I've been having problems for a while with a DIY power amp I've been partly responsible for building.
It's a good amp so I want to get it right. Currently NOT working.
The current spate of problems started when I lost a channel. When I picked it up to take it to my desk to start investigations I noticed it had burnt my wood floor it had got so hot.
So as part of repairing the obviously blown components (some resistors) I decided to upgrade the heatsinking by fitting a big plate of 1050 aluminium suspended between the steel chassis floor and the PCB.
I bolted the transistors down to the aluminium heat sink. Turned it on, it fizzed, popped and smoked a bit. So that wasn't ideal. The fuse had blown so probably ok.
I started again with a new board and new components as I didn't want to have to keep fixing up a board which had burnt-out once already. I thought a brand new board with brand new components would solve all my problems.
Got it all reassembled with new everything except the power supply. Turned it on, same thing, pop fizz, smoke, blown fuse.
I suspect both my boards are ok.
So the difference between when it did work and now it does not work is the way I've attached the transistors to the heat sink and I'm wondering if I've made a crucial error. I'm a general engineer rather than an electronics specialist so what I don't know about electrical engineering makes quite a long list!
In the past the transistors have been bolted to the steel amplifier case with some blue sheet between the transistor and the steel. It's worked like that for a couple of years.
With the new heat sink plate suspended inside the chassis the transistors have CPU heatsink paste instead of the blue sheet I used before.
Could it be that the blue sheet was some sort of electrical insulator which was stopping the transistors shorting on the chassis or is the difference between blue sheet and CPU paste irrelevant to my problem.
Could it be the power supply which has blown? It's the only other part not changed.
It's a good amp so I want to get it right. Currently NOT working.
The current spate of problems started when I lost a channel. When I picked it up to take it to my desk to start investigations I noticed it had burnt my wood floor it had got so hot.
So as part of repairing the obviously blown components (some resistors) I decided to upgrade the heatsinking by fitting a big plate of 1050 aluminium suspended between the steel chassis floor and the PCB.
I bolted the transistors down to the aluminium heat sink. Turned it on, it fizzed, popped and smoked a bit. So that wasn't ideal. The fuse had blown so probably ok.
I started again with a new board and new components as I didn't want to have to keep fixing up a board which had burnt-out once already. I thought a brand new board with brand new components would solve all my problems.
Got it all reassembled with new everything except the power supply. Turned it on, same thing, pop fizz, smoke, blown fuse.
I suspect both my boards are ok.
So the difference between when it did work and now it does not work is the way I've attached the transistors to the heat sink and I'm wondering if I've made a crucial error. I'm a general engineer rather than an electronics specialist so what I don't know about electrical engineering makes quite a long list!
In the past the transistors have been bolted to the steel amplifier case with some blue sheet between the transistor and the steel. It's worked like that for a couple of years.
With the new heat sink plate suspended inside the chassis the transistors have CPU heatsink paste instead of the blue sheet I used before.
Could it be that the blue sheet was some sort of electrical insulator which was stopping the transistors shorting on the chassis or is the difference between blue sheet and CPU paste irrelevant to my problem.
Could it be the power supply which has blown? It's the only other part not changed.
mica sheets and blue sil pads are an absolute must in most amp designs.
failing to test for leakage to ground at the devices case or tab is critical before powering on.
and rather than risk another potential fire it would behoove you to build a "dim bulb tester"(homebrew current limiter) before your next batch of components vanish into the ether with a whiff of smoke...
with only heat sink compound between the transistors and heatsink you'll likely find arc spots on the heatsink after removing the compound.
in your previously running version where shoulder washers employed as part of the assembly?
failing to test for leakage to ground at the devices case or tab is critical before powering on.
and rather than risk another potential fire it would behoove you to build a "dim bulb tester"(homebrew current limiter) before your next batch of components vanish into the ether with a whiff of smoke...
with only heat sink compound between the transistors and heatsink you'll likely find arc spots on the heatsink after removing the compound.
in your previously running version where shoulder washers employed as part of the assembly?
Last edited:
You need an insulator with thermal compound between the transistors and the heat sink it they have a metal tab. Avoid PC thermal compounds like Thermaltake, they are electrically conductive. Wakefield makes some good compounds for your application. https://www.mouser.ca/ProductDetail...eqYBWXqKiDPSCPyB6dnCYOuVYZ8cfLZcaAtUUEALw_wcB
My PCB has nylon washers on top of the mounting posts. No part of the circuit on the PCB goes anywhere near the mounting holes if that's what you're asking?
My main question really is whether or not I've misunderstood the role of the blue sheet material for the transistor mounting supplied by the manufacturer.
I felt like I was keeping everything the same as my previously working amplifier but just upgrading the heat sinking. As part of the upgrade I changed the blue sheet to CPU paste thinking paste would provide a more uniform and better conducting bond with my new heatsink.
If the blue sheet is supposed to be some sort of insulator as well as a conductor then I'd understand what I've done wrong.
Let me know what you think.
My main question really is whether or not I've misunderstood the role of the blue sheet material for the transistor mounting supplied by the manufacturer.
I felt like I was keeping everything the same as my previously working amplifier but just upgrading the heat sinking. As part of the upgrade I changed the blue sheet to CPU paste thinking paste would provide a more uniform and better conducting bond with my new heatsink.
If the blue sheet is supposed to be some sort of insulator as well as a conductor then I'd understand what I've done wrong.
Let me know what you think.
The actual name of the part you need is thermal pad.
The color is not always blue, pink tan and white are quite popular.
Here is a link to one stocked at farnell usa https://www.newark.com/sensata-crydom/hsp-1/thermal-pad-57-1-x-44-2-x-0-076/dp/41X7820
that fits rectangular TO-247 TO-263 transistor packages. You have to measure your transistors to buy ones made to the right dimension.
If your transistors are diamonds with 2 screws on the ends instead of one, that is TO-3 package.
If using silicon rubber pads, the full name of the product "sil", you do not need heat sink compound between transistor and pad and pad and aluminum. If using mica or ceramic aluminum oxide, you do.
If there is a chance of the legs of the transistor touching the aluminum, or if on TO3 the case of the transistor the screws will touch the aluminum, then plastic shoulder washers must be installed to insulate the transistor case from the screw and the aluminum. Farnell sells a mica washer kit for TO3 that has the washers included.
Furthermore I am dubious that your aluminum sheet is going to lower temperature. Heat needs to be transmitted to the air and that requires fins.
Here is a picture of a finned heat sink with flat spot for TO3 that is not drilled. https://www.newark.com/wakefield-solutions/403k/heat-sink-natural-convection-55/dp/58F505
If your amp worked okay with the steel arangement, perhaps putting it on tall rubber feet may save your floor.
If there is air flow from bottom to top in your case, heat sinks may be enough. The case in that case needs to sit up off the floor on feet. If your case is sealed, you need to cut 2 holes in some side that is not crucial and install a fan to suck air out, and a filter and finger protection grill on the inlet. Such fans can be powered by the mains supply off the power switch. To prevent electrical noise, I recently installed an AC powered fan outside the case of an amp pulling air up over the fins. It had tall rubber feet to allow airflow from bottom to top while sitting horizontal on a shelf, but I need to install the amp vertical in an organ case, which destroyed the natural bottom to top airflow.
BTW you didn't list a location, so I linked to farnell that has warehouses in about 50 countries. Digikey & mouser are other US distributors. RS & Reichelt operate in Europe.
The color is not always blue, pink tan and white are quite popular.
Here is a link to one stocked at farnell usa https://www.newark.com/sensata-crydom/hsp-1/thermal-pad-57-1-x-44-2-x-0-076/dp/41X7820
that fits rectangular TO-247 TO-263 transistor packages. You have to measure your transistors to buy ones made to the right dimension.
If your transistors are diamonds with 2 screws on the ends instead of one, that is TO-3 package.
If using silicon rubber pads, the full name of the product "sil", you do not need heat sink compound between transistor and pad and pad and aluminum. If using mica or ceramic aluminum oxide, you do.
If there is a chance of the legs of the transistor touching the aluminum, or if on TO3 the case of the transistor the screws will touch the aluminum, then plastic shoulder washers must be installed to insulate the transistor case from the screw and the aluminum. Farnell sells a mica washer kit for TO3 that has the washers included.
Furthermore I am dubious that your aluminum sheet is going to lower temperature. Heat needs to be transmitted to the air and that requires fins.
Here is a picture of a finned heat sink with flat spot for TO3 that is not drilled. https://www.newark.com/wakefield-solutions/403k/heat-sink-natural-convection-55/dp/58F505
If your amp worked okay with the steel arangement, perhaps putting it on tall rubber feet may save your floor.
If there is air flow from bottom to top in your case, heat sinks may be enough. The case in that case needs to sit up off the floor on feet. If your case is sealed, you need to cut 2 holes in some side that is not crucial and install a fan to suck air out, and a filter and finger protection grill on the inlet. Such fans can be powered by the mains supply off the power switch. To prevent electrical noise, I recently installed an AC powered fan outside the case of an amp pulling air up over the fins. It had tall rubber feet to allow airflow from bottom to top while sitting horizontal on a shelf, but I need to install the amp vertical in an organ case, which destroyed the natural bottom to top airflow.
BTW you didn't list a location, so I linked to farnell that has warehouses in about 50 countries. Digikey & mouser are other US distributors. RS & Reichelt operate in Europe.
Last edited:
well it would be good to know what the amp design is,what type of output device, topology,etc...
i once spent an infuriating amount of time and money rebuilding a BGW 750 because of a faulty transistor socket, the long a short of that story is that it taught me the importance of isolation between output device and chassis ground (which is where your heatsink should live) has a devastating effect...i also killed a bottle of scotch working on an amp that in the end had heatsink compound that had been contaminated with metal fillings and the resulting high impedance "short" only appeared when all the output device screws where tightened(small amount was creeping past a shoulder washer to form a collector to ground short!!
i once spent an infuriating amount of time and money rebuilding a BGW 750 because of a faulty transistor socket, the long a short of that story is that it taught me the importance of isolation between output device and chassis ground (which is where your heatsink should live) has a devastating effect...i also killed a bottle of scotch working on an amp that in the end had heatsink compound that had been contaminated with metal fillings and the resulting high impedance "short" only appeared when all the output device screws where tightened(small amount was creeping past a shoulder washer to form a collector to ground short!!
Here's some images to hopefully clarify.
Be tame with any criticisms as this amp has been working very well for years, bear that in mind before any sharp intakes of breath and exclamations over all the wrong things. Hahaha.
Not all amps use external fins for heat dissipation. Some use the technique I've gone with here which is why I chose it. It suits my case. The transistors had been attached to the floor of the case which is only a thin steel pressing.
The transistors as they are currently with CPU heatsink paste
These are the unused sheets supplied with the transistors - possibly my issue?
All parts in place with CPU paste for the transistors
The heatsink
Be tame with any criticisms as this amp has been working very well for years, bear that in mind before any sharp intakes of breath and exclamations over all the wrong things. Hahaha.
Not all amps use external fins for heat dissipation. Some use the technique I've gone with here which is why I chose it. It suits my case. The transistors had been attached to the floor of the case which is only a thin steel pressing.
The transistors as they are currently with CPU heatsink paste
These are the unused sheets supplied with the transistors - possibly my issue?
All parts in place with CPU paste for the transistors
The heatsink
Those pads with the holes in them in your 3rd picture appear to be just the correct thing. Thermal insulating pads.
The sanken MT200 package is going obsolete and it would be difficult to buy the pads from somewhere else. The case allows screws to touch the plastic top without touching the metal flange on the bottom, so insulating washers may not be required. Care must be taken to avoid sliding the transistor sideways and touching the metal flange of the transistor to the screw which is grounded to the metal case or heat sink.
I think tall rubber or wood dowel feet would solve your whole problem of burning the floor.
The sanken MT200 package is going obsolete and it would be difficult to buy the pads from somewhere else. The case allows screws to touch the plastic top without touching the metal flange on the bottom, so insulating washers may not be required. Care must be taken to avoid sliding the transistor sideways and touching the metal flange of the transistor to the screw which is grounded to the metal case or heat sink.
I think tall rubber or wood dowel feet would solve your whole problem of burning the floor.
Last edited:
It already has rubber feet but not tall enough. The heat coming from it must have been pretty strong to burn the floor at a distance without direct contact.
What I really must discover is whether or not what I've done with CPU paste is potentially causing the fuses to blow. Have I created some kind of short circuit by using paste instead of the insulating sheet?
Also, could my power supply be causing the problem instead? It's the only part not changed, can they short-circuit underneath all the wrapping (bits melting into other bits ect?) or is this really unlikely / never happens?
What I really must discover is whether or not what I've done with CPU paste is potentially causing the fuses to blow. Have I created some kind of short circuit by using paste instead of the insulating sheet?
Also, could my power supply be causing the problem instead? It's the only part not changed, can they short-circuit underneath all the wrapping (bits melting into other bits ect?) or is this really unlikely / never happens?
boy would i ever love to have an thremal image of this amp in operation...heat rises...no?
and Indy is quite correct that without fins or the dividing down of the thermal mass you got squat!!
if only a 90 degree bend where put in place between the pcb plane and the orientation of the output devices your "aluminum plate" would allow "backside radiation" of heat.
if been a long term fan of Bryston amps (a little Canadian pride) as i think they are the schiznitz of passive cooled amp designs.
why does this case look so familiar...??(looks to be something Crest or Crown!?)
and by the way your construction is not frightening, at least you have a case, which is more then some who come seeking answer's to problems....
and Indy is quite correct that without fins or the dividing down of the thermal mass you got squat!!
if only a 90 degree bend where put in place between the pcb plane and the orientation of the output devices your "aluminum plate" would allow "backside radiation" of heat.
if been a long term fan of Bryston amps (a little Canadian pride) as i think they are the schiznitz of passive cooled amp designs.
why does this case look so familiar...??(looks to be something Crest or Crown!?)
and by the way your construction is not frightening, at least you have a case, which is more then some who come seeking answer's to problems....
Are we saying because of heat or because of shorting?
When I originally built this a few years back and knew even less I didn't heatsink the transistors at all. It worked for around four hours and didn't blow the fuse. It blew the components on the board. Transistors flapping in the wind (inside the case)
Shorting. The case of almost all transistors is one of the terminals, usually the collector. Take a ohmmeter to the collector pin and test the resistance to the case of the transistor. It will likely read 0. I'd suggest after mounting the transistors, verify the the case of the transistors is not shorted to the box by either the screws holding the transistors or the mica insulator was not installed properly.
i would say that depending on the input impedance of the meter i'm using anything lower then 10 Meg ohms is to be viewed with suspicion...pending clarification of the exact output device....
four hours without a heatsink...? what devices where these? and can you supply an ambient temperature reading?
and yes, shorts are the question but shorts of high impedance are dumfounding arent they?
four hours without a heatsink...? what devices where these? and can you supply an ambient temperature reading?
and yes, shorts are the question but shorts of high impedance are dumfounding arent they?
Last edited:
sorry after a scotch and some contemplation it occurs to me that the knowledge of S.O.A (or soar) safe operating area region is the principle that real world transistors have to live by...
so in this case lack of isolation due to improper mounting procedure and no apparent work on thermal dissipation characteristics are the potentials for failure not obvious...sorry...
so in this case lack of isolation due to improper mounting procedure and no apparent work on thermal dissipation characteristics are the potentials for failure not obvious...sorry...