Flex 6600M Heat sin Issue - replacement

  • 2
  • Question
  • Updated 2 months ago
Hello Flexradio
I just installed yesterday my two new heatsinks according your installaltion procedure. One of the old heatsink was fallen already away.
I'm using an Expert 1.3k amplifier with my Flex 6600M. So I don't need the 100 W power of the flex only about 7-8 Watts. The chip on the ADC Board is getting hot in any case doesen't matter how I'm using my Flex or only if the Flex is operating with full 100W power?

What I not understand is, why to install the new black heatsink on the back of the circuit board. Because the circuit board is an isulating layer which is not taking the heat away. How this will work?

Should I find a solution to mount the two golden old heatsinks on top of the chip as before with an additional rubberband or a better glue?

I feel at the moment not so sure with this solution regarding the health of my flex 6600M.





73
Wolfgang




Photo of Wolfgang

Wolfgang

  • 22 Posts
  • 1 Reply Like

Posted 2 months ago

  • 2
Photo of David Decoons wo2x

David Decoons wo2x, Elmer

  • 1201 Posts
  • 247 Reply Likes
The new heatsink on the back of the ADC board will lower the ADC chip temperature enough for it to operate in the safe range of temperature.

I have modified my 6600 several months ago as per the instructions and I have not had problems.

Dave wo2x
Photo of Paul

Paul

  • 475 Posts
  • 143 Reply Likes
Hi Wolfgang. A while ago there was a long discussion on here about how effective (or otherwise) the heatsink would be on the back of the board. If I recall, Flex stated that there are more than enough pads and via holes to conduct the heat from the chip to the back of the board. So I assume they are confident that providing you fit it correctly there should be no further overheating problems.

Regarding your other question; the PA produces the output RF not the ADC. So it shouldn't matter what power you operate your TX at. The PA has it's own independent cooling so the two heat sources (ADC and PA devices) are dealt with by separate cooling mechanisms.

Hope this helps.
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 2971 Posts
  • 653 Reply Likes
The back plain of the ADC board is a cooling plain, it has been discovered that the cooling from the back has a much better cooling effect then on the chip itself. This is in part that the chip has a plastic coating on it that hinders cooling. If they had realized this from the designe stage they would have done it this way from the start.
Photo of Paul

Paul

  • 475 Posts
  • 143 Reply Likes
Bill, it sounds like you're saying Flex did't read the data sheet properly.  The AD9467 data sheet indicates clearly that the back of the chip is used for cooling.

Page 31; http://www.analog.com/media/en/technical-documentation/data-sheets/AD9467.pdf

The fact that Flex implemented the paddles & vias suggests that they were aware of this. They just ran out of luck when they chose to put the heatsink on top of the chip instead of the back of the board. In a way, the failure of the adhesive led them to a more appropriate (correct) solution.
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 2971 Posts
  • 653 Reply Likes
It is all part of the hundreds of choices being made in the designe. The ADC board is not a size of a cooling plane that would allow the amount of cooling needed as described in the ADC manual. For this reason a heatsink is needed. A small heat sink was added to the top of the Chip that gave plenty of cooling. But the thermal adhesive was incorrect and the heat sinks lost connection.

If the ADC board was larger a heat sink would not be needed at all. For axample, the Flex 6500 has no heat sink on the ADC chip. The entire board is a cooling back plane and provides all the cooling needed in those radios.

When Gerald worked on this I think it may not have dawned on him to add a heat sink to the back plane intead of the top of the chip. Maybe some day if he has time he can go into this more. I was not part of the designe and I don't know what he was thinking about. But I know Flex does not designe anything without much consultaion and thought.
Photo of WQ2H - Jim Poulette

WQ2H - Jim Poulette

  • 120 Posts
  • 29 Reply Likes
Hi Bill,

Clearly arm-chair quarterbacking - but still an interesting scenario.

I have thermally modeled devices like these, and have familiarity with AN-772. The issue I've seen is that the proposed 'stencil', as thin as it may be, will often cause a void between the paddle and the ground plane on the substrate. In essence the paddle never even comes in contact with the substrate - so you're forced to live with dead-air convection rather than physical conduction.

A remedy that is often used to mitigate this is to apply dots of a thermally conductive adhesive, like Eccobond 56C, in each of the stencil panes to fill the voids and join the two surfaces. I'm sure the Tc of the molding material is not good - so my guess is that heat sinks on the top would do very little.

Thank you all for this interesting thread.
73
Jim, WQ2H
(Edited)
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 2971 Posts
  • 653 Reply Likes
Un like many kinds of boards, the boards used in the flex have cooling back plans to cool components. As I mentioned, the 6500 needs no heat sinks on the ADC because the back plane cools the ADC more then It needs.
Photo of WQ2H - Jim Poulette

WQ2H - Jim Poulette

  • 120 Posts
  • 29 Reply Likes
You're right, AN-772 is pretty clear about how the correctly lay out, mount, and interconnect to these devices to provide the most efficient cooling. Choosing to do otherwise would certainly risk performance degradation, failures, or even a product re-work or re-call. We do all live and learn. Thanks!

73
Jim, WQ2H
(Edited)
Photo of Wolfgang

Wolfgang

  • 22 Posts
  • 1 Reply Like
Hi All
Thank you very much for your answer :-)
73, Wolfgang
www.hb9ryz.ch