For science on a Sunday

[ScottishColin]

I ran a test today for 9 hours on my 3016 just so none of you had to - I ran the PETTESTER v 04 ROM all day today to test speed and temperatures.

In summary, there were 267 RAM tests run in 9 hours which equates to 29.67 tests/hour (or pretty much 1 every two minutes).

Here's a table of temperatures. Nothing was plugged into the PET - all devices were removed. It was running with the original 6502 and the BASIC v 2 ROMs (with the exception of the test ROM). UD7 has been replaced with an EPROM and I have the temperatures for that too (although they're not any different to the other ROMs).

The room temperature was measured as 19.8 c at the beginning. At all times expect when measurements were taken, the PET lid was closed and no keys were pressed.

The thing that piqued my interest was the difference in temperature between Row 2 (UI) of the RAM chips and row 1 (UJ) of the RAM chips - do the RAM tests exercise the UI row of RAM chips more than the UJ row?

I guess for this to even approach any kind of scientific rigour, I should load a 16K game and leave that running to see if there's a similar difference in temperature with that running too.

Any - this may well all mean nothing, especially with the number of replacement chips and sockets in my PET - but the RAM chips are the originals - any comments welcomed.

Device	Starting temperature	End temperature after 9 hours
6502 (original chip)	20.1 c	31.9 c
UD6 (original chip)	20.3 c	37.0 c
UD8 (replacement - PETTESTER v04)	21.3 c	42.0 c
Top of the transformer - original	21.3 c	44.2 c
Row 2 RAM (UI) - all original chips	21.4 c	36.7 c
Row 1 RAM (UJ) - all original chips	21.2 c	32.3 c
VR3 - original	22.1 c	45.6 c
VR4 - original	22.0 c	51.6 c
VR5 - original	22.2 c	49.7 c

 

[Hugo Holden]

Very cool!

Sometimes thermal data can be helpful in fault finding. Many IC's generate more heat when they are operating at high frequencies vs not and have multiple stages in them. If they fail internally their temperature can go up or down compared to normal, depending on what went wrong in them. There are some IC's that are prone to a latch up state, some OP amps do this. If they get into it they start drawing excessive power from the supply and heat up significantly. They can be "reset" by switching off the supply and switching it back on again. After they have done that though I never trust them and discard the IC. Recently I found some LM324 OP amps that did this, but I have heard of it in other IC types.

An IC can run slightly hotter in a socket as there is a little less heat conduction away from its pins into the pcb track-work.

It would be a useful thing to have an infra-red photograph of the pcb in the PET, just in its booted state, as a reference to show the relative temperatures of the voltage regulators components & IC's. It could come in handy one day.

 

[daver2]

Excellent (and interesting) work.

I'll have to borrow an IR camera from work...

The UI row of memory is the lower 16K bank whilst the UJ row of memory is the higher 16K bank.

Extensive use is made of page 0 and page 1 (stack). So, for every memory access, an access to a double-byte memory address in page 0 has to be performed. Also, the mask byte has to be updated and this will result in the memory address in page 0 being incrementing or decrementing every eight memory bits tested.

From this knowledge - it is not surprising that the temperature for the UI bank of devices is slightly warmer than the UJ bank (assuming the temperature is a function of the number of times the device is accessed and hasn't peaked out (as a result of the heat production being equal to the heat dissipation)).

Dave

 

[Ruud]

Very interesting!

I know that heat is not good for ICs. But what is acceptable and what is even bad? I just know this: the cooler, the better. Would you mind to perform the test with a little fan inside the 3016? Preferably one that sucks cool air from the outside but it would already be interesting (maybe even more) if the fan just keeps the air moving inside the case. The idea behind the last thought: the case cools the moving air.

Thank you in advance!

 

[ScottishColin]

Update. 3 hours in. I'll update more tomorrow. UD7 is strangely colder but it's a replacement one from Sirius; maybe that affects it.

 

[Hugo Holden]

Update. 3 hours in. I'll update more tomorrow. UD7 is strangely colder but it's a replacement one from Sirius; maybe that affects it.

Wow, great photos .

It looks to me like resistance has developed on one of the power connector fittings, the 4th wire up from the bottom of the connector in the photo. When this happens they can heat up. It would be interesting to clean that connector plug and remove any oxides from the pcb pin, slip it out of the plastic housing if possible , re-tension it for a tighter contact and re-do the IR photo. You could also, before doing this measure the DC voltage drop across it.

In essence with this photo you have captured a failure in evolution. So it goes to show, that IR photos could be a technique to detect failures in system critical equipment, before the failure actually occurs. It is like the movie Minority Report, where they detect the crime before it happens.

Can you tell me what model camera did you use for these images ?

 

[SiriusHardware]

Actually, that particular connection on that JP8 connector has already given Colin trouble in the past. I thought Colin had chosen to bypass it with a wire. Are those male / female connectors still available NOS anywhere? They could do with being replaced. As well as Colin's I have seen one other case where that particular connection (pin 4 of JP8) went high resistance. It's in the path between the +9V output of the bridge diodes and the big offboard capacitor. My theory is that the JP8 pin 4 connection gets hammered by the huge inrush into the 23,000uF offboard capacitor at switch-on.

UD7 is cooler because it is an EPROM replacement for the original UD7 (basic ROM 2 of 2) which had failed. The other ones are original Commodore PROMs. I would say this could be a good argument to remove and preserve the original ROMs and replace them with cooler running EPROM equivalents.

 

[ScottishColin]

Sirius - I have spares to be able to replace the connector (both the plastic housing and metal connectors) but I have been leaving it as the PET works.

The only reasons I'd change it is if it's either dangerous or has potential to damage the PET; do you think the temperature shown is enough of an issue for me to change it?

Colin.

 

[ScottishColin]

I bought a Topdon TCView TC001 camera to do this (well, it's really to work out where the heat is escaping from the house). It plugs into the USB C socket on my Android (Pixel 6) phone and uses an app on the phone. You can plug it into a computer via a USB cable and use the PC instead.

Only Day 2 of owning it, but it's interesting and looks like it should help.

Colin.

 

[ScottishColin]

Ok - I ran the tester for 17 hours (1f6 successful tests) and have stopped now. Photos attached. A slight increase in temperature but only by a few degrees that I can see.

I have replaced the ROMs with new BASIC 4 EPROMs that I burnt and once it has cooled down, I will load Invaders as it has a demo playing if I don't touch the keyboard to see if that makes any difference.

As a reminder, my PET has had many chips replaced during its resurrection so it may vary wildly if you're doing this at home with your PET with original chips:

UA7
UA8
UB2
UB3
UC3
UC4
UC9
UE7
UE8
UE10
UF7
UF8
UF9
UG5
UH1
UH3
UH6

plus two sockets
UC4
UD9

 

[daver2]

Colin,

That is real cool work . Yes, pun intended!

Thanks for sharing what your configuration is. I was looking for a thermal attachment to one of our portables for a while now (as we have sent our camera at work to one of our other sites).

Mainly I was also going to use it to look for escaping heat from the house... But, if we have some experience with PETs as well!

Dave

 

[Hugo Holden]

A couple of things in those photos that seem interesting and unexpected to me.

I would have thought that the regulator bodies would be hotter than the heat sink, due to the thermal resistance of the regulator's body to the heatsink, at least for the hottest running regulator there, the 12V one I think.

It looks like the two 5V regulators are acquiring heat from the heatsink that is hotter than they are, due to the 12V regulator heating up the heat sink.

This appearance of the 12V regulator body top could be explained by the TO-3 body shell and upper surface is losing heat by radiation & convection and fractionally lowering the top surface body temperature with respect to the mounting flange of the TO-3 body and the heat sink.

If the camera could see the junction in the regulator, it would (should) be hotter than the heat sink.

If this is the case, when we measure regulator temperature, in a TO-3 case device, it might be better to do it on the heatsink near it or the flange of the TO-3 case, but not the top surface.

But one very striking thing in the images:

The lead wires on the diodes on the bottom corner of the pcb appear to be significantly cooler than the diode body and pcb area and show up as blue.

This suggests the pcb area there has heated up from radiation from the diodes, not from conduction of heat from their lead wires. Most cooked pcb's I have seen where axial power rectifiers are, appear to have sustained more heat damage around the solder pads where the diode leads are, suggesting it was heat conducted down the wires the cooked and browned the pcb.

I have managed to cool axial diodes by attaching fins to their lead wires. Time to revise the belief system perhaps? And it would be better to apply the heat fin to the diodes plastic body ?

But in this case I have thought of a possible optical reason the diode leads "look cold" in the photo and the photo might not be "telling the truth".

It could be of the shape of the diode leads I think. We are looking at a light photograph, albeit IR. With the small round lead cross section every point on its surface, if we consider the radiation leaving it perpendicular to the surface, this spreads the light energy out. By the time it arrives at the camera only a small amount of the radiation at the lead temperature (and frequency) is captured and, if it is too small, then lower temperature (frequency) background waves could dominate the image. In addition, the hotter area of the pcb below the lead, will be masked by the diode lead on top of it, making those leads look "cold" when they were not. In effect the diode lead shape has refracted the IR light away from the camera and appears like the equivalent of a shadow. (just a hunch here). If it was true it might be possible to make an IR cloaking device for a fighter jet by a applying a myriad of hemi- spherical shaped or hemi-cylindrical pattern work to its surface, to make it IR invisible.

The way to tell would be to take a zoomed up photo of those diodes so it fills a much larger part of the image and see if it changes the apparent temperature of those diode leads or not ?

I just checked on my PET and I got 57.7 deg C on the diode body and 56.7 deg C on the lead wire, as one might expect, as heat is being conducted from that lead to the pcb.

 

[SiriusHardware]

do you think the temperature shown is enough of an issue for me to change it?

An ideal (or normal) plug and socket connection has near zero resistance. Heating of any such connection indicates that it has higher than zero resistance and is not functioning correctly. It could be the pin to socket connection which is oxidised / high resistance or it might also be the connection between the wire end and the knife-edge contacts on the top of the plug which the wire was pushed down when the cable assembly was originally made. For anyone suggesting the pin and socket just need to be cleaned up, I think that has already been tried as far as possible in this case. If you showed a normal photo of the JP8 connector body shell (as you have in the past elsewhere) we would see that it is browned / heat damaged where that connection is.

What concerns me is that sudden complete failure of that connection high resistance or open circuit will remove the +9V input to the two +5V regulators leaving no +5V supply but all of the other supply rails still up - that would be a dangerous state of affairs which I know has already happened to you once and somehow everything survived, but you don't want to push your luck.

If you aren't ready to replace the connectors then I would suggest you reinstate your bypass wire going from the +9V output of the two diodes to the + of the offboard capacitor as you have done before. I know it's a nuisance having to disconnect the bypass wire every time you want to take the PCB out for any reason, maybe you could insert a car type 'bullet' male / female connector in the bypass wire to make it easier to disconnect and reconnect.

 

[Hugo Holden]

Sometimes the crimp to the wire can be poor and as noted the resistance can be up there. I never rely on crimps on their own and solder the wires to the connector pins when I replace the connector or build something using one.

I had a diabolical one in my SOL-20, one of the feeds with a spade terminal, connected to the chassis mount bridge rectifier in the supply was poorly crimped from new. Covered in the usual insulation so difficult to see.

They had basically crimped onto the wire's insulation and the bare wire was sitting in there just touching the connector. The power would intermittently crash, but only for brief episodes. It corrupted some floppy disks, the rapid power cycling upset the N* controller card. The thing was though, it did not know initially it was power cycling doing it, as there were no LED's anywhere.

But later I had installed a small LED voltmeter and when the fault cropped up, I was lucky to see it blink off momentarily and then I knew what the problem was. But it was a lot of hunting and inspection to find the defectively crimped terminal. At one point I wondered if it might be defective terminations in the power transformer, but fortunately it was not.

All in all, I'm not much of a fan of crimping. It can work very well if the specific tool suits the connectors & wires, but a lot of the time the wrong tools and wire for the connector get used and there is hell to pay later.

 

[SiriusHardware]

I'm also not a fan of crimped connections, they were the bane of my life when I was a field service engineer for 13-14 year period. When I'm replacing a crimped - on contact on the end of a wire I lightly tin around 3mm of the end of the wire and then use fine nosed pliers to carefully fold the crimp part of the connector tightly around the tinned wire end, then I apply a further dab of solder to the crimp area externally such that it runs down inside the hollow of the crimped area. It's a good idea to do this with the crimp connector pointing at the ceiling and the trailing wire pointing at the floor so the solder doesn't run into the hollow female part of the crimp contact, rendering it useless.

 

[Hugo Holden]

I'm also not a fan of crimped connections, they were the bane of my life when I was a field service engineer for 13-14 year period. When I'm replacing a crimped - on contact on the end of a wire I lightly tin around 3mm of the end of the wire and then use fine nosed pliers to carefully fold the crimp part of the connector tightly around the tinned wire end, then I apply a further dab of solder to the crimp area externally such that it runs down inside the hollow of the crimped area. It's a good idea to do this with the crimp connector pointing at the ceiling and the trailing wire pointing at the floor so the solder doesn't run into the hollow female part of the crimp contact, rendering it useless.

Yes, Gravity is your friend, it can oppose the surface tension effects that drag solder into places you don't want it. In other applications on bigger objects polyimide tape does just as well.