Controller builders: I was not going to share this with you

T

Teh Stork

1 kW
Joined
May 25, 2011
Messages
463
But now I'm doing it.

One of the keys to handling high power is handling the thermal challenges with it. China controllers use silicon pads or polyimide (kapton) tape to insulate the mosfets from the heatsink. I've found the perfect insulator - aluminium oxide. By growing a oxide layer 30-50 nm thick it safely insulates against 100-200V. The trick is growing this, you need battery acid, high amperage and refrigerator temperature - without the moisture. I've not been able to score any contract with a chinese company for manufacturing of these in small quantities - so I've settled for making my own little production facillity. I used these chemicals with expencive equipment at school, so if anyone have any tips as to how I can do this safely at home - I'm all ears :)

Other users of this board has electically connected the mosfets to the heatsink, well good job - you've now got yourself a RF antenna for the noise to efficietly escape the controller.

This is, by far, the best method for connecting mosfets to a heatsink.
 
Years ago, before TO220 packages became so popular, aluminium oxide insulators for the old TO3 packages were common, in fact I still have some here in my junk box:

AlO2 insulator.JPG

They are certainly very good indeed and were routinely used on pretty much all the power stuff that I worked on in the 1970's. A lot of military grade hardware uses this method to mount high reliability components. Hard anodising (pretty easy to do at home with some battery acid and a power supply) will do the job pretty well, too.
 
Yes, aluminium oxide is a very good compromise between thermal conductivty and electrical isolation. The commcercial process for this is Plasma Electrolytic Oxidisation or PEO for short. One company that I've used for this exact purpose is Keronite (www.keronite.com). My use for making a commercial product for super high brightness LEDs for use in street lights, etc. You won't really need anything beyond 30 um in thickness. I doubt you'll find it China.
 
PEO sounds pretty cool. Maybe even within the the capabilities of our production level.

Granted I've basically just read the short wiki page on it, but it doesn't sound too complicated like an actual plasma process, unless that is occurring on the low level. Most things with plasma oxidation that I've seen involve placing the object in a chamber where oxidized radicals of various gases are bombared on the object. Those are expensive. But can do things like covalently bond a silicone polymer to glass (awesome!!).

From the wiki page, it made it sound as simple as electrolysis, with a high voltage, and with using that high voltage to penetrate the initial oxide layer. As I typed this, however, I realize that applying 200 volts could not be applied to some simple liquid because other things in solution would be effected. I also don't see how the oxygen penetrates the metal unless oxygen penetrates Alu the same way carbon penetrates iron (Alu's small radius compared to Fe makes me think there's no room in the structure to fit oxygen.) So where is the oxygen coming from? Does it just diffuse through the oxide layer? Wikipedia listed a few papers by a cambridge professor, probably worth reading. The thing is, 200V is not hard to produce with a used gel electrophoresis psu, which are quite cheap on ebay. And it sounds like the depth of the oxide layer would be largely controlled by voltage.

Anyway, I'd like to hear back from people who actually know what they're talking about (not me.) I apologize for rambling.
 
Stork, I hate to say it, but the secrets been out for awhile... :mrgreen: In fact CamLight has been advocating it: http://endless-sphere.com/forums/viewtopic.php?p=213967#p213967

If you want hard coat anodize, do it at home systems and/or chemicals go to caswell plating here: http://www.google.com/url?sa=t&rct=...poCoDA&usg=AFQjCNGpT32BUbKJe_t5v9ODZDaU-hzEFQ

The price for the kit is a little steep, but you don't need all of it. Instructions are on the web how to do hard coat with respect to temp, currents and such.
 
bigmoose said:
Stork, I hate to say it, but the secrets been out for awhile... :mrgreen: In fact CamLight has been advocating it: http://endless-sphere.com/forums/viewtopic.php?p=213967#p213967
Click to expand...
Yea, what he said about what I said. :mrgreen:
I do agree that it's, by far, the best method for insulating active electronic devices (MOSFETs, whatever) from a heat spreader or sink. I tested the 1mil Type III hardcoat (0.5 penetration, 0.5mil surface coat) I had put on on the heat sink used for my CC-400 electronic loads and got a thermal resistance value of 0.02C/W...incredible. And the stuff is incredibly hard so be sure to do any machining first!! It chews up drill bits real quick.
 
Teh Stork said:
Other users of this board has electically connected the mosfets to the heatsink, well good job - you've now got yourself a RF antenna for the noise to efficietly escape the controller.
Click to expand...
Actually, bonding the FETs to the heat sink can decrease the amount of radiated energy. By insulating the FET from the sink, you create a capacitance between the FETs and sink. In a controller, with a high switching frequency, this creates a lot of EMI/RFI. You need to carefully RF-ground the heat sink if you choose to insulate the FET from the sink.
 
One downside of having a super thin insulator is less tolerance to surface irregularities. You have to be real careful around any holes to make sure there is no burr that could penetrate the coating or cause an air gap.
 
fechter said:
One downside of having a super thin insulator is less tolerance to surface irregularities. You have to be real careful around any holes to make sure there is no burr that could penetrate the coating or cause an air gap.
Click to expand...
Darn good point! Deburr all holes and only use very sharp bits with slow feed speeds to prevent "mounding up" of metal around the hole (especially when it's aluminum). But, I'd hope that would be done even if using SilPads as almost any burr would probably seriously affect even the thickest pad's effectiveness.

Not that the thickest pads are very effective in the first place though. :mrgreen:
 
fechter said:
One downside of having a super thin insulator is less tolerance to surface irregularities.
Click to expand...
FYI
I have an interesting thing about "surface irregularities"...
I work at Vincotech, and we use phase-change material for thermal conducting between power module and heat sink.
Please take a look (there is a video and an App.Note also):
phase-change material
 
Nice...great thermal conductivity spec for your compound! I've been using Aavid's UltraStick phase change thermal interface compound and love it. These phase change materials don't dry out or pump out over time and just blow away the silicone greases.

They don't handle anything more than the sub-surface voids on an already smooth surface (like any grease/compound does), but IMHO they're the best choice for interfacing heat sinks and active devices (FETs, etc).
 
You must log in or register to reply here.
Back
Top