• Please review our updated Terms and Rules here

Memory chip temperature observation

Scruit

Experienced Member
Joined
Apr 9, 2022
Messages
105
No conclusions here, just and observation regarding my black screen C64...

I measured the temps of the various chips with a cheap infra-red non-contact thermometer, and found the Kernal ROM to be running at 160F (yikes), and the memory chips to be between 90F and 118F.

A subsequent test of the memory chips in my dramarduino showed that the 3 chips that were in the low-90s in temp passed the testing. The 5 that were in the 105-118F range all fail testing. Repeating the memory test the next morning, to eliminate temperature as an acute variable, did not change which chips tested good or bad.

So, thought it was an interesting observation and maybe useful for quick-and-dirty triage.
 
That's the famous finger test! Where you put your finger on memory chips and try to feel if one is significantly hotter that the rest. And you are right, it always denotes a faulty chip
 
That's the famous finger test! Where you put your finger on memory chips and try to feel if one is significantly hotter that the rest. And you are right, it always denotes a faulty chip

The "Fingometer" test shows that the Kernal ROM was definitely suspect. The Ram chips were all roughly the same, though, and I was having trouble telling the difference between a chip that was warm because it was generating heat, versus a chip that was warm because my finger was on it a few seconds ago.

Oh, and the 160F kernal rom gives an "overcurrent protection" error in my TL866II+ eprom programmer. A multimeter shows that one of the data lines (D1 / pin 10) is shorted to 5+ (pin24). 12 ohms between them.
 
Sometimes when IC's fail they can heat up. other times they cool down.

The dissipation heat in logic IC's tends to increase the faster they are clocked. Cmos IC's particularly because in a static state the current consumption is extremely low. However even TTL's will run hotter with rapid clocking. If you have a group of IC's in some frequency divider chain, the more rapidly clocked ones at the beginning of the chain will run hotter than the ones further down the line.

An infra-red photo or temp readings on the IC's can be helpful, but it really requires a reference image from a known working normal pcb, unless one of them is really cooking.
 
Sometimes when IC's fail they can heat up. other times they cool down.

The dissipation heat in logic IC's tends to increase the faster they are clocked. Cmos IC's particularly because in a static state the current consumption is extremely low. However even TTL's will run hotter with rapid clocking. If you have a group of IC's in some frequency divider chain, the more rapidly clocked ones at the beginning of the chain will run hotter than the ones further down the line.

An infra-red photo or temp readings on the IC's can be helpful, but it really requires a reference image from a known working normal pcb, unless one of them is really cooking.

Understood, and I appreciate your more learned insight. Once I get this board up and running I'll re-measure things (getting known good waveforms on address/data/clock lines etc) and I fully expect the memory chips to run at a different temp than I observed here. I thought the memory / temp observation was interesting and useful in direct comparison for the ram chips to each other. I'm not aware of a reason why good ram chips would naturally run at different temps than each other at any given moment, but that may be my inexpereince. I understand it as each chip is responsible for one bit, and that all 8 chips are accessed for every read/write of any memory address, so I don't believe it is a case of "This 8k block of memory is not currently being used so one of the ram chips is going to be cooler than the others."

With the bad kernal rom there was no life on the board. The good chips were doing nothing (cooler) and the bad chips were doing nothing but apparently partially shorted in some way. I expect "good chips that are actually doing something" to run at a different temp that "good chips dong nothing", maybe even higher than "bad chips doing nothing"

The bad ram chips all passed the portion of the DramArduino test where a "0" was written to it and read back. However, the tests where a "1`" was written it also read back a "0". My hyper-simplistic, self-taught-in-electronics brain is telling me that the data out pin is being pulled down to ground, and maybe that pin is just shorted. I will grab those chips and do some multimeter testing to see if pin combinations given any significant differences in readings. I also socketed the ram slots on the board and before I put the new ram chips ones in I will put the old chips back in their original positions and repeat the multimeter tests and see how the readings change while in place with other components. Maybe the bad chips give readings that would be helpful when triaging other boards.

Or maybe the bad kernal rom and pla will impact the readings... I'll have to compare again once the good pla arrives...
 
Last edited:
Funny thing, the Kernal was scalding hot in my board too. Like OUCH can't keep your finger on it.
 
Back
Top