Tough PIC

[Charles Linquis]

One of my circuits uses an 18LF2523. When we were bringing up the board for the first time, a technician noticed that the circuit was working, but the voltages were off. He measured VCC and found that it was 11.5V! A diode had been installed backwards.
The PIC had been working fine for 15-20 minutes before the problem was noticed. He reversed the diode and the circuit (and the PIC) worked normally.
I didn't try all the peripherals, but the USART, the PWM and all the I/O pins I tried were working normally.

Of course, we replaced the PIC before shipment for reliability reasons.

[spcw1234]

Impressive! I might try that tonight with one of my PIC's ;)

[Ioannis]

Please put on your safety glasses, a helmet and gloves. It may blow!

Post your interesting result hen... :)

Ioannis

[Charles Linquis]

I have mentioned it before, but I once used 16C73 (I'm embarrassed to say that I ever used 16 - series chips). The 16C73 was one in a ceramic package and a quartz window (it was UV-erasable, remember those?). Because it had to be put into an eraser for 10 minutes every time I wanted to reprogram it, I was constantly plugging it into, and pulling it out of, a nylon 3M proto-board. After one such insertion, I smelled something and looked over at the proto-board. Smoke was rising! I shut off the power and (stupidly) used my finger to see just which component was overheating. I touched the PIC and got a blister on my finger. After the PIC had cooled down, I removed it from the protoboard, and the 'groove' under the chip was melted. The nylon was dark brown as well. I noticed that I had installed the PIC backwards.
I was sad that I had to throw the PIC away (the erasable ones cost $20 each back then), but before I did, I put it back in the protoboard in the correct orientation. It worked!!!

About the same time (also with a PIC16C73), I needed a 20Mhz crystal. I didn't have any, but I had lots of TTL 'can' oscillators. I didn't have any 20Mhz parts, but I did have a 24Mhz oscillator. I tried it. It worked - no problem. So I kept plugging in higher and higher frequency oscillators. It ran fine at 57.6Mhz, but didn't work at 64Mhz. I can't say that I tested everything at that frequency, but it did a fine job of reading the A/D and adjusting an EEPOT in a "tracking" mode.
VCC was 5V. Who knows what it would have done if I had raised the voltage to 11V!

So - you can overclock PICs.

[Darrel Taylor]

Here's an 18F2550 that was powered with 15V for ~15 minutes.

http://support.melabs.com/DT/18F2550_15V.jpg

A 30F3012 was on the same board. It got hot, but didn't melt like the 2550.
Neither of the chips worked afterwards.

http://support.melabs.com/DT/30F3012_15V.jpg

Keep those safety glasses on!

[Ioannis]

On the last picture, there is "melabs.com" and 30F3012 chip. Can I speculate that something you are cooking there at Melabs?

Ioannis

[Darrel Taylor]

I wish !!
No, it was just built on an melabs PicProto3 board.

http://support.melabs.com/DT/PP3.jpg

[DocDingwall]

In general, these chips can really take a beating. It's a real kindness for newbs like me.

Before I switched to ICSP, and was removing the chips from the board to re-program, I occassionally got them in backwards. It's really funny how your nose gets tuned to the "hot" smell. I never wrecked a chip but I sure warmed their behinds! Sadly micro mmc cards are not so tough, you just hear a little pop and it's game over--no magic smoke or anything.

Cheers,
Doc.

[Demon]

Yup, they really can take abuse. You may not notice anything wrong after a screw-up, but are you using all features? It's quite possible that 'something' broke.

But yeah, I've also done my share of 'Official Microchip Volunteer Stress Tester'. :D

Robert

[Art]

What would be the first problem with overclocking?
I don't think it would be heat.
Maybe reading it's own program memory to execute the program it's running?

Have run 16Fs at 25MHz many times here.

[Demon]

The biggest danger when overclocking a PC cpu is overheating. I don't see why it would be any different with a PIC.

Maybe we should liquid cool a PIC and see how fast it can run? :D

Robert

[Art]

You could cool one with a peltier device, but I doubt you could run it fast enough to get it hot.
PC CPUs don't have to read internal memory.
Pics are more than a CPU, but certainly Do go screwy when the outside environment heats up...

I have a 16F84A running in a vehicle that goes haywire in the heat every time (digital speedometer),
but resumes as normal when the vehicle cools down without needing any reset.
The 16F877A does not seem as fragile though (so far).

I'm still guessing it's the internal interface between it's core and internal memory that would fail.
Not that my theory could be tested. It's not like the CPU can be isolated to do anything without
using it's program memory.

[Demon]

CPUs do read internal cache, that may be technically equivalent to ram.

If I'm not mistaken, isn't the A series for automotive applications?
EDIT: Nevermind, I think it's low voltage programming.

Robert

[Art]

I meant flash memory where the program is stored in a pic.
The 16f84 I have had running a speedometer in a car for a few years always goes batty if the cab gets hot.
The chip is running, but the value displayed is all over the place.
This isn't any sort of proof that the chip is failing to read it's program memory though.
It could be any number of things like the program counter jumping all overt the place, but flash is still my guess.

[Demon]

What would be the first problem with overclocking?
I don't think it would be heat.
...

Art, was rereading this thread and realized I should have been more specific:

No problem with heat if you play with clock speeds only, the problem is when you increase voltage to keep the chip running. That's when you run into "opinions" from guys called Fourrier, Joule, Ohm, Watt and company.

Watts = volts x amps

All that extra energy has to go somewhere. :D

[Art]

If you ever have some hobby time to waste, set a 12C509 to toggle an LED every second,
use it's internal clock, and connect a high value potentiometer between the pic and a 12 Volt supply.
Yeah, HEAT, but result was no less than impressive :D