0

I'm repairing quite the pricey device, where a strain gauge failed. It's a pressurized device, and the strain gauge measures the internal pressure by means of it. The crux of the matter is that the strain gauge goes on a small metal cylinder epoxied to the PCB itself.

To install a new one, I will need to use hot curing resin, but I want to avoid having to remove the metal cylinder to reduce risks of damage and potential surprises. This means baking the whole PCB, MCU included, which worries me a lot since I have no clue whether it could become corrupted and I can't reprogram it.

I have 3 options for hot curing with Adhesive A:

  • 1 hour at 190 °C
  • 3 hours at 170 °C
  • 6 hours at 160 °C

And other 3 with Adhesive B:

  • 30 minutes at 200 °C
  • 1 hour a 180 °C
  • 3 hours a 150 °C.

Now, the max storage temperature I saw in the datasheets of some ICs on the board is 125 °C which is what is deterring me from attempting this.

I found this question that basically asks the same, but I'm looking for a more formal resource if anyone has one.

I'm also thinking if there's a possibility of insulating the rest of the PCB since the topology allows it.

Thank you very much in advance.

garvamel
  • 55
  • 5
  • I did some one-off tests where we needed to operate an MCU (from Microchip) at \$180^\circ\text{C}\$. We had to write data into EEROM, too, while depending upon the code running in flash memory. Way out of spec. But we were consistently successful running for 3-4 hours in that situation. (Threw away the parts and replaced them with new ones before the next cycle of testing, of course, since there was no way we'd trust them later.) Metal migration is an issue. Ask IC designers about it and they can provide predictions to help, if you ask. That's what I did. You are on iffy ground. – jonk Nov 09 '19 at 02:31
  • Any chance you could selectively cool the MCU during curing? – Spehro Pefhany Nov 09 '19 at 04:20
  • @jonk thanks a lot! I thought so. – garvamel Nov 10 '19 at 16:01

1 Answers1

0

A rough back of the envelope calculation indicates that a device with a minimum 20 year retention at 110’C would last about 1/256 of that time at 190’C so you’d be knocking off a year or two from the life. 200’C is a wash with the halving for 10’C approximation so I would pick the lower temperature I think. An hour at 190’C is much worse than a few seconds at reflow temperature when it comes to retention. At 230’C, 30 seconds is the equivalent of only 4 minutes at 190’C.

Of course my estimate using that rule of thumb may be completely worthless— perhaps you could procure a sample of the MCU chip, program it, bake it for (say) 10 and then 20 hours and see if it loses bits.

Or find some way to selectively cool the MCU chip.

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
  • 1
    Putting your comment together with your answer ... fi a jet of cool air can knock even 30C off part of the PCB, that widens your options (or margins) –  Nov 09 '19 at 11:57
  • Something like [this](https://contestimg.wish.com/api/webimage/57328e37b4686e5d36033f7b-large.jpg?cache_buster=0cdd8b3631c2dfb42e15b8f23ada82da) will even let you add fine mist of liquid to the air flow. It's for mist coolant on a machine tool such as a lathe or milling machine. Rather inexpensive. – Spehro Pefhany Nov 09 '19 at 23:58
  • I will investigate all of your recommendations, great ideas. Thank you very much. I ended up recommended getting a complete replacement from the manufacturer, since my client was on the clock and in need of a working part within 10 days, and since this isn't my particular area of expertise, I did not want to offer something I couldn't accomplish with certainty. Thanks a lot again. – garvamel Nov 10 '19 at 16:05