Freezing chips for debugging

[Kohanbash]

Hi all
I have heard many people reference freezing chips/boards for debugging (including in the latest eevblog video #565).

What is the purpose of freezing the chips/boards to help with debugging?

Thanks!

[w2aew]

Hi all
I have heard many people reference freezing chips/boards for debugging (including in the latest eevblog video #565).

What is the purpose of freezing the chips/boards to help with debugging?

Thanks!

One common reason is to help isolate micro-fractures in solder joints as well as package connections. Under operation, devices that dissipate power will heat up. As they heat up, they expand, and can open up these fractures/cracks, causing the circuit to fail. Typical symptoms include a device that works when first powered up, then fails once warmed up.

Also, failing semiconductor devices can often have excessive leakage currents when warmed up, disrupting normal operation.

These are two examples where cooling down a device or region of a circuit may restore operation, thus helping to isolate the area of the fault.

[Bored@Work]

It should be noted that the technique is not without problems. Rapidly cooling a hot component from, lets say, 100 deg C to -10 deg C^* can mean that a 100% working, faultless component cracks.

Second, many cooling sprays contain flammable substances. Their autoignition temperature is around 400 deg C ^**, so it can become dangerous if you have something very hot in the system. Or if you system e.g generates sparks (switches, relays), which can ignite the gas, because its flash point is much lower than 400 deg C ^**.

There are non-flammable cooling sprays, which I prefer.

* Some sprays proudly claim they reach -55 deg C, -10 deg is a conservative figure.
** Don't quote me on this.

[Kohanbash]

Thanks for the replies

[TerraHertz]

Or, you can have thermally sensitive failures in which the device works most of the time, but fails in a narrow temperature range. Perhaps due to some fractured bonding wire or metalization on the chip.

Actually, this is a good place to ask a question about a problem I face at the moment.
I have a system (HP 54121T scope) in which one board has an intermittent thermal fault. Probably one of the DRAM chips, of which there are 16 so desoldering and socketting all to allow swapping isn't a good option.
But I have no way to exercise the board except in it's slot, in which case I can't get at the thing to use spray freeze of hot air jet.

Has anyone seen a product that consists of a pad that can be clipped to the top of one IC at a time, with wires/tubes allowing the board to be run in place and the suspect IC heated and cooled via the wires/tubes?
I'm thinking I can probably improvise something, but if it already exists and is cheap it might save some bother.

Kind of like overclocker's water cooling rigs, only for 16 pin DIPs and alternate heating/cooling.

[SeanB]

Just get some bendy straws and tape them so you can do a chip or two at a time, so you then spray into the one end of the straw at a easy to get place while the outlet is on one chip you want to chill.

[dannyf]

What is the purpose of freezing the chips/boards to help with debugging?

To tell the boss that it is not your fault that you cannot debug the board - it is broken.

[peter.mitchell]

Or, you can have thermally sensitive failures in which the device works most of the time, but fails in a narrow temperature range. Perhaps due to some fractured bonding wire or metalization on the chip.

Actually, this is a good place to ask a question about a problem I face at the moment.
I have a system (HP 54121T scope) in which one board has an intermittent thermal fault. Probably one of the DRAM chips, of which there are 16 so desoldering and socketting all to allow swapping isn't a good option.
But I have no way to exercise the board except in it's slot, in which case I can't get at the thing to use spray freeze of hot air jet.

Has anyone seen a product that consists of a pad that can be clipped to the top of one IC at a time, with wires/tubes allowing the board to be run in place and the suspect IC heated and cooled via the wires/tubes?
I'm thinking I can probably improvise something, but if it already exists and is cheap it might save some bother.

Kind of like overclocker's water cooling rigs, only for 16 pin DIPs and alternate heating/cooling.

A peltier? A peltier with a heatsink should do fine. The amount of heat a memory chip would produce could fairly easily be reduced below ambient with a heatsink and a peltier, and then just swap the polarity (i think) to heat the chip.

BUT as SeanB says, just get some of the bendy straws, no need to make this more complicated than necessary.

[tszaboo]

These are two examples where cooling down a device or region of a circuit may restore operation, thus helping to isolate the area of the fault.

Interesting. I've only seen (and did) people using the freezer to make "poor mans thermal imaging camera".
Make a "frost" on the board, and the broken IC usually heats up abnormally and melts the ice, so you see what is broken.

[TerraHertz]

Or, you can have thermally sensitive failures in which the device works most of the time, but fails in a narrow temperature range. Perhaps due to some fractured bonding wire or metalization on the chip.

Actually, this is a good place to ask a question about a problem I face at the moment.
I have a system (HP 54121T scope) in which one board has an intermittent thermal fault. Probably one of the DRAM chips, of which there are 16 so desoldering and socketting all to allow swapping isn't a good option.
But I have no way to exercise the board except in it's slot, in which case I can't get at the thing to use spray freeze of hot air jet.

Has anyone seen a product that consists of a pad that can be clipped to the top of one IC at a time, with wires/tubes allowing the board to be run in place and the suspect IC heated and cooled via the wires/tubes?
I'm thinking I can probably improvise something, but if it already exists and is cheap it might save some bother.

Kind of like overclocker's water cooling rigs, only for 16 pin DIPs and alternate heating/cooling.

A peltier? A peltier with a heatsink should do fine. The amount of heat a memory chip would produce could fairly easily be reduced below ambient with a heatsink and a peltier, and then just swap the polarity (i think) to heat the chip.

BUT as SeanB says, just get some of the bendy straws, no need to make this more complicated than necessary.

Actually if I have to improvise, a peltier is one possibility I was thinking of using. But what I asked was did anyone know of an existing product?

Nevermind, I'm pretty sure it's going to be improvise or nothing.
As for bendy straws, I'm not sure that would work. Spray freeze cools at the place it expands and evaporates. You'd just be freezing the straw, not the end where the chip is. Besides, I don't like to use spray freeze. Too wasteful and environmentally iffy.
There's an alternative, since I have some big dewars for liquid N. Just need to work out a way to apply small controlled amounts and have it evaporate at the chip.

[all_repair]

Normally I try heating, less wasteful.  Heating like blowing hot air over it and using a hot air station is the easiest.

[wilheldp]

Normally I try heating, less wasteful.  Heating like blowing hot air over it and using a hot air station is the easiest.

Wouldn't heating a chip that is failing due to overheating be counter-productive?