Thermal paste for +2 ULA?

For experts to discuss very technical stuff and newbies to ask why the Spectrum they bought off ebay doesn't work.
Post Reply
User avatar
Alessandro
Manic Miner
Posts: 630
Joined: Wed Nov 15, 2017 11:10 am
Location: Messina, Italy
Contact:

Thermal paste for +2 ULA?

Post by Alessandro » Wed Feb 26, 2020 10:09 pm

Recently I decided to dig my faulty +2 out again and examined its behavior. The issues - graphic and memory corruption while pressing the keys of the 0x7FFE port half-row, up to crashing and reset - seem to point to an overheating of the ULA. I then decided to remove the heatsink from the chip and noticed three very thin drops of thermal paste, probably too scarce and old to be effective:

Image

A new layer of thermal paste could solve the problem - hopefully! I am reluctant to undertake the replacement of the +2's capacitors at this time, although I am confident that with some exercise and advice from my Dad, who's been working with electronics for 50 years and more, I will be able to do it myself. Just in case, I mapped the capacitors according to the list on this page. There is no schematic of the Z70700 issuse 1 available yet unfortunately. Here it is for anyone who could be interested:

Image

Back to the thermal paste on the +2's ULA: did someone already face the same situation? Has anyone some advice about which paste would be suitable for the job? Thanks.
0 x

User avatar
1024MAK
Manic Miner
Posts: 683
Joined: Wed Nov 15, 2017 2:52 pm
Location: Sunny Somerset in the U.K. in Europe

Re: Thermal paste for +2 ULA?

Post by 1024MAK » Wed Feb 26, 2020 10:40 pm

Any good quality thermal paste / heatsink compound / heat transfer paste / thermal compound / silicone grease will do.
For example like this Silicone Heat Sink Compound. Although you only need a little bit, so you will never use 25g worth! It is worthwhile shopping around.

A solvent cleaner should be used to clean both surfaces. The objective of heat sink compound / grease is NOT to act as the filling in a sandwich, but to fill in all the small holes and unevenness in the surface of the chip and in the surface of the ,
metal heatsink. This increases the surfaces that are in contact, hence increasing the amount of heat transfer.

So, put a very thin smear (less than 0.25mm in depth) across both surfaces with your finger. The actual semiconductor chip is in the centre of the plastic chip package and is only around one cm square. Hence it does not matter about heat sink compound going to the far ends of the plastic chip package, or going to the very edge next to the pins. If you put too much on, it will ooze out the sides when the heatsink is fitted. If so, just clean the excess off with a clean tissue.

Then the heatsink should be fitted. If the heatsink is not tight, adjust the fixings until it is very firmly pressed onto the chip.

Mark
0 x

User avatar
Guesser
Microbot
Posts: 136
Joined: Wed Nov 15, 2017 2:35 pm
Contact:

Re: Thermal paste for +2 ULA?

Post by Guesser » Wed Feb 26, 2020 11:08 pm

I just put a small amount in the middle of a chip and squeeze it out with the heatsink when applying compound. Smearing it around just makes a mess in my view :)
0 x

User avatar
Ast A. Moore
Dynamite Dan
Posts: 1464
Joined: Mon Nov 13, 2017 3:16 pm

Re: Thermal paste for +2 ULA?

Post by Ast A. Moore » Thu Feb 27, 2020 12:26 am

Hmm. Wouldn’t a chunk of metal stuck to a heated IC inside of a closed, unventilated case just increase the thermal mass of the whole thing? Sure, it’ll take a few seconds longer to heat up, but it’ll also prevent that “sandwich” from cooling down quicker.

If would make sense if the heatsink’s surface was much, much larger, and the whole case had better ventilation. I can’t see the fins in the picture, as the heatsink is turned upside down, but it seems it’s just a piece of a U-shaped aluminum profile. Or is this part of a special two-piece passive heatsink?
0 x
Every man should plant a tree, build a house, and write a ZX Spectrum game.

Author of A Yankee in Iraq, a 50 fps shoot-’em-up—the first game to utilize the floating bus on the +2A/+3,
and zasm Z80 Assembler syntax highlighter.

User avatar
Alessandro
Manic Miner
Posts: 630
Joined: Wed Nov 15, 2017 11:10 am
Location: Messina, Italy
Contact:

Re: Thermal paste for +2 ULA?

Post by Alessandro » Thu Feb 27, 2020 9:51 am

There are no fins of sort, it's just a long and thin U-shaped aluminium heatsink in fact (see the thread linked in the first post).

If it was designed that way from the start (we are talking about the case hosting a standard Amstrad-manufactured Spectrum mainboard, not an ATX one with multiple fans in fact), and until now, to my knowledge, nobody has ever reported a similar issue, then probably it's enough to substitute the old and thin paste with a layer of new, more efficient substance.

I will get some thermal paste, put the heatsink in place, and see what happens.

Edit: would a new heatsink with fins be better? Here is an example of finned heatsinks placed into a C64. However this poses the problem of keeping the heatsink firm in place. I could use some thermal adhesive tape but it won't be as efficient as thermal paste. The ubiquitous Chinese-made tape seems to be a waste of money (see here).

Re-edit: this seems a viable solution.
0 x

User avatar
Ast A. Moore
Dynamite Dan
Posts: 1464
Joined: Mon Nov 13, 2017 3:16 pm

Re: Thermal paste for +2 ULA?

Post by Ast A. Moore » Thu Feb 27, 2020 11:03 am

Alessandro wrote:
Thu Feb 27, 2020 9:51 am
There are no fins of sort, it's just a long and thin U-shaped aluminium heatsink in fact (see the thread linked in the first post).
Yeah, I don’t think it’s going to be doing much in that case. It is more likely, however, that the problems you describe are due to a bad solder joint. Thermal expansion/contraction will cause a solder joint to lose conductivity intermittently, whereas a ULA misbehaving when it heats up suggest there’s a fault with the ULA itself and no amount of heat sinking will remedy the situation—the chip will have to be replaced.

You could try narrowing down the issue by taking the ULA out of its socket, cleaning the pins and the socket with isopropyl alcohol, and reinstalling the ULA. Then, after powering the computer on, try wiggling the ULA or any other chips and see if that causes a fault to occur. You could also try to gently bend the board. A dry solder joint could be anywhere, not necessarily under or around the ULA.

If you have a spare 128K/+2 ULA (they are identical—the Amstrad-branded ones are just relabeled Ferranti chips), try swapping them around.
0 x
Every man should plant a tree, build a house, and write a ZX Spectrum game.

Author of A Yankee in Iraq, a 50 fps shoot-’em-up—the first game to utilize the floating bus on the +2A/+3,
and zasm Z80 Assembler syntax highlighter.

User avatar
Alessandro
Manic Miner
Posts: 630
Joined: Wed Nov 15, 2017 11:10 am
Location: Messina, Italy
Contact:

Re: Thermal paste for +2 ULA?

Post by Alessandro » Thu Feb 27, 2020 11:23 am

The +2 was in almost pristine condition when I bought it so I exclude dirt having accumulated under the ULA chip. It still smells as new in fact! I don't have a spare ULA, and replacements at £ 30 with postage are not exactly cheap.

Furthermore, I noticed that the issue does not arise under every circumstance. For example, the BASIC prompt seems to be immune to it. 128K games are affected, while some 48K ones aren't.

If the culprit is a dry solder joint, and assuming the ULA has not been damaged yet, it will be hard to find anyway. Maybe I should look near the ULA? I must learn how to use a multimeter, it seems... :(
0 x

User avatar
Ast A. Moore
Dynamite Dan
Posts: 1464
Joined: Mon Nov 13, 2017 3:16 pm

Re: Thermal paste for +2 ULA?

Post by Ast A. Moore » Thu Feb 27, 2020 12:02 pm

Alessandro wrote:
Thu Feb 27, 2020 11:23 am
I noticed that the issue does not arise under every circumstance. For example, the BASIC prompt seems to be immune to it. 128K games are affected, while some 48K ones aren't.
Might not be the ULA, then.
0 x
Every man should plant a tree, build a house, and write a ZX Spectrum game.

Author of A Yankee in Iraq, a 50 fps shoot-’em-up—the first game to utilize the floating bus on the +2A/+3,
and zasm Z80 Assembler syntax highlighter.

User avatar
1024MAK
Manic Miner
Posts: 683
Joined: Wed Nov 15, 2017 2:52 pm
Location: Sunny Somerset in the U.K. in Europe

Re: Thermal paste for +2 ULA?

Post by 1024MAK » Fri Feb 28, 2020 9:36 pm

Heatsinking semiconductors and airflow inside a case are more complex than people think.

In the case of a ZX Spectrum +2 (gray), the ULA is the same as the one in a ZX Spectrum 128K (toast rack) but Sinclair did not fit a heatsink.

The purpose of a heatsink is to more effectively transfer heat from the semiconductor to the air. By moving heat away from the semiconductor faster than if no heatsink was used, the temperature rise of the semiconductor is slower and will level out at a lower value. But only if the heatsink can continue to transfer heat to the air at a sufficient rate and if the temperature difference between the heatsink and the air is sufficiently large.

How fast the air inside a case rises in temperature depends on many factors, including intended ventilation slots, cut-outs for connectors (which allow limited ventilation), the size, shape and volume of the case, the material the case is made from, the ambient air temperature outside the case plus what other heat sources (both inside and outside the case) are also adding to the temperature rise.

Did I mention that it’s complex?

Anyway, as all we are doing is refitting the original heatsink as provided by Amstrad, all of the above is rather mute as there are countless working ZX Spectrum +2 machines that don’t suffer from heat related problems.

I do agree however that a likely cause of random faults is poor connections. Either due to poor contact between a socket and a pin, a damaged socket contact, a dry solder joint, a crack in a PCB track, or possibly a short circuit caused by a fine solder splash (some can be as thin as a human hair).

So use a good magnifying glass under a good light to visually examine the board on both the top and bottom.

Mark
0 x

Post Reply