ZX Spectrum ULA

[payty]

Hi,

I am trying to figure out a method of cooling the ULA of one of my Spectrums (rubber key case).

One method with good results is to reduce the heat inside the case by replacing the linear regulator.

Personally, I do not think the heat sink on the ULA is a good method because the rubber key case has a poor air flow inside and I do not like the idea of a soldered, non removable ULA.

So I've been wondering if there is something that can be done with the two VCC pins that are powering the ULA.

One is pin 14 which powers the peripheral cells that need to run at 5V.

But the other, pin 13, powers the internal cells that run at 0.85-0.95V.

Is it possible to lower the power at pin 13 with a small switching power module in order to lower the power that needs to be dissipated by the internal regulators of the ULA?

Would that give any benefits (in case the ULA would work ok with such modification)?

[rastersoft]

Although the air flow is not great, adding a heatsink to the ULA will help a lot. The ULA gets VERY hot, much more than the inner air, so helping to dissipate that heat quickly always will be good. Also, if you replace the linear regulator with a switched one too, the interior will be much cooler, and thus the heatsink will work much better.

Remember that air is a very bad heat conductor, so even if the air inside is not moving a lot, adding a heatsink will greatly help to move that heat from the chip into the air.

[1024MAK]

Hi,

I am trying to figure out a method of cooling the ULA of one of my Spectrums (rubber key case).

One method with good results is to reduce the heat inside the case by replacing the linear regulator.

Personally, I do not think the heat sink on the ULA is a good method because the rubber key case has a poor air flow inside and I do not like the idea of a soldered, non removable ULA.

So I've been wondering if there is something that can be done with the two VCC pins that are powering the ULA.

One is pin 14 which powers the peripheral cells that need to run at 5V.

But the other, pin 13, powers the internal cells that run at 0.85-0.95V.

Is it possible to lower the power at pin 13 with a small switching power module in order to lower the power that needs to be dissipated by the internal regulators of the ULA?

Would that give any benefits (in case the ULA would work ok with such modification)?

Pin 13 should already be at around 2.8V to 3.5V (actual voltage depends on the ULA version) due to resistor R34. There is no reason why R34 could be replaced with a suitable DC-DC converter/switching regulator. Internally this supply feeds lots of series pass voltage regulators which provide the cell voltage supply of 0.85V - 0.95V. I don’t know what the drop-out voltage is for these, but for most series pass voltage regulators, it’s typically 2V to 2.5V. So the input voltage at pin 13 would have to be at least 2.8V.

Personally, I haven’t fitted any heatsinks on any ULA chips in any of my Sinclair computers (ZX81, TS1000, ZX Spectrums). Hence the only ones that have heatsinks are the ULAs in the Amstrad made +2 (grey) machines.

I said this before, the ULA is designed to run hot. The ULA chips fitted in the ZX81, TS1000, ZX Spectrum 16K/48K/+ are working within their designed temperature range. The internal semiconductor junction temperature can be up to 125°C (even higher for some devices). So the middle of the plastic DIL package getting too hot to hold your finger on, is normal. It’s nothing to worry about.

The biggest killer of ULA chips is edge-connector abuse. Beyond this, the majority of faults with ZX Spectrum 16K/48K/+ machines are caused by one or more DRAM chips failing. Or the DC-DC converter based around TR4, TR5 and ‘the coil’ malfunctioning. Using an external PSU with the incorrect polarity, an AC type, or one with too high an output (for issue 1 to issue 4a boards that have not been modified) will also wreck various semiconductors.

So my recommendation is always leave the ULA alone. Remove the 7805 and fit a good quality fixed output DC-DC converter / switching regulator in it’s place (I use Recom types because I bought ten at what was a good price about five years ago). Replace the electrolytic capacitors that are part of the power supply (DC-DC converter based around TR4, TR5 and ‘the coil’) or all of them. And if needed carry out this modification.

There is further discussion in this topic and in the topics that are linked to from that topic.

Mark

[1024MAK]

Remember that air is a very bad heat conductor, so even if the air inside is not moving a lot, adding a heatsink will greatly help to move that heat from the chip into the air.

The epoxy resin that the plastic DIL chip package is made from is also not a very good conductor of heat. DIL chips primarily dissipate the heat from the silicon chip via the lead frame that forms the chips pins. So fitting a heatsink to such a chip does not actually aid in cooling the silicon core as much as people think. Especially in a ZX81, TS1000 case or a ZX Spectrum in a rubber key case where the level of airflow is very low.

The power consumption of an entire ZX Spectrum board (excluding the 7805 voltage regulator losses) is only about 3 to 3.5 Watts.

The 7805 typically dissipates about 3.5 to 4 Watts when supplied by a Sinclair UK1400 PSU. Total power consumption being about 6.6 to 7.7 Watts.

Mark

[rastersoft]

The epoxy resin that the plastic DIL chip package is made from is also not a very good conductor of heat. DIL chips primarily dissipate the heat from the silicon chip via the lead frame that forms the chips pins. So fitting a heatsink to such a chip does not actually aid in cooling the silicon core as much as people think.

Mmm... But then, why did Amstrad fit a heatsink in the Plus 2 ULA?

[1024MAK]

Mmm... But then, why did Amstrad fit a heatsink in the Plus 2 ULA?

I don’t know the answer to that. But I do know that on the two +2 (grey) machines I have, the heatsinks don’t fit very well. So they are not very effective.

If you are convinced that ZX Spectrum ULA chips are weak sensitive chips that die often, search this forum, search Sinclair ZX World and search World of Spectrum forums. Then try to count the number of ULA chips that have been reported as having failed due to overheating, or suspected overheating. Then compare to DRAM chip faults. But I know of no one who suggests fitting heatsinks to 4116 DRAM chips (some of which run just as hot as ULA chips).

If you really want a ULA chip to run cool, use a vLA82 or a Nebula (both for 16K/48K/+ machines, there is a different type for the 128K toast rack and the +2 grey).

Mark

[rastersoft]

If you are convinced that ZX Spectrum ULA chips are weak sensitive chips that die often, search this forum, search Sinclair ZX World and search World of Spectrum forums. Then try to count the number of ULA chips that have been reported as having failed due to overheating, or suspected overheating. Then compare to DRAM chip faults. But I know of no one who suggests fitting heatsinks to 4116 DRAM chips (some of which run just as hot as ULA chips).

That's a good point, indeed.