Heat sink for 20 amp cylindrical diode?

Thread Starter

Saluki

Joined Feb 23, 2019
80
Ok, so I have some 20 amp 1000v PIV silicone diodes. I'm looking at using one to half wave rectify 120 VAC powering a "quartz" heater, to give it a "half" power setting. In a trial (with heavy clip leads making pretty solid connections per measured very small voltage drops @ the connections), I measure 6.0 amps through the diode. That measurement by clamp meter. With the diode switched out of its series connection with the load, ie., normal operation, current to the heater measured 11.5 amps, again per the clamp meter. All this is pretty much as expected. (I don't expect the clamp meter to be wonderfully accurate on a half-wave signal.) The rectification looked pretty good on my o'scope. (Yes, I used high voltage probes.)

However, with the diode in series and basically just hanging out in free air, the diode gets pretty hot, in said operation. I'd estimate about 190-195 deg. F. Ambient was about 25 deg. C. So, I'd like to heat sink the diode to the fairly heavy metal (steel) handi-box it will be in. Obviously I have to take care to avoid any shorts to the box!!

The diode measures about 8.5mm diameter: Slightly smaller than a TO-39, and a bit bigger than a TO-5. I was "sure" I'd find some sort of heat-sink-to-larger-heat-sink / metal enclosure on Amazon or eBay, but have not. I tried Mouser but didn't see anything right off the bat. Possibly my search was not defined well.

Are these diodes normally operated with little to no sinking? I can't see operating them at much over 1/3 their current rating, if so.

It seems to me that essentially a small "U" strap around 8.7 mm ID and some thermally conductive / electrically non-conductive "potting" compound would do the trick. Heat sink grease should work too, but, I can't really "encapsulate" the diode with it.

It also "seems" to me that I might be able to use 2 diodes in parallel, as any "unbalancing" due to variance in Vf of the two diodes in parallel, in this type operation, is swamped by the high voltage into the load itself. (I think?)

Has anyone tried anything like this? I think I can bend / fabricate a pretty good U-clamp out of soft aluminum or copper material I have on hand in small amounts. However, an easier solution if inexpensive is good too. :)

Thanks!

Edit: I just looked at thermal potting compounds (epoxies) on Amazon. Yikes!!! And I thought car prices had really gone up since I dabbled in this stuff! Well, I don't need CPU grade cooling here. Heat sink grease @ mechanical contact surfaces would surely suffice, and I have some of that... somewhere...
 
Last edited:

schmitt trigger

Joined Jul 12, 2010
889
If these are plastic body leaded diodes, the heat is dissipated through the leads! Dissipation thru the body itself is very low.

Solder to each lead a 1” square copper foil, as close as possible to the body itself.
 

schmitt trigger

Joined Jul 12, 2010
889
Ok, so I have some 20 amp 1000v PIV silicone diodes. I'm looking at using one to half wave rectify 120 VAC powering a "quartz" heater, to give it a "half" power setting. In a trial (with heavy clip leads making pretty solid connections per measured very small voltage drops @ the connections), I measure 6.0 amps through the diode. That measurement by clamp meter. With the diode switched out of its series connection with the load, ie., normal operation, current to the heater measured 11.5 amps, again per the clamp meter. All this is pretty much as expected. (I don't expect the clamp meter to be wonderfully accurate on a half-wave signal.) The rectification looked pretty good on my o'scope. (Yes, I used high voltage probes.)

However, with the diode in series and basically just hanging out in free air, the diode gets pretty hot, in said operation. I'd estimate about 190-195 deg. F. Ambient was about 25 deg. C.

It also "seems" to me that I might be able to use 2 diodes in parallel, as any "unbalancing" due to variance in Vf of the two diodes in parallel, in this type operation, is swamped by the high voltage into the load itself. (I think?)
Absolutely no. Forward conduction losses have no bearing with reverse voltage withstanding.
 

Thread Starter

Saluki

Joined Feb 23, 2019
80
Can you not make up one by drilling holes in a piece of copper or aluminum stock?
Yes, in fact if I hunt around I might have some grounding blocks that would be just about right. However, a poster downstream notes that most of the heat actually sinks out via the leads. That suggests a rather different approach, which I'll comment on, back to that poster, shortly...

Thanks for the reply!
 

Thread Starter

Saluki

Joined Feb 23, 2019
80
If these are plastic body leaded diodes, the heat is dissipated through the leads! Dissipation thru the body itself is very low.

Solder to each lead a 1” square copper foil, as close as possible to the body itself.
Ah, that suggests all sorts of possibilities. I have a couple big rolls of 63/37 solder, and solder flux for aluminum, too. Double sided heat conductive electrically insulative tape is available @ reasonable cost, if I want to sink to the metal box. A couple dabs of silicone, or, a couple small nylon screws for extra security...

Edit / update: Well, I couldn't find my little bottle of flux for Aluminum and Aluminum to Copper, etc. Rats! But, it has been 15+ years since I'd used it... So... I went looking for soft Copper I can "coppersmith" with. Heh, this should last me the rest of my life: https://www.menards.com/main/plumbi...il-tubing/22100076/p-1444436527694-c-8565.htm

They have an 11% rebate going too... :)
 
Last edited:

Thread Starter

Saluki

Joined Feb 23, 2019
80
Absolutely no. Forward conduction losses have no bearing with reverse voltage withstanding.
Oh, I wasn't even considering the reverse voltage rating, I was just looking at the forward voltage across the load vs. the diode drop. The signal could even be DC to start with. However, I guess that's not the way to look at it. Instead:

From the data sheet, a typical one of these (diodes) passes about 2 amps @ Vf = 0.87v, and 4 amps @ 0.92v. If the next diode slightly shifts that curve, it could easily be conducting 4 amps @ Vf = 0.87 volts -- so the fact there might be half sine waves @ the load peaking at around 160v doesn't really matter: I have one diode conducting 4 amps and one conducting 2 amps, right? (6 amps being what I measured with the clamp meter, going into the load.)

The above would still be an improvement, my "hotter" diode is now conducting 4 amps instead of 6, but, the improvement is not ideal.

I could also go "crude", since the 1st diode survived in a 2nd test the 6A for 8 hours: Hook up diode pairs until I get a pair where both are equally warm after a few minutes... Install with heat sinking and don't worry about it too much. ;-)
 

MisterBill2

Joined Jan 23, 2018
18,442
OK, indeed those diodes transfer the heat out by the leads. So you an add some copper wire "flags" soldered to the leads to improve the heat transfer to the air. OR mount them in the cool air stream to improve the heat transfer.
 

MisterBill2

Joined Jan 23, 2018
18,442
One more thing is that a real specification sheet will describe exactly the conditions under which the diode will be able to pass 20 amps with no problems. That will include the ambient temp and probably the junction temperature. I see none of that in the sale page. So what makes sense is to mount them so that they are in the inlet air path of the heater. AND add to the wire leads so that they can conduct more heat away. Diodes in parallel may help if they have very similar voltage drops.
 
Last edited:

Tonyr1084

Joined Sep 24, 2015
7,886
That will include the ambient temp and probably the junction temperature. I see none of that in the sale page.
I saw "China". Makes me dubious about the diode's ability and ratings. I can make anything and say it can handle billions of mega-amps. That doesn't mean it's so.

On the notion of heat sinking: small copper clamps similar to the plastic wire holder clamps could be easily made out of a sheet of copper or even out of a piece of copper pipe. Just cut the pipe to length then split. Splay it out flat and then bend it around a bolt that is roughly the same diameter as the diode. This will pull some heat away, but if the body is cheap "Chinezium" then heat sinking the body will do little if anything to dissipate heat. Others have suggested sinking the lead. That's the best advice I've seen thus far.

For quick and heavy duty heat sinks I take old LED lamps that have given up the ghost. They have heavy aluminum bodies. I've cut and shaped some to fit inside a glass tube. With thermal paste the heat is transferred to the tube which is cooled by water. THAT project ran low voltage UV-C LED's in a fish pond filter. Still working like a champ.
 

Tonyr1084

Joined Sep 24, 2015
7,886
I can swear I've seen this in copper AND where it wraps completely around something, typically wire bundles. Here's one from Grainger, made of steel: You could easily make something like this out of copper.
Screen Shot 2024-02-27 at 8.45.34 PM.png
 

Thread Starter

Saluki

Joined Feb 23, 2019
80
OK, indeed those diodes transfer the heat out by the leads. So you an add some copper wire "flags" soldered to the leads to improve the heat transfer to the air. OR mount them in the cool air stream to improve the heat transfer.
I'm thinking, yes, copper or Aluminum* flags, those to be attached with the heat conductive tape to the steel "handi-box". ~1 sq. inch of contact for each flag to the box. Another option would be a paste of silicone caulk with very fine sand mixed in it. Might be more durable.




*Since I have flux for Aluminum, originally used for soldering Aluminum wire in loudspeaker voice coils to the "tinsel" lead outs.
 

Thread Starter

Saluki

Joined Feb 23, 2019
80
I saw "China". Makes me dubious about the diode's ability and ratings. I can make anything and say it can handle billions of mega-amps. That doesn't mean it's so.

On the notion of heat sinking: small copper clamps similar to the plastic wire holder clamps could be easily made out of a sheet of copper or even out of a piece of copper pipe. Just cut the pipe to length then split. Splay it out flat and then bend it around a bolt that is roughly the same diameter as the diode. This will pull some heat away, but if the body is cheap "Chinezium" then heat sinking the body will do little if anything to dissipate heat. Others have suggested sinking the lead. That's the best advice I've seen thus far.

For quick and heavy duty heat sinks I take old LED lamps that have given up the ghost. They have heavy aluminum bodies. I've cut and shaped some to fit inside a glass tube. With thermal paste the heat is transferred to the tube which is cooled by water. THAT project ran low voltage UV-C LED's in a fish pond filter. Still working like a champ.
Agreed on the ratings - that's why I ran tests as noted above. Granted my temperature estimate was of the diode body only, as I have no good way to test the temperature of a "hot" (120 VAC) lead. (The old Mad Magazine tester, available from "Dept. Bzzztt, Blackhand, South Dakota", hasn't captured my fancy either...)
I saw "China". Makes me dubious about the diode's ability and ratings. I can make anything and say it can handle billions of mega-amps. That doesn't mean it's so.

On the notion of heat sinking: small copper clamps similar to the plastic wire holder clamps could be easily made out of a sheet of copper or even out of a piece of copper pipe. Just cut the pipe to length then split. Splay it out flat and then bend it around a bolt that is roughly the same diameter as the diode. This will pull some heat away, but if the body is cheap "Chinezium" then heat sinking the body will do little if anything to dissipate heat. Others have suggested sinking the lead. That's the best advice I've seen thus far.

For quick and heavy duty heat sinks I take old LED lamps that have given up the ghost. They have heavy aluminum bodies. I've cut and shaped some to fit inside a glass tube. With thermal paste the heat is transferred to the tube which is cooled by water. THAT project ran low voltage UV-C LED's in a fish pond filter. Still working like a champ.
Agreed on the ratings - that's why I ran tests as noted above. Granted my temperature estimate was of the diode body only, as I have no good way to test the temperature of a "hot" (120 VAC) lead. (The old Mad Magazine tester, available from "Dept. Bzzztt, Blackhand, South Dakota", hasn't captured my fancy either...)
 

Thread Starter

Saluki

Joined Feb 23, 2019
80
datasheet: https://semiware.com/uploads/datasheet/20A05-20A10.pdf

Typical Thermal Resistance (1) 20 degree C/watt, you're running about 6 watts -> 120 degree C increase in temperature!

I wonder what the saturation current of the power feed transformer is? Your drawing just DC...
This is AC line voltage dropping (getting rectified) through the diode to the heating elements load). Same as found in many quartz heaters where at "half power" the element or elements dim, instead of switching out one of two elements, leaving the other "bright". The dimmed elements method greatly extends element life. (I've never seen one fail.)

°I "could" rig a crude test by using an insulated probe to place a small drop of water on a diode lead close to the diode in operation, heat sunk only by the clip leads at the far ends of the diode's leads. That would tell me if the diode's leads heated to 100° C or more. (The violence of the boil, if the water boiled, would give some indication of "how much hotter than 100° C is it".) At present all I know is the diode body exterior surface stayed a bit under 100° C (similar conditions).
 

Thread Starter

Saluki

Joined Feb 23, 2019
80
datasheet: https://semiware.com/uploads/datasheet/20A05-20A10.pdf

Typical Thermal Resistance (1) 20 degree C/watt, you're running about 6 watts -> 120 degree C increase in temperature!

I wonder what the saturation current of the power feed transformer is? Your drawing just DC...
That rating is at Ta = 60° C. The mfgr.'s notes state:

"Thermal resistance from junction to ambient and from junction to lead at 0.375" (9.5mm) lead length, P.C.B. mounted with1.1” x 1.1” (30 x 30mm) copper pads"
I'd say that definitely confirms the cooling via leads. :cool:

I suppose the mfgr. must be assuming some sort of forced, ie., fan, cooling to keep the "immediate ambient" volume around the diode itself at 60° C, or, an effectively infinite volume around the diode? In typical usage in an enclosure, 20° C/watt is going to build up the immediate ambient in a hurry. At the rated current through the diode in a small space with no further sinking...
1709137366733.png

Well, all I really do know is that with the diode suspended in a large space (roughly 1000 cu. ft) by two 18 ga. clip leads at the far ends of the diode's leads, ambient = ~ 22° C, and with the clamp meter reading 6 amps*, the outer surface of the diode maxes out a little under 100° C.

*Less the diode, current to the heater read 11.5 amps. The quartz heater is rated at 1500 watts, but I think that's some sort of compromise rating between the steady state power and the initial, somewhat slow turn-on. Plus the heating elements have some use on them, I'd estimate 5000+ hours, and may have boiled off a little material by now... My measurement results pretty much seem "reasonable"... I think?
 

Attachments

Last edited:
Top