Best way for mounting high voltage transistors on heat sink

[Saman]

Hello, I have a board that has 8 transistors mounted on a heat sink. each
TO-220 transistor(IRF 630) dissipates about 10W. for lower voltages I
usually used rubber silicon pads for isolation. I have tested them and they
can dissipate up to 15W without making transistor too hot. Now I have a high
voltage board. the voltage of metal flange(drain) of each transistor is
100V. I haven't worked with such high voltage yet, so I'm worry about
safety. The whole heat sink assembly is inside a metal box that has earth
wire and also both live and null of input 220V line are fused. Still I'm not
sure it is enough. your experience is appreciated. thanks

 

[Vladimir Vassilevsky]

Hello, I have a board that has 8 transistors mounted on a heat sink.
each TO-220 transistor(IRF 630) dissipates about 10W. for lower voltages
I usually used rubber silicon pads for isolation. I have tested them and
they can dissipate up to 15W without making transistor too hot. Now I
have a high voltage board. the voltage of metal flange(drain) of each
transistor is 100V. I haven't worked with such high voltage yet, so I'm
worry about safety. The whole heat sink assembly is inside a metal box
that has earth wire and also both live and null of input 220V line are
fused. Still I'm not sure it is enough. your experience is appreciated.

Nothing to worry about. The 100V and 15W is not much for TO-220 with the
rubber silicon pad.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

 

[Saman]

Thanks Bill,

How should I isolate mounting bolts? I have bought an epoxy that the
retailer says it can resist up to 300 degree centigrade but I'm not sure if
I spray an epoxy layer on top of transistor then it can exchange heat,
because though most of heat are exchanged through silicon pads but some heat
are also exchanged through top of TO-220 body. another problem that I have
encountered in past using these pads is that when I tighten the screw,
sometimes it causes sharp edges of TO-220 to cut the pad and make a short
circuit with heat sink. Thermal conductivity of these pads are directly
related to the pressure on them. I am worry that after a while and because
of different mechanical stresses even if I do not tighten the screw too much
it cuts the pad. One of the hardest part of my job is power dissipation
because I have a very limited space(a fixed size box that is too small to
accommodate big heat sinks. so I have to use every cubic cm of box space.
Are these silicon pads the favorite choice? I mean isn't there better
alternatives?

regards,

Hello, I have a board that has 8 transistors mounted on a heat sink. each
TO-220 transistor(IRF 630) dissipates about 10W. for lower voltages I
usually used rubber silicon pads for isolation. I have tested them and
they
can dissipate up to 15W without making transistor too hot. Now I have a
high
voltage board. the voltage of metal flange(drain) of each transistor is
100V. I haven't worked with such high voltage yet, so I'm worry about
safety. The whole heat sink assembly is inside a metal box that has earth
wire and also both live and null of input 220V line are fused. Still I'm
not
sure it is enough. your experience is appreciated. thanks

Check the voltage rating on your insulating silicone rubber mounting
pads - most of them seem to be good for a couple of kilovolts, so 100V
shouldn't be a problem.

Make sure that you've also got proper insulators on the mounting bolts
that hold your transistors onto the heatsink.

 

[Jamie]

100 is not really high voltage. The silicone pads are rotten
thermally, but will not be bothered by 100 volts.

John

I like the mica sheets, not the plastics.


http://webpages.charter.net/jamie_5"

 

[Saman]

Have you compared thermal conductivity of mica and silicone pads? Once I did
a small experiment and I noticed it is practically impossible to use these
pads over 20W dissipation(If I remember correctly in my experiment and by
direct mounting of transistors on heat sink and applying silicone paste it
could dissipate happily up to 40W ). I never compared mica sheets with these
pads.

 

[Vladimir Vassilevsky]

How should I isolate mounting bolts?

Do not put any screws through the mounting hole of TO-220. Use the
spring loaded mounting clamp over the TO-220 body.

Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

 

[Saman]

Thanks, good idea! Why I didn't think about that!
But how to isolate top side of flange?

 

[Saman]

I use nylon shoulder washers a lot but as you pointed out the screw is not
isolated and I don't know is it safe to leave a bolt at 100V inside the case
non-isolated.
nylon screws as you said should stand high temperature. I think only Teflon
can stand these temperatures safely.
Clips are also very good idea, thanks but I have to design them and order
them because I don't think I can find anything ready that fits my heat sink.
as I indicated before because of high density and low space I'm very limited
at choosing heat sinks and I found a suitable small heat sink that a 8x8
fan can be mounted on it and has very good dissipation.

PS: my lost post was in response to Vladimir post that mistakenly posted in
response of myself, sorry.

 

[Saman]

There's two issues: with higher voltage, may come higher power (and
you might want more power dissipated). Also, this is above shock-
hazard
threshold (but if it's inside a grounded box, that's already
handled).

yes because of higher power I used 8 parallel MOSFETs.
So you are suggesting having a live conductor inside a grounded box is Ok?

Easiest would be to use thermal compound and direct bolt-to-heatsink
(so the heatsink is electrically live, but it's inside a grounded box,
so you're OK there). The only insulation problem is how to connect
the
heatsink to the box (lots of solutions, at real HV we used a lot
of ceramic standoff posts).

I'm not sure about saftey but if I could do such a thing it would save me
from
a lot of troubles when trying to accomodate big heat sinks inside a small
box.
Because as I said earlier if I could directly mount transistors on heatsinks
then
each transistor could dissipate considerably more and I could use fewer
transistors.
Another problem with this scheme is when I have 8 parallel transistors
I either have to use a separate heatsink for each one or isolate them.

There's IRF630 variants that have factory insulation in the package,
the IRFI630G costs circa $1.53 in a fully isolated 'fullpak' package,
as opposed to IRF630N at $1.05, and the cost of mounting washers
and insulators makes the apply-your-own-insulation route
more expensive.

Thanks I will investigate about it.

It's also convenient, sometimes, to put a bunch of transistors
onto a small heatsink (like an aluminum L section) that fits onto
the circuit board, then bolt the small L section to a big heatsink
when the board is mounted in the box. The L section can be
electrically live, it's relatively easy to put a layer of Kapton tape
on the flat mating surface and fasten the big sink with nylon screws
or other nonconducting fasteners.

good idea, I have seen some L shape heat sinks on PCBs but I do not
know how much they can dissipate. I remember once (about a year ago)
I made some measurements about power dissipation of different heat sinks
and I came to this conclusion that for anything above 10W I need either a
huge heat sink(and for 20W even that's not enough) or a fan cooled heat-sink
the shape of heat sink is very important in its efficiency and honestly I
don't
think these L shaped aluminum sections can dissipate beyond a few Watt.

 

[Saman]

I use nylon shoulder washers a lot but as you pointed out the screw is not
isolated and I don't know is it safe to leave a bolt at 100V inside the
case
non-isolated.

If you want to connect to the tab using the screw, the shoulder
washer
keeps that screw from the sides of the heatsink hole. If you
don't want the screw live, the shoulder washer keeps the
screw from the sides of the tab hole.

the sholder washers that I've seen have a small cylindrical body
and usually their tube-shape section is long enough to only cover
transistor hole not the heatsink.

If it isn't the screw that's at 100V, it's other items inside the box.
So, close the box before powering it up. Apply a bit of lock
adhesive (Loctite 2240 or similar) to the nut so it doesn't
loosen.

Good advice about adhesive
thanks

 

[Jasen Betts]

I'm not sure about saftey but if I could do such a thing it would save me
from
a lot of troubles when trying to accomodate big heat sinks inside a small
box.
Because as I said earlier if I could directly mount transistors on heatsinks
then
each transistor could dissipate considerably more and I could use fewer
transistors.
Another problem with this scheme is when I have 8 parallel transistors
I either have to use a separate heatsink for each one or isolate them.

hard anodise is good for a about 200V insulation