Flat heatsink for F1 ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
If I looked at the design right he has 13.5 watts burning in each output transistor. IRFP240's being what they are, it's probably ok, but I would personally be interested in the case temperature rather than the temperature at the edge of the plate directly above the transistor, especially in the usual 20C room rather than 7C. 1/8" material at that distance is going to show a very large gradient between the case and the plate edges. I would recommend 3/16 or 1/4 with the same area plate at that dissipation. Obviously this would cost more and take up greater space. Or you could run 1 transistor per 1/8" plate.
 
I don't think you gentlemen have read and understood the first few posts in the AX100 thread completely.

The main heat sink is comprised of 3/16" thick aluminum plates. The .125" thick plates are used only for the end plates. The end plates don't have any transistors on them.

When the amps were first powered up I ran 13.5 watts of dissipation per transistor for a week or two to see how the sink would perform. All was good. I THEN increased the dissipation to 15 watts per transistor. The sink temperature rise was still modest and is given in the thread. These amplifiers have been in constant use since they were built in 2006, both in summer heat as well as winter. The measured temps were made with a Fluke infrared hand held thermometer pointed down into the heat sink at the transistor cases. This not a rigorous methodology but IMO it's close enough. So I would rate the heat sinks a success even though they are, as I said in the thread, very ugly.

I have used this heat sink construction method many times and I am currently building a pair of 200W class A monoblocks with 16 main plates, each 18" x 7". The plan is to once again to use 3/16" plate and put 2 transistors on each plate. I am shooting for 18W of dissipation per transistor this time. We'll see how it goes. I will increase the plate thickness to .25" if I need to, and if I'm willing to deal with the added weight.

The thickness of the plates makes a huge difference on the efficiency. I have tried .125 and even .1 and I wouldn't recommend either thickness for class A. There is a heat sink calculator somewhere on the web for flat plate stock but I don't remember where it is. Rod Elliot's site maybe.

I have found that the dominant thing in heat sink pricing is the cost of the raw aluminum. Whether extruded stock or simple flat sheet stock, the cost is pretty close. The sheet stock, however, is far easier to find and in my view, is much easier to work with. All you need is a table saw and a drill, and there are no threaded holes -Yay.

Graeme
 
Last edited:
Hi pixpop,

To help the discussion, I have attached pics of the Jungson, and McIntosh MC1 2000W monster amp which use a type of aluminum plate output transistor heat sink. You can estimate physical plate fin sizes from the pics.

There are a few heatsink simulation programs on the web. I have used R-Tools. There are also simple tables for heat spreading on the web which can give reasonable estimates for a single fin.


I have also attached a pic of highly regarded extruded aluminum heatsink profiles to aid in comparison. Heatsink USA is a DIYaudio vendor. It is sometimes valuable to have all transistor junctions operate near the same temperature. Bipolar Class A/B amps bias optimization often assumes a uniform temperature on all transistors. ThermalTrak bipolars include an on-header diode which provides fast time constant response, but still only averages thermal drops over several transistors.
 

Attachments

  • jungson_09.jpg
    jungson_09.jpg
    83.2 KB · Views: 116
  • mc-1.jpg
    mc-1.jpg
    37 KB · Views: 111
  • mc-2.jpg
    mc-2.jpg
    33.8 KB · Views: 114
  • Al_HS_Extrusions.jpg
    Al_HS_Extrusions.jpg
    74.9 KB · Views: 113
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.