the heatsink, making the build process of the unit easier. Back then, electrically isolating the IC from the heatsink was a rather cumbersome process involving mica sheet and thermal grease, using isolated screws was considered more elegant. Some of these designers never arrived in todays world and might still go this way.

isolating the IC from the heatsink adds thermal resistance, isolating the heatsink doesn't

[voltz]

Having the whole heat sink at potential isn't very common and considered as bad practice really.
Its true the thermal connection is better but worse in terms of its too easy to short out as there's a lot of area at potential. I have seen this in other equipment where the heat sink is small, pcb mounted or not physically located near anything else. But his particular design is begging to cause problems as its just mm away from earth. It happens, just have to be on your guard! Can't think of any other Marshall amps that do this, and i've worked on quite a lot.

[Armadillo]

Having the whole heat sink at potential isn't very common and considered as bad practice really.

I don't particularly like the idea of tying down an in practice heatsinking solution to an unfounded personal fear, really. Designer could choose other means of protection. For example, protection from physical contact by separation. In this particular case, protection by the casing which separate the user from contact with the heatsink. The serviceman is supposed to be trained anyway. Manufacturers always state "Warranty void if open"!

[b_force]

I didnt notice this initially and ended up shorting the heatsink to ground by not having the back panel screw insulators tightly in place. How common is this? I was totally surprised and blown away that a large heatsink in the case could be energized and additionally the screw heads!

That style was quite common in the days then. Todays style would be to isolate the Amplifier from the heatsink, making the build process of the unit easier. Back then, electrically isolating the IC from the heatsink was a rather cumbersome process involving mica sheet and thermal grease, using isolated screws was considered more elegant. Some of these designers never arrived in todays world and might still go this way.

isolating the IC from the heatsink adds thermal resistance, isolating the heatsink doesn't

Depending which silicon therman pad you use.
But it's not an excuse. It's not difficult to design a good performing solution with a thermal pad.

[Armadillo]

I didnt notice this initially and ended up shorting the heatsink to ground by not having the back panel screw insulators tightly in place. How common is this? I was totally surprised and blown away that a large heatsink in the case could be energized and additionally the screw heads!

That style was quite common in the days then. Todays style would be to isolate the Amplifier from the heatsink, making the build process of the unit easier. Back then, electrically isolating the IC from the heatsink was a rather cumbersome process involving mica sheet and thermal grease, using isolated screws was considered more elegant. Some of these designers never arrived in todays world and might still go this way.

isolating the IC from the heatsink adds thermal resistance, isolating the heatsink doesn't

Depending which silicon therman pad you use.
But it's not an excuse. It's not difficult to design a good performing solution with a thermal pad.

Let's face the reality. They are not going for any design award or needing your appraisal, obviously more so for commercial gains.
Yes, its not difficult to design a good performing solution with thermal pad, I agree, we can say and sing anything now, but its not the business answer in the commercial world.
So they exist both with thermal pad and without in the market. We just need to be aware of it.

[Cyberdragon]

PFFT, plastic sleeves , the stuff I've seen used plastic screws.

[voltz]

Having the whole heat sink at potential isn't very common and considered as bad practice really.

I don't particularly like the idea of tying down an in practice heatsinking solution to an unfounded personal fear, really. Designer could choose other means of protection. For example, protection from physical contact by separation. In this particular case, protection by the casing which separate the user from contact with the heatsink. The serviceman is supposed to be trained anyway. Manufacturers always state "Warranty void if open"!

"Unfounded personal fear". Nope. 30 years + of repair/design experience including Marshall amps Its simply about cutting cost.

[Armadillo]

"Unfounded personal fear". Nope. 30 years + of repair/design experience including Marshall amps Its simply about cutting cost.

I would say respectably many good years there.... 

For those that repair branded equipment power supply like the lecroy for example PS1724 knows that their heatsink is tied to 330V dc bus, that is lethal voltage I would say.
Now compare this to his low voltage amplifier, I would say is peanut.

[sixstringartist]

Thanks all for the responses.

Maybe you should picture how it "Hide the screw heads"

This is actually pictured in the 5th image. The black insulation at the screw head has a lid that closes.

The serviceman is supposed to be trained anyway.

Nobody told me that! 

Really appreciate the insight so far. This amp was manufactured around 2003-2006 so its been around a bit but not that long in the grand scheme of electronics or guitar amplifiers for that matter. Marshall may have changed some of their designs since then. Anyway's I still need to fix this thing so I'll keep chugging away at it.

[David Hess]

How common is this?

It is very common or even universal with junction isolated integrated circuit processes but it only becomes an issue with power devices where the integrated circuit is bonded to the exposed metal pad which is intended to be used with a heat sink.  Because of junction isolation, the IC substrate *must* be connected to the most negative supply voltage to maintain isolation between the different devices on the integrated circuit.  (1)  Some ICs include a special pin for the substrate connection.  MOSFETs may include a substrate connection also but most connect the substrate to the source.

The above is why the pinouts of the 7805 positive regulator and 7905 negative regulator illustrated below are different.  The most negative supply voltage *must* go to the center pin which connects to the tab.

Dielectric isolated integrated circuit processes do not have this restriction but are more expensive.

(1) In theory it could be the most positive supply instead of the most negative if the N and P fabrication was reversed but offhand I do not know of any ICs made this way.  NPN transistors work better in silicon than PNP transistors and this determines the doping for the junction isolated processes.

[duak]

I could be wrong on this but I repaired an hp 6002A supply recently and I believe the pass transistors are not electrically isolated from the heat sinks.  OK because they're inside the case and there is a fan for cooling.  Makes for some unusual threaded plastic spacers, standoffs and washers to hold the heat sinks in place.  I suppose the thermal resistance is minimized but it must have cost a few bucks to engineer the insulated heat sink mounting hardware.

A note about polymer hardware such as screws and shoulder washers.  Unless they are glass filled, polymers such as nylon and teflon will creep (stretch or squeeze) and relax the clamping force on the device usually leading to its failure.  I remember in the 70s building a power amp with Motorola transistors, the (I think) approved mounting kits and the nylon shoulder washers liquified, turned brown and squeezed out within a few months.

Cheers,

[floobydust]

The IC is rated for 50W dissipation, and seeing two on that tiny heatsink with fan, well Marshall is pushing it.
They could not afford the thermal resistance of an insulator there, they would need a larger heatsink.

[voltz]

Quite some difference between lab equipment SMPSU and a stage guitar amplifier.

Yes of course smpsu designs use a non grounded heat sink at potential. Thats because its safe to do so in an earthed case, and normally used in a dry, dust free environment. But a guitar amplifier has a good chance of user abuse, damp conditions, dust and beer spillage! Along with physically dropping the unit with a good chance of breaking the plastic insulators. Heat sinks are heavier than just a chip. Plus the case is not shielded. Its wood although the voltages go up to 80v, not hundreds. Large heat sinks at AC (anything up to 20Khz in this case) will radiate EM fields. More chance of interfering with other circuitry within the amp, for example causing positive feedback causing HF oscillations. Those then have to be taken into account. If not, the whole amp could become unstable. I could go on...

Marshall do have far better designs certainly not using live heat sinks in their higher range of products. The heat sink is normally placed at the back externally or internally fan assisted. In nearly all cases that i've seen over the years, the heat sink is at ground. Even with designs using this same chip but insulted at the chip - not the heat sink. But in OP's case its not. This is purely a cost cutting exercise and i bet even the Marshall techs had to bite their lips with this one. I would have said something to management if it was me designing it - knowing full well they would have replied 'sorry, we have to'.. Maybe someone reading from Marshall UK could comment for a definitive answer.. But thats how i see it.

[Armadillo]

Gryphon amplifier should be well known expensive amplifier. At over 90kg, is huge, mega, gigantic... Just look at the custom built Heatsink... isn't it lovely.
What I want to say is, there are differences between those that must look big and those that must look sleek and compact. No investor is going to come to you if you are going to design a handphone like the size of a army 77 signal set for example. Conversely, if you are going to design a audio amplifier like the big tanky like "TRANSFORMER" look, yeah I think it will sell.
I think is a matter of industrial and circuit exposures. Those that view without thermal pad as a risk, and those that will brush it off as just nominal. 

[voltz]

Some professional amplifiers for example from Lab Gruppen which run into Kilowatts, do indeed have high potential heat sinks internally. With fan assisted cooling. The reasoning for that is, these amplifiers are very compact and light weight with SMPSU's rather than linear PSUs with a massive heavy transformers. So keeping the heatsink inside the unit rather than outside helps make the amp smaller.

The big difference is - these amplifiers cost £1000ss . The heatsinks inside are solidly bolted flat to the PCB - NOT on flimsy plastic insulators! Also, the circuit is designed fully with HF oscillations in mind. There are none!

Now in the case of those amps, i am fully in favor of the high pot heatsink design.. Best design in class in fact. So it depends on how its implemented. The OP's amp, well thats a totally different story.

I think i will end this here. Sorry for swamping the OPs thread and good luck with your amplifier.

[IanMacdonald]

Had to laugh, those screw covers are actually for use on car number plates.

Anyway, thermal pads do make quite a difference to the power ratings of tab power devices like this. Having only one bolt makes it hard to get an even pressure across a washer, and consequently it's likely that only a small area of the washer is actually conducting heat. With metal to metal that's not so important as a tiny area of contact can pass a lot of heat that way.

For any powerful instrument amp, they may be old fashioned but TINSFTO3.