Inquiry about passive heatsinks for the ACA1234 (future upgrade to 50 mhz)

diegoarevalo · Feb 11th 2022

Hello, nice to greet you. I wanted to make an inquiry.

I was thinking of purchasing the ACA1234 upgrade.

I have an ACA1234 at 33mhz and I was thinking of taking it towards the value of 50mhz.

Before starting to carry out the tests that the ACATOOL allows to carry out, in order to verify if there are instabilities, I was thinking of adding some passive heat sink.

I imagine that due to the increase in frequency, more heat will be generated, which must be dissipated.

May I ask you if it is necessary to add a copper heatsink to the 68030 microprocessor?

I was looking at a 14x14x5mm copper passive heatsink model for Raspberry. What do you think about this option?

Would it also be appropriate to use an aluminum heatsink?

Would it be necessary to add a heat sink to another component to ensure proper operation?

What would be the height of the heatsink so that it can close the trapdoor? (I think 5mm thickness would be enough)

Thank you very much for the answers.

Kind regards, Diego

Jens · Feb 11th 2022

Quote from diegoarevalo

I imagine that due to the increase in frequency, more heat will be generated, which must be dissipated.

It's a small increase, but I have never actually measured it. If the mask if F91C, you probably get away without an additional heat sink.

Quote from diegoarevalo

Would it also be appropriate to use an aluminum heatsink?

Absolutely. No need to go fancy.

Quote from diegoarevalo

What would be the height of the heatsink so that it can close the trapdoor?

I don't have a plastic 68030 at hand, but the ceramic stands 5mm above the PCB surface. The white connector stands 9mm above the PCB surface, so you have 4mm if you want to be equal to that.

Since you won't really have air flow with the trap door closed, you might want to gi for a piece of sheet metal. This will spread the heat and allow for better irradiation. Fins won't really do anything if you close the trapdoor.

diegoarevalo · Feb 12th 2022

Thank you very much Jens for the answer.

I was looking at the ACA 1234 - 33 mhz and I find it has a label with the code G40W. Is this the mask number?

I hope I can get it to 50mhz.

So I decided to put the heatsink on the 68030. It is true what you say, if I put the trapdoor on the Amiga 1200, the air flow is interrupted.

I have found on Thingiverse some Trapdoors that have slots. I'm going to print some of them and I'm going to put them on the Amiga 1200.

Thank you very much for the advice,

Kind regards, Diego Arévalo

Jens · Feb 14th 2022

Quote from diegoarevalo

I was looking at the ACA 1234 - 33 mhz and I find it has a label with the code G40W. Is this the mask number?

Yes, that's a 1.0µm mask. It will most likely work without a heat sink up to 40MHz. 50MHz operation in the black, plastic package is probably unstable - worth a try, but I'd really give it a good test run and critical observation.

diegoarevalo · Feb 14th 2022

Thank you very much Jens for the answer.

To test it then at 50 Mhz, what types of tests and critical observation can be done?

Using this 50 MHz frequency could damage ACA 1234?

Or would it just be some instabilities which would not represent permanent damage?

I wanted to know this so I could handle myself in a safe range.

If it represents a high risk, I will only do an upgrade to 40 MHz. It all depends on how ACA 1234 performs in the tests.

Thank you very much, best regards, Diego

thebajaguy · Feb 15th 2022

Taking the 40MHz plastic-encased CPU past ~10% overclock is starting to approach the design limit of the chip package - And by that, I mean it's ability to dissipate heat from the internals of the chip. I have known some to succeed, so it's not impossible. It's a YMMV situation. What Jens is suggesting is to make sure it's stable at higher speeds in the long-term. This means closing up the unit and letting it run something fairly 'busy' for awhile so that the entire case can heat up inside and not affect the stability of the CPU.

Jens · Feb 15th 2022

Quote from diegoarevalo

To test it then at 50 Mhz, what types of tests and critical observation can be done?

This question comes up really often - and the answer is always "load the bus!". That means: The more memory accesses you make, the busier it will be. Further, you should switch on all caches and burst, as burst operations are the most critical ones.

Install a compiler and "make clean" a large project. Create an LHA archive of your boot partition in RAM: and test it over and over again (endless loop in a shell script over night). If it survives that without a crash, you can be sure that the setup is stable at the chosen speed.

Creating a few AIBB modules is also a good exercise, although this can't be done in an endless loop. Street wisdom appears to be that if a card can make two AIBB modules in a row, it's stable. I personally prefer the overnight test, though.

Quote from diegoarevalo

Using this 50 MHz frequency could damage ACA 1234?

No. Worst thing that can happen is a crash, and you'll be back to normal either by using the "safe" jumper setting, or reducing the speed with ACAtool.

diegoarevalo · Feb 15th 2022

Thank you very much Jens for the answer.

I have put some aluminum heatsinks covering the surface of the 680030. These have a thickness of 5 mm. I have made a 3d impression of a trapdoor with slits. Here he left the photos of how this turned out.

I tell you that I use adapted PC power supplies, and I see that these have sometimes strange behavior depending on the load that is applied to them.

For example, a PC power supply powering an A500 plus, with the ACA 500 PLUS + ACA 1234, takes almost 1 minute to boot up. And with an original A500 power supply, to which I changed the capacitors, it starts up in just a few seconds. The difference is abysmal. The floppy drive also behaves strangely.

In my Amiga 1200 so far I had no problems with the PC power supply, but it may be because so far I have not loaded it too much.

I was thinking that it seems to me that it would be convenient to change the power supply before overclocking the microprocessor.

I think I'm going to wait for you guys to start selling the Amiga power supply, to buy one, and do these tests with a good power supply.

Thank you very much, best regards, Diego.

diegoarevalo · Feb 16th 2022

Hi!. My queries are the following:

Would it be convenient to use a power supply like the one you develop at Icomp, to be able to use the ACA1234 accelerator with this increased frequency in the microprocessor?

Would the modified PC power supply be able to ensure the correct currents without voltage drops, when demanding it with a higher load?

The problem that occurs is that one may be believing that the ACA1234 is having a certain behavior, but in truth it is caused by the use of an improper power supply.

Thanks and regards, Diego

Jens · Feb 16th 2022

Quote from diegoarevalo

Would it be convenient to use a power supply like the one you develop at Icomp, to be able to use the ACA1234 accelerator with this increased frequency in the microprocessor?

"convenience" is probably not the word you're looking for - except if you're looking at features like "switch is near the computer itself" and "no need to open the computer for installation". I generally do not recommend using PC power supplies, but if you have one that is DC-DC converter based (where each rail gets it's own regulation) and the path from PSU to the main board is really short (max 20cm of a good-grade copper cable), you can use it.

From a technical point of view, I haven't seen any PSU with 5-pin square DIN connector that complies with all Commodore specs - except for the CA-PSU.

Quote from diegoarevalo

Would the modified PC power supply be able to ensure the correct currents without voltage drops, when demanding it with a higher load?

Only with a very short cable between PSU and computer. YOu cannot achieve this in the standard A1200 case. I only see this working in a tower setup.

Quote from diegoarevalo

The problem that occurs is that one may be believing that the ACA1234 is having a certain behavior, but in truth it is caused by the use of an improper power supply.

This is often what we are seeing here: The "last thing installed" is ours, and it causes the machine to mis-behave. However, it's often a voltage drop in the cable of the "new PSU" that causes trouble.

diegoarevalo · Feb 16th 2022

Thank you very much for the answer, it helps me a lot.

Kind regards, Diego