Why pull and redo power transistors on the heatsink

[I LIKE MUSIC]

Clipping.... raises the DC content in the signal

I mean no disrespect, my OCDness makes me do things like this.

There is no DC content in a clipped waveform (regardless of what it looks like on an oscilloscope), it still remains a complex waveform''

If the output voltage stays the same, the power will increase because of a change in the crest factor of the waveform. A fully clipped sine wave (which may become close to a perfect square wave) will produce almost twice the power for a given load and output voltage as an unclipped sine wave..


BTW, for transistors that use sockets, it is not a bad idea to pull the transistors and clean the socket contacts and the pins of the transistors regardless of the performance of the heat sink compound.

 

[jstang]

Keep in mind that clipping blows tweeters....

All depends on the time frame for the sample or how long it stays clipped.... and the freq response of the driver & cross over... You may want to call it complex, but the time that the signal stays clipped may be long enough to look like pure DC to the Tweeter & its cross over, the flat top of the clipped wave charges the crossover caps like is DC..... and then that swings the VC wildly in the opposite direction once the "nearly square wave transitions ground and clips on the other side...." it has the same affect as applying large amounts of DC to the tweeter and poof if it is allowed to take place for too long...

A 30 hz clipped sine wave at full power is DC for a driver that has a freq response up to 20K hz. When it swings from full power + to full power - ( either way ) .... The time constant of the crossover cap applies voltage at whatever is the Peak to Peak clipped voltage to tweeter in what appears to the tweeter to be DC..... Melts the VC.....

It almost acts like a voltage doubler.... if the Clipped waveform is 40V peak to peak at 30 hz.... as far as the Tweeter concerned, it is could be 80 volts for long enough that it sees it as DC....and melts or mechanical damage.

 

[Mike Sweeney]

Back to the topic of why pull transistors.. here is the results of pulling them. You can clearly see the silicon has migrated down from the heatsink onto the underneath of the heatsink and into the connector. You can also see where the compound has migrated away from the center of the transistor and is down the legs of the transistor. None of this is good for longevity and heat transfer. If the amp is rarely used or used at a low volume, you would probably never notice an issue. But, if you crank it or drive 4 ohms vs 8 ohms or you drive four speakers, you could have a problem of overheating unless you fix it. And like I said before, once you are this far into the job of "restoring", then why not do it right:?

 

[jstang]

Is it dried out? And cracked....?

And agreed.... do it....

 

[Mike Sweeney]

Is it dried out? And cracked....?

Uh.. yeah.. parts are in fact. Typically around the edges and the thicker part surrounding the legs of the transistors. Not all of it and not everyone. And "dried out" is a mis-statement.. it's more of a soft chalky consistency. Which is what I would expect if the silicon carrier wandered off somewhere else.

 

[whoaru99]

So just the areas that are not really contributing to thermal transfer anyway are the "dried out" areas?

 

[Mike Sweeney]

So just the areas that are not really contributing to thermal transfer anyway are the "dried out" areas?

You didn't look closely at the pictures. The edges and the center voids around the pins were crumbly. The centers were a massive void of none to little thermal compound from it migrating over the years between the pressure and heat/cooling cycles. The silicone actually left the compound and ended up on the bottom of the heatsink and into the connector attached to the transistor. I don't think it "dried" out as much as the silicon left the party. Which is exactly the same behavior we saw at Hughes with our military equipment. And we reproduced it in the heat chambers with forced aging. I would like to think the newest compounds we use today work better

 

[jstang]

I have used the Dow 340..... This doc provides its electrical characteristics beyond the thermal specs. Which actual might be more interesting than the thermal characteristics.

https://users.obs.carnegiescience.edu/crane/pfs/man/Misc/DowThermalMaterials.pdf

You have to be careful with tuner lube and RF.... But can there be an better heat sink compound for final audio stages? Clearly you need nonconductive.. But should we be concerned with the dissipation factor and other electrical specs of the compound? Do other brands even provide such levels of characterization?

Not sure its worth the effort... Will have to noodle this....

Jk

 

[Mike Sweeney]

I like this line
Evaporation @ +200°C, 24 Hours, %/Wt. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.6

So we lose .6 % of the liquids when baked at 392 degrees for 24 hours.. thats a touch warmer than my amp will ever get But at least you know it does slowly evaporate .. very slowly

 

[Mike Sweeney]

This document from ON semiconductor was very interesting about torque, thermal transfer, mounting types etc.. https://www.fairchildsemi.com/application-notes/AN/AN-4166.pdf

 

[jstang]

But 45yr x 365dy x 24hr is 394200 hrs....

I am still thinking about the "capacitor" that is formed by the case of the power xter, the mica insulator, the heat sink compound and the heat sink... The xter body is one plate of the capacitor and heat sink is the other plate....

Most likely meaningless compared to other stray capacitance and the power levels.....the heat sink compound is not likely going to affect it much... As a guess....10 pf a xter? Or so. Plus what choice do you have other than a dielectric grease, none without a short....

 

[ConradH]

But 45yr x 365dy x 24hr is 394200 hrs....

I am still thinking about the "capacitor" that is formed by the case of the power xter, the mica insulator, the heat sink compound and the heat sink... The xter body is one plate of the capacitor and heat sink is the other plate....

Most likely meaningless compared to other stray capacitance and the power levels.....the heat sink compound is not likely going to affect it much... As a guess....10 pf a xter? Or so. Plus what choice do you have other than a dielectric grease, none without a short....

That capacitance is an interesting topic. I once designed a high voltage amplifier with power MOSFETs back when they first came out. Nothing I did would stabilize it for every condition. The amp was designed specifically for capacitive loads and used an output stage a bit different from out typical audio amps. Anyway, the problem was the capacitance between the TO220 devices and the heat sink. The cure was changing from Silpads to thick alumina ceramic insulators and grease to cut the capacitance. No more oscillation.

 

[whoaru99]

You didn't look closely at the pictures. The edges and the center voids around the pins were crumbly. The centers were a massive void of none to little thermal compound from it migrating over the years between the pressure and heat/cooling cycles. The silicone actually left the compound and ended up on the bottom of the heatsink and into the connector attached to the transistor. I don't think it "dried" out as much as the silicon left the party. Which is exactly the same behavior we saw at Hughes with our military equipment. And we reproduced it in the heat chambers with forced aging. I would like to think the newest compounds we use today work better

I did look very closely. It is fairly normal appearance for when you separate those parts. There should be very litttle compound between as its purpose is only to fill the minute surface inmperfections, not have a layer like frosting on a cake.

Was just going to leave it alone but, if I was a betting man, I'd bet more stereos have been damaged by people attempting to re-goop than saved by it. Now, of course the re-gooping itself wasn't the problem, yet in the process something goes awry such as this recent example.

http://audiokarma.org/forums/index.php?threads/power-amp-repair-please-help.810454/#post-11366931

 

[Mike Sweeney]

I wont disagree with that given the "more is better" mind set.. and I wouldn't take the bet

And with regard to the post.. ::faceplate onto my workbench::

 

[jstang]

Conrad .... Why? Why did you have press the audio OCD button.... Now I need follow those bread crumbs...Interesting .... Very interesting...

That capacitance is an interesting topic. I once designed a high voltage amplifier with power MOSFETs back when they first came out. Nothing I did would stabilize it for every condition. The amp was designed specifically for capacitive loads and used an output stage a bit different from out typical audio amps. Anyway, the problem was the capacitance between the TO220 devices and the heat sink. The cure was changing from Silpads to thick alumina ceramic insulators and grease to cut the capacitance. No more oscillation.

 

[jstang]

So thinking about the stray capacitance issue from the mount and heat sink compound....Kind of doubt BJT will be affected the same as a MOSFET.... But something to keep in mind.

http://www.ti.com/lit/an/sboa020/sboa020.pdf

A neat little doc from TI....

Page 4...
"Although the grease may dry up over time, this does not degrade the thermal resistance provided the mounting joint remains tight."

So.... Tightening the screws may be sufficient.... But I still would refresh the heat sink compound as cracks and gaps likely have developed. Also mentioned earlier in this thread , is cleaning the leads/socket to refresh connection quality. If the unit is not worth this effort...it's not worth a recap either....

Jk

 

[ConradH]

Excellent link. My guess is most people use too much grease and then over-torque the hardware.

 

[jstang]

They nailed it, TI has been a very good steward of the electronic industry...

By the way... The zinc oxide heat sink compound is a close relative to zinc oxide sun screen....

And seemingly used to slow the aging process.... I don't think I would apply it like this and go out in the sun....

Go figure....

 

[Mike Sweeney]

Funny thing is that Marantz used split lock washers which they expressed said "don't use" I don't think they were going for max transfer.. it was probably more of "this will work" The 7 inch pounds is what I worked out yesterday reading up on things.. and I am guess most go way over when they tighten it up. 7 inch pound doesn't feel like very much

 

[jstang]

Well like many things.... Recommendations are just that ....