microwave resistors for a large dummy load

[Opcom]

Just ordered 8 of these 100 Ohm 500 W 1 GHz chip resistors.

They can be placed in series parallel for 50 Ohms, 4KW. I don't know what to use as a heat sink and have yet to read all the appnotes. suggestions? 500W dissipation in a 0.5x1.25 inch area will need a good spreader.

These have two 'leads' in the form of tinned metal strips. The connection is optimally to be by PCB but for HF could be done other ways.

I don't know the proper layout yet till I figure out the heat removal. Prefer conduction and convection but it might not be completely practical to go without a fan.

Florida RF Labs: RF Power Resistor - SMT - Flange /Tab & Cover: 31-XXX1: 31-1123-100-5
part # 31-1123-100-5
100 ohm resistor 500 W 5% BeO Cap Max: 1.50 pf 2 s
   
Manufacturer:   Florida RF Labs
Status:   Standard
Datasheet:   31-1123-100-5 DataSheet
RoHS Compliance :    No

http://www.rflabs.com/microwavepassivecomponents-a1010.html

[WQ9E]

Patrick,

How about a modern Cantenna?  Mount them on small heat sinks and immerse the whole assembly in oil.

This will provide a no noise solution unless you are trying to come up with a load that will allow you to do long term testing at 1 KW out in which case there is a fan in your future. 

[N4LTA]

Use a 1/4 inch thick copper spreader to couple them to a big heat sink. I have done that in the past with large RF Power FETS. You can buy the copper sheet online cut to your dimensions.

Pat
N4LTA

[WA1GFZ]

These resistors are quite common. I have a nice stash of different resistances and power ratings. Yes a copper spreader is a good way to go at higher power levels.

[w3jn]

Can of earl with a corrugated radiator, you betchya.

[WD5JKO]

I like the idea of a heat spreader in oil as others have already said. Keep in mind though that these resistors need to stay below 100 degrees Celsius, so the spreader needs to be thick enough and have enough surface area to the oil to do the job. If the spreader is not up to the job, then the resistor gets way too hot before the can begins to feel warm.

Might be time to look at those thermal cooling equations. This gets complicated since the spreader fins orientation will either aid or hinder natural oil circulation.

I once was given a grocery sack of 47K 1/2w CC resistors. I got the half cocked idea to make a dummy load by making 10 circles of 100 resistors all in parallel! That took a sh_t load of time and solder I put the monster rings in a paint can full of ATF. From the outside it looked like a Cantenna.

I learned a few things besides this was a stupid idea. First off the heat required to solder those rings ratcheted up the final resistance from 47 ohms to about 80 ohms, and the shunt capacitance of 1000 47K's all in parallel was immense. It was fun though

Jim
WD5JKO

[N4LTA]

I have a dummy load made from 40 - 2 watts metal oxide resistors. These are the ones from Mouser that the were tested in QEX (If I remember correctly) and found to be very flat up to 100 Mhz or so.

I use it dry for loads up to 100  watts and it works very well and I have loaded it for 30 minutes at 80 watts.

I am surprised at the performance as I have abused it. It is made from 40 - 2K  resistors in a circular form with a SO239 in the center  - between 2 copper sheets about 4" square.  The SO239 is mounted to the top sheet and the bottom sheet is connected with a piece of #12 copper bare wire to the center pin. The wire is the same length as a resistor.The resistors go between the two sheet with essentially no leads.

Pat
N4LTA

[Ed/KB1HYS]

I took the dummy load resistor (big tubular job) out of an MFJ tuner that had failed, mounted it vertically in a metal paint can using copper strip and such.  RTV on the PL-259 etc. 

just like a cantenna. the load was originally rated for 100watts. It would get pretty hot running tests on a Valiant.  don't know what it will do in the oil. 

[WA1GFZ]

130 degrees per watt per square inch.
So 1000 watts with a 1000 square inch heat sink area gives you 130 degrees of temperature rise. so a heatsink with 2000 square inches would be good for 1000 watts. 65 degrees plus 25 degrees ambient to stay below 100C
I would run 3/8 or 1/2 inch copper spreader on a 400 watt part. The thicker the copper the better it spreads heat. A smart guy once told me it is useless to go more than 10X thickness wide and most of the heat is concentrated to 3X the thickness. On 1/4 inch spreader that is 2.5 inches and .75 inches. I guess this is why my 1/4 inch spreader material is 4 inches wide and the 3/8 is 6 inches wide.

[vk3he]

Hi Patrick

What did those resistors cost you per piece?

Craig

[Opcom]

The reel-end of 56 Ohm 2W 5% carbon resistors I was given made a nice dummy. Each row of ten is soldered between two strips of 1.5"x10" double sided PCB. 9 rows stacked and it is FB to about 200 MHz and usable to 460MHz. Air cooled. It's seen 100W for many hours and survived for 15 years or so. I built it when i built my first leenyar, a pair of 6146's at 100W PEP outpoot.

These new pieces cost $25 each from the distributor (in Texas, so add the tax). The dist. was very helpful and polite.


Posted by: WA1GFZ
130 degrees per watt per square inch.
So 1000 watts with a 1000 square inch heat sink area gives you 130 degrees of temperature rise. so a heatsink with 2000 square inches would be good for 1000 watts. 65 degrees plus 25 degrees ambient to stay below 100C
I would run 3/8 or 1/2 inch copper spreader on a 400 watt part. The thicker the copper the better it spreads heat. A smart guy once told me it is useless to go more than 10X thickness wide and most of the heat is concentrated to 3X the thickness. On 1/4 inch spreader that is 2.5 inches and .75 inches. I guess this is why my 1/4 inch spreader material is 4 inches wide and the 3/8 is 6 inches wide.


Is this for convection cooling in air? It is preferred to be air cooled.
For each pair of resistors, 1KW.
2000 sq. inches is needed for a 65 deg rise.
-so-
1000 sq inches for each 500 watt resistor.

That is right at 32" on a side, with the resistor in the center.

This 32" square piece of copper for each 500W resistor would have to be how thick? I am not sure I understand about the problem. I am confused about the size of the copper spreader, then the size of the aluminum heat sink. Then the size of the fins, 1" or 2"..

The device has an operating temp of 100 deg C. It has a footprint of 1.25"x0.25" (0.625 sq in.), and its little built in spreader is 0.125" thick.

I have no objection to this being physically large. I do intend to try to have it able to take the 4KW that is the total rating of the 8 resistors.

I'm trying to understand what you have posted. I was reading the attached paper from Thermalloy as well.

[WA1GFZ]

You want to use a heat sink with 3 or 4 inch fins. Look at the specs and find one that is around 100 square inches per linear inch. Then mount two of them side by side. I would just put a hunk of 3/8 copper spreader across them. 6 inches wide by say a foot long. Then mount the resistors so you have very short leads 
Example my solid state linear project I have 4 heat sinks each a foot long. I plan to use two 1200 watt Erbteck MRI amplifiers. They are about a foot square and the heat sinks are about 7 inches wide. My heatsinks are about 98 square inches per linear inch. So I will have just under 2400 square inches per 1200 watt amplifier. The efficiency is about 50% and I will not need to run them wide open to make 1500 watts between two modules so there is plenty of cooling. I plan to mount the two modules in a block with fins facing each other. This way I can blow some air through them.  I have a 4 foot by 6 inch slab of 3/8 inch copper that will cover the heat sink decks. 
Wait till you see the price of copper.

[Opcom]

I see. I had disregarded the area of the fins in the calculation and so of course a simple flat plate would be enourmous.

[WA1GFZ]

It would be cheaper to buy a dummy load than a 32 inch square hunk of copper plate. BTW CTR Surplus on ebay has some copper heat sinks.

[Opcom]

Something like the drawing? Where the copper sits on a larger heat sink?

It could be broken up into smaller units if necessary due to material availability, but better to make in one piece so the leads are shortest. I should be able to get the heatsink(s).

There is one from Thermalloy, standard aluminum extrusion type 65535, which is 12.2" wide, 2.88" tall, and 115.5 square inches per inch.

It looks like a fan is necessary to make anywhere near full power.
The reason for this high power rating is for smoke-testing the "JawsII" 3CX3000 amplifier.

They have nothing over 3" except exotic shapes.

The thermalloy site gives temp rise per watt, according to the length of the extrusion. It's interesting to note that the "length*wattage" is not really equal to the temperature rise they way they calculate it. No chart on their site shows rise at over 250W either.

length-watts-rise

3" 50W 37deg
6" 70W 37deg
9" 88W 37deg
12" 95W 37deg
24" 137W 37deg
I always thought it was linear.

I hope the drawing reflects your advice. If not please let me know what's wrong. This is really interesting!

[WA1GFZ]

I understand your problem with the specs. I have the same concern. My problem is cooling 8 MRF150s in a line along the 12 inches of the Erbteck board. Similar to your problem. I plan to cover a 12 by 14 inch heat sink with 3/8 Copper to help spread the heat over the whole surface. Then mount the heatsink with the fins vertical with the FETs along the bottom edge. Heat wants to rise so the heat will move away from the FETs in a natural flow. Since both heatsinks will face each other there will ba a channel . I plan to put a chassis on the bottom end and
pessurize it with a blower to help the heat move up.
You actually have a bigger problem because you want short leads to avoid adding inductance to the load yet you need to spread out the heat on the deck. I see a number of rigs with the heatsink positioned fins down. This is a bogus way to control temperature rise. We have a couple broadband amps at work that I have studied. Each one has the fins facing up with air blowing through the fins.
My best advice is to use the heatsink with the tallest fins so you can concentrate the heat. Also use the thickest copper you can afford. Then there is the CTRsurplus copper heat sink which could be a good choice for you. Copper has 1/2 the thermal resistance as aluminum.

[Opcom]

That's great! Ok, I do plan to mount the heatsink so the fins are vertical, just my drawings are easier to do the way I did it. I guess there is no way to know for sure except to do the best job I can and monitor the temperature at one of the resistors. A fan or two is definitely part of the plan no matter how things are done. I had in mind this could be mounted to the backside of a rack panel and a rather large SPDT antenna switch could be used to select the load or the dummy. The switch can be set up to only select the load if there is power to the fan(s). I'm also looking at the STLM20, a small temp sensor from STMicroelectronics. It's analog and can be hooked right to an op-amp.

CTRsurplus has a Heavy Copper Heatsink 12.75” x 9.375” x 4” 48lbs. $225!! yow! I think a copper plate on an aluminum sink will do for me.

I don't need to take this over 30MHz and if the SWR is a tiny little bit off I am not going to gripe, so the lead length should be OK as long as I don't get sloppy.

This is interesting, but probably $$ for their products.
http://www.novelconceptsinc.com/calculators.htm

Thanks,
Patrick

[WA1GFZ]

Just don't get them too hot because they have a BeO substrate.

[Opcom]

I ordered the copper heatsink from CTR Surplus. 12.75” x 9.375” x 4” 100 fins 48lbs.

9" x 4" is 32 sq."

32 sq." x 100 fins is 4800 sq. in. Hope this will work, should be OK with a fan?

FEB 2015: correction 9.375" x 4" for one side is 37.5 sq. in.. but there are two sides on each fin, so each is 75 sq. in, and 100 fins is 7500 sq. in.

[k4kyv]

I learned a few things besides this was a stupid idea. First off the heat required to solder those rings ratcheted up the final resistance from 47 ohms to about 80 ohms

It took me a long time to figure this out, but carbon composition resistors need to be  handled just like transistors as they are soldered in place.  I sometimes clamp a haemostat over the wire lead that is being soldered, to sink the heat away from the body of the resistor.  Many times I have painstakingly sorted resistors to find a matched pair, or one with the exact resistance I need, only to find that the resistance has changed considerably after I soldered them into place.

Not only do those resistors drift with age (even n.o.s. ones that  have never felt the heat of a soldering iron), but I suspect much of the resistance error we find when checking out old equipment is a result not of ageing, but from soldering heat dating back to the day the equipment was manufactured, and they had always been off-value throughout the life of the unit.

Off-value resistors seem to be an extremely common problem with older equipment, suggesting that 90% of the circuit resistances in vacuum tube equipment are non-critical, and that the majority of errors up to +/- 50%, and sometimes even more, have little or no effect on equipment performance.

My dummy load  consists of 12 each, 18" X 1" Glo-Bar resistors in parallel, mounted in a circle, vertically.  No fans or oil, they self-cool by convection, taking advantage of the chimney effect of the tall vertically mounted hollow resistors.  I am not sure of the power rating of those resistors, but assuming at least 100 watts per resistor, that would make the load good for 1200 watts continuous duty 24/24, and it is completely silent.

[Opcom]

well Don, I probably won't need a real noisy fan.. The radio have blowers anyway. I can appreciate what you have there in the winter though. Are the glo-bars things that are still currently aavilablke or ones you found surplus? I think you posted a picture of that load before, can you post it again?

[Pete, WA2CWA]

Globar: Bulk Ceramic Non-Inductive Resistors:
http://www.globar.com/ec/bulk-ceramic-resistors