Rectifier Box

[kitestrings]

Greetings,

I've been working on the rectifier box for our new axial.  I'm getting ready to do final mounts and wiring.  I thought it would be a good time to gather comments before I do.

There are heatsinks on both the front sides and ends of the enclosure.  The rpm gauge is mounted on the front 1-5 VDC coresponding to 100-500 rpm.  The center panel hinges from the top down to allow access to the turbine-side (50A 300V) breakers, and the Omron sensing relay.  The rpm board is mounted on the inside of the center panel.  Here it is mostly, loosely, assembled:
" alt="" class="bbc_img" />
" alt="" class="bbc_img" />

To access the diodes and wiring terminals I've hinged the front sinks.  Thumb screws secure them closed, but each pair is somewhat isolated from the metal portion of the box due to the anodizing and nylon shoulder washers and stops.  I'm not sure that this is critical, but Jim (Sencenbaugh) had suggested that the hard anodizing provided a shock barrier if the heat sinks were touched in a ground fault scenario. 
" alt="" class="bbc_img" />

Basically there are two sets of (6) diodes, each providing an isol;ated input to one of the two Classics CC.  The diodes are stud mounted in aluminum blocks - just one set of three is installed in the photos.  The blocks will be electrically separated from the heatsinks by sil-pads, and are connected to the sinks with nylon screws:
http://www.farnell.com/datasheets/313126.pdf

Also mounted on the side heatsinks are two SS relays. One is for our load diversion, or clipper and will be controlled by the Aux 1 output of the Classic.  The second is a hi-voltage limit diconnect, which I hope never has to opperate.  This was a scheme presented here by Rob Beckers.  The dump will be paired with the upper range CC, which overall I expect will be dissapating considerabley less heat.
" alt="" class="bbc_img" />

I would appreciate comments, and I have a few questions:

Is it recomended to also use thermal grease with the sil-pad?
How much overlap, or extention of the mmaterial should there be beyond the diode blocks?
Should the sil-pad be used under the relays as well (I think I have enough)?

Thanks for any input.

~kitestrings

[DanG]

Neat stuff looking at the spec sheet - but I would be hard put to trust nylon screws, can you accurately torque each 1/4-20 screw to 8 Inch-Pounds exactly? Nylon sleeved machine screws with a washer perhaps? The greater the pressure the lower the impedance to heat flow and nylon screws won't be good for very much.

As to using the pad as a heat spreader it doesn't appear to be very good at it - the oversized example pictured I think is as an assembly aide rather than increasing thermal transfer. Twice or three times thickness to accept torque deformation, and no sharp edges on the device periphery.

For the SSR's - any void filling paste to exclude possible air pockets would work except if the device surface is not neutral.

And, as an opinion - I don't like the diodes under the SSR's - the smokestack effect of that warm assembly will forever be heating the SSR bodies. I'd be tempted into placing the SSRs at extreme lowest position on the heat sinks with the diode block close above, let the heatsink only spread heat back to the SSR chassis instead of the passive thermals inside the case baking the SSR?

[clockmanFRA]

Kitestrings,
For what its worth, with my 3.7m 12footers each at max 42a and 61v, i found my original heatsink woefully inadequate and uprated accordingly although you can see in Pic 2 my heatsink for Turbine No2 is now very large, and just gets warm.

Pic 1, shows No1 Turbine controll equipment now with added heatsinks on my 3 rectifiers.
Pic 2, shows No2 Turbine controll gear in a more orderly and tidy arrangement, No3 turbine has the same.

[BrianSmith]

The heatsink / enclosure looks really good.  In regards to what Dan said about cooking the SSRs, not sure about that from what I could make out from the picture, but if the box is being mounted indoors, I would put some ventilation holes in the top and bottom to allow some convection air flow thru the inside of box and let the hottest stuff out the top.  If the inside top of the box starts getting too hot, you could always flip it over along with the meter and mount it upside down....  

If you need to spread the heat better, you will need something with a lower thermal impedance than the heatsink like a bigger piece of copper to mount to the rectifiers, then thermally bond the other side to the big heatsink.  If you don't mind using something active, you can also mount a thermal snap disk to the heatsink and when it gets too hot and closes, activate a fan to blow thru the heat sink fins.  That will lower the thermal impedance of your heatsink dramatically and double or triple its capability when needed. 

[kitestrings]

Thanks for comments, and I regret not following up sooner.

The sil-pad is marketed as a "thermally conductive insulator".  I suspect it is a challenge to do one or the other well.  For this size material, it had the lowest thermal impedance that I could find (.41  deg C-in^2/W @50psi).  Your point is well taken though Dan, that the thermal peformance is best under higher pressure.  I'm going to start by swapping out the screws for (hex) cap heads.  the latter has a max torque before deformation of 15-16 in-lbs vs. 9-10 fo rthe screws.  If we're still not satisfied we can go larger, or use insulated metal fasteners.

The material is suprisingly tough, however, notice it also has a breaking resistance of 30 lbs/in (5 kN/m).

The original heatsink design used 35 amp diodes on our 12-volt system, and I rarely saw the temperature of them be more than mildly warm to touch.  In our design, they are 85A diodes (1,200 PIV), but the load is split between the two rectifiers (so the amperage will be in a comparable range).  I'm further assuming that at the higher voltages that the Classics will allow that there will be less amps, and therfore less heat than if direct tied, but maybe this is not the case?  Maybe it is simply a function of power?  In any event, increasing the passive cooling seems easy enough, and adding fan cooling if it becomes necessary.

Jim once described the stud mounted diodes as a "brute force approach....[that was] perfect for wind."  I like it simply because you can swap out one or two if they get hammered in a lightning surge.

Clockman, thanks for sharing the pics.  What is the meter and digital display indicating?  It looks good, except for I'd be nervous having it all mounted in a wood box.

No one has said, "Stop you damned fool"! so I gues I will press on.  Thanks again.

Regards,

~kitestrings

[Flux]

Silpad doesn't need any additional grease but I presume it wouldn't hurt.  i agree with not using nylon screws.

You look to have plenty of heat sink so I think all will be well. Using proper stud diodes rather than the potted bridges means you can get away witl less heat sink. I seriously doubt that you will need any fan cooling.

Flux