Splitters/combiners

Discussion in 'Amateur Radio Amplifiers' started by AA7QQ, Jul 3, 2017.

ad: L-HROutlet
ad: l-rl
ad: Subscribe
ad: K5AB-Elect-1
ad: L-MFJ
ad: MessiPaoloni-1
ad: Left-3
ad: Left-2
  1. AA7QQ

    AA7QQ Ham Member QRZ Page

    It does seem the best option.

    So, for my water cooling idea:
    GE MASTR II continuous duto amplifiers have 3"X6" heat sinks perimiter bolted to the aluminum spreader.
    At the tips of the fins, there is .380" & at the base, it is .310".
    I intend to get some 3/8"(.375") soft aluminum tubing (Race car fuel line) .035 wall & use the hydraulic press at work to press it (Hopefully) all the way into the base of the heat sink.
    I have some of the fuel line coming in 7 days to give this a try.
    Now, I am sure that many of you know the exact heat sinks I am looking at (I could use some more if anyone has some that may get scrapped). I will router a chamfer on the ends of the fins.

    Here comes the major question in this part of the amp build:
    I will mount 1-BLF278 LDMOS (300w [PDISS 500 W @ TC = 25 *C]) per 3"X6" heat sink. Each has 5 channels for the tubing.
    Let's call them L to R 1,2,3,4,5. If using just 1 3/8" feed per heat sink, water input comes in 3, turns 180*(Have to experiment with tight turns), to 1, to 4. to 2, to 5? I am thinking that running down the center first, would draw a decent amount, then distribute some to channel 1, then back to 4 to draw some more, then to 2 to draw some more & 5 to dissipate a bit, then out to a manifold that will have 3/4"-1" out. It would go through a radiator (Car radiator) placed outside (Las vegas 30* to 117*) & into a tank, From the tank, it would possibly go through another exchanger inside to cool it some more when needed (Fan controlled by arduino).
    In this scenario, there would be 4 BLF278s, each fed by 3/8" tubing, each. the amp would be fed by one or a pair of 2.1 GPM pump(s). THere would be valves at the input of each pump to regulate flow & cheap inline flow meters on the outputs. An arduino would read the temperature of each heat sink & the flow out each module. I might add PWMing the pumps, but doubt it. The arduino(s) would also be able to switch on a small 2" fan per module, if needed in a pinch, or if I get long winded.

    I will grease up a piece of tubing & try to press it in with the hydraulic press at work. If it forms to my satisfaction (Test sample) I will continue on. & either use heat sink grease, or thermal transfer epoxy. Once done, I will heat the assembly & test what it can dissipate for heat.

    Now, if this works, I will press tubing into my 4-BLF278 HF amp (MRI pallet) & my Larcan 6-BLF278 6 meter amp & add them to the system.

    I will use a sacrificial zinc anode to ease electrolisys. I will also use distilled water & propelene glycol for coolant.

    What say?

    Ed
     
  2. AF6LJ

    AF6LJ Ham Member QRZ Page

    You going to use something like a length of square tool steel to apply even pressure to the tubing as it is pressed in?
    I have thought about doing this with .250 copper line.
    Connecting them together at both sides to a manifold avoids sharp bends and lowers flow resistance.
    It is also five times the work....
     
  3. AA7QQ

    AA7QQ Ham Member QRZ Page

    If it all works, I will press 1/4" tubing into the other amps. I could use copper, but the wall thickness is larger & that would inhibit flow.
    As for the bends, I will figure it out.
    As for pressing the tubing in, I was thinking of skinning some steel with uhmw. If need be, I can pre flatten the tubing in the vice, or a large stamp press we have.
    A manifold gets expensive & this amp will show just what can be done when you are a tight wad...
    Each 1-BLF278 pallet amp cost $50 shipped from Australia. No, don't look, I got the last 8 of them... 4 will be for replacement LDMOSs for my 3 amps. I LOVE surplus junk.

    Ed
     
    AF6LJ likes this.
  4. AF6LJ

    AF6LJ Ham Member QRZ Page

    The bends may be easier to make by filling the tubing with something like fine sand or baking soda. You will have to pack it tight and make sure it has nowhere to go.
    This may work....
     
  5. AA7QQ

    AA7QQ Ham Member QRZ Page

    Yeah, that is a possibility. We'll see how the Speriments work out. It's great having a whole machine shop at my disposal.

    Ed
     
    AF6LJ likes this.
  6. W0AAT

    W0AAT XML Subscriber QRZ Page

    AF6LJ likes this.
  7. AA7QQ

    AA7QQ Ham Member QRZ Page

    OK. good info there. I changed my order from .035 wall aluminum to .020-.025 copper. I do agree that the epoxy will be the way to fo, especially if I have an expansion coefficient issue.
    We will see how this all turns out.
    This will also make doing the 180* bends easier, as I can silver solder them all.
    I just didn't know they made copper with such a thin wall.

    Thank you,

    Ed
     
    Last edited: Jul 6, 2017
    AF6LJ likes this.
  8. WB1E

    WB1E Ham Member QRZ Page

    I have some Aluminum Nitride powder that I use to mix my own thermal epoxy. Hardens like a ceramic at 30% weight to resin ratio. Ed, I can send you some. Let me know.

    Doug
     
    AF6LJ likes this.
  9. K7JEM

    K7JEM Ham Member QRZ Page

    I don't think those heat sinks will have enough mass to remove the heat fast enough, not at that level. You're talking about trying to remove 300 watts or so of heat using a relatively lightweight sink. Even if the sink were immersed in cool water, I don't think it would keep the transistor package cool enough.

    I think what you will need is some sort of relatively heavy block of AL with holes drilled in it to route the water. Kind of like how a car engine is cooled.

    I suppose you could mount 300 watts of resistors on one and see how fast it heats and how hot it gets.
     
    AF6LJ likes this.
  10. AF6LJ

    AF6LJ Ham Member QRZ Page

    One way to aid in the removal of all that heat is to use a copper heat spreader, make sure the surfaces of the heatsink and the spreader are lapped to a level of flatness on the order of a tenth of a thousandth or less.
     

Share This Page