Maximal restoration of ZX81

[The Byte Attic]

Hi Lardo,
I didn't look up the specific retention time of the Flash chip used in the ROMel. In reality, however, retention depends on the particular circumstances of use, such as average temperature of operation (higher means shorter retention time), the number of times a memory position is re-written, etc. Typically, retention times -- without refreshes -- can vary between a couple of months and about 10 years, but never decades. The ZX81, of course, has no house-keeping firmware to do Flash memory refreshes in the ROM.
Cheers, TBA

[1024MAK]

The AT49F001AN datasheet does not specify a data retention figure, but modern EEPROM/flash ROM devices typically have data retention times of 25 years plus if used as a ROM.

1980s EPROMs often were said to have data retention times of 10 years, but some are now still working at 30 to 40 years...

Mask ROM chips should not suffer from loss of data, but are still subject to service life failures.

Mark

[The Byte Attic]

Oh my goodness, I inadvertently seem to have pushed some buttons. Relax, man, this is a hobby.
A chip's packaging is expressly done to help heat dissipation, and there is a lot of technology behind it. The package will reach thermal equilibrium with the die inside. Naturally, the die will always be a little hotter than the outside of the package, as it is the source of heat, but the difference won't be extraordinary, otherwise nobody would bother with heat-sinks (today there are even fluid-cooled heat-sinks and ideas for doing cooling in between dies in 3D advanced packages).
As for your positions on fluid dynamics, who knows, perhaps you know more than me! But I guess fluid dynamics isn't as simple as imagining that you have only vertical air flows inside a case... Just a thought.
Cheers, TBA.

Warm air convection currents form even in relatively narrow cases, as EDA software models show, and they can flow horizontally upon hitting a barrier. By your account, the heat sink used in the 7805 should be useless too... Try to run the regulator without it. And no, the measurement I did wasn't meant to reflect the details of the situation after the case is closed, it was obviously meant as a mere indication. But as such, it was certainly telling, as a difference of over 20 C doesn't seem circumstantial.

Just about every test I have seen re ULA heat-sinks always show the thing in an open or ambient air situation with a hand held jobby aimed at it, so how about a thermal device shoved UNDER the chip as per most modern motherboards, then put the zeddy back in its nice reasonably air tight case in other words a real world environment,give it something to do for four/five hours or soed with the 81 doing something meaningful over a few hours then let us know how you get on. then get back to us with an accurate DIE temperature as opposed to an exposed heat sink in open air after 15 minutes.

@Lardo.

That really is a thing of beauty, I do hope after you find a solution for your switch mounting ,that you hold on to this one.

Noticed the the Atari console strategically placed too, when's that up for sale?

Where to start?

Right over 45 years of designing and building cooling solutions for audio amplification both internal and external is where my knowledge and experience comes from ok and just shoving an heat-sink on output transistors without fan or airflow doesn't cut the mustard regardless of textbook theory.
Be it transistor or chip the principle is still the same.

Internal heat-sinking without air flow through never works out well in an enclosed environment which regardless of your opinion the ZX81 case is such an environment.
You pointing a device at an heat-sink after 15 minutes in ambient conditions reflects only the HEAT-SINKS temperature not the DIE temperature so is totally irrelevant as after so short a time I doubt whether it reached any kind of thermal equilibrium with the ULA so if you are going to make bold claims then go to the time and trouble of doing the test as has previously suggested, ie test the DIE temperature with and without the heat-sink in a fully enclosed case with the probe on the underside of the ULA with the 81 doing something meaningful over a few hours then let us know how you get on.

Finally contrary to popular belief the the biggest cause of ULA failure is abuse by the user trying to connect/disconnect interfaces whilst powered up or a fault within the interface blowing the ULA's internals,faulty Memotech RAM modules seem prone to this in my experience.

I have yet to meet ANYBODY who can show me with scientific measured PROOF that their own ULA's failed due to heat damage, the ULA doesn't run at the furnace like temperatures that some people seem to think they do.

Also as for the regulator running without a heat-sink, just take a look inside many of the ZX81 peripherals that have the same regulator inside their own cases running off the main power supply with no heat-sink fitted yet still work, the Memotech Z80 assembler module being one such example which shows the same 11.5 volts under load as the 81's own regulator yet manages perfectly well without an heat-sink so not " your account " as you call it but measured proof.


For the record at the suggestion of a forum member and for the sake of experimentation I have an internally fan cooled 81 drawing air from the rear to the underside and no part of the upper case gets warm never mind hot during use with rather proves the point re airflow I would have thought.


In the end everyone has their own opinion about these things so I think it's best that you go to your church and I'll go to mine as a wise man once said.

[Moggy]

Oh my goodness, I inadvertently seem to have pushed some buttons. Relax, man, this is a hobby.
A chip's packaging is expressly done to help heat dissipation, and there is a lot of technology behind it. The package will reach thermal equilibrium with the die inside. Naturally, the die will always be a little hotter than the outside of the package, as it is the source of heat, but the difference won't be extraordinary, otherwise nobody would bother with heat-sinks (today there are even fluid-cooled heat-sinks and ideas for doing cooling in between dies in 3D advanced packages).
As for your positions on fluid dynamics, who knows, perhaps you know more than me! But I guess fluid dynamics isn't as simple as imagining that you have only vertical air flows inside a case... Just a thought.
Cheers, TBA.

Warm air convection currents form even in relatively narrow cases, as EDA software models show, and they can flow horizontally upon hitting a barrier. By your account, the heat sink used in the 7805 should be useless too... Try to run the regulator without it. And no, the measurement I did wasn't meant to reflect the details of the situation after the case is closed, it was obviously meant as a mere indication. But as such, it was certainly telling, as a difference of over 20 C doesn't seem circumstantial.

Where to start?

Right over 45 years of designing and building cooling solutions for audio amplification both internal and external is where my knowledge and experience comes from ok and just shoving an heat-sink on output transistors without fan or airflow doesn't cut the mustard regardless of textbook theory.
Be it transistor or chip the principle is still the same.

Internal heat-sinking without air flow through never works out well in an enclosed environment which regardless of your opinion the ZX81 case is such an environment.
You pointing a device at an heat-sink after 15 minutes in ambient conditions reflects only the HEAT-SINKS temperature not the DIE temperature so is totally irrelevant aTBAs after so short a time I doubt whether it reached any kind of thermal equilibrium with the ULA so if you are going to makeTBA bold claims then go to the time and trouble of doing the test as has previously suggested, ie test the DIE temperature with and without the heat-sink in a fully enclosed case with the probe on the underside of the ULA with the 81 doing something meaningful over a few hours then let us know how you get on.

Finally contrary to popular belief the the biggest cause of ULA failure is abuse by the user trying to connect/disconnect interfaces whilst powered up or a fault within the interface blowing the ULA's internals,faulty Memotech RAM modules seem prone to this in my experience.

I have yet to meet ANYBODY who can show me with scientific measured PROOF that their own ULA's failed due to heat damage, the ULA doesn't run at the furnace like temperatures that some people seem to think they do.

Also as for the regulator running without a heat-sink, just take a look inside many of the ZX81 peripherals that have the same regulator inside their own cases running off the main power supply with no heat-sink fitted yet still work, the Memotech Z80 assembler module being one such example which shows the same 11.5 volts under load as the 81's own regulator yet manages perfectly well without an heat-sink so not " your account " as you call it but measured proof.TBA


For the record at the suggestion of a forum member and for the sake of experimentation I have an internally fan cooled 81 drawing air from the rear to the underside and no part of the upper case gets warm never mind hot during use with rather proves the point re airflow I would have thought.


In the end everyone has their own opinion about these things so I think it's best that you go to your church and I'll go to mine as a wise man once said.

Trust me you have pressed no buttons so no need for the cheap "relax" shot and yes I know it's a hobby but accurate observations go a long way on this forum as you may find when you have been here a little longer,and yes I'm fully aware of non vertical air flow hence my air cooled 81 drawing air horizontally through the case but nice to be patronised all the same

We will never see eye to eye so best we leave it at that.

[1024MAK]

Lots of quotes...

Lots of quotes...

There is no need to keep quoting large amounts of text. Use the “Post Reply” button instead or at least edit the quoted text to a reasonable amount that is relevant.

Otherwise I will have to step in and aggressively edit posts.

Mark

[Moggy]

Lots of quotes...

Lots of quotes...

There is no need to keep quoting large amounts of text. Use the “Post Reply” button instead or at least edit the quoted text to a reasonable amount that is relevant.

Otherwise I will have to step in and aggressively edit posts.

Mark

Sorry Mark, in my case it's lack of knowledge,I always cock it up being an old fart blah blah so feel free to hack away at my posts if needed.


You look lovely when your'e angry by the way XX

[1024MAK]

There is lots of useful information on semiconductors and heatsinking here, including the calculations needed to work out the temperatures or heat flow values.

However the important points are these:

• The ULA runs warm because that is due to the way it was designed, so it’s normal. The only ULA chips that I am aware of that need heatsinks are some versions of the VideoULA in Acorn BBC computers (here it is needed due to a design flaw). The heat output of the ULA is well within the specifications for a DIL plastic package, hence why there is no designated location on the plastic package for a heatsink (and no metal area for good heat-flow). The lack of a metal area means the heat from the chip die has to transverse through the epoxy plastic resin encapsulation.
• There is an option with the type of ULAs used to have two power input pins, one (+5V) for the internal/external interface circuitry and a lower voltage one (+1.3V to +5V) for the internal logic matrix circuitry (which is CML and operates at between 0.84V and 0.95V and is powered by on chip series regulators). Using two separate supply voltages reduces device power dissipation. However Sinclair wanted the maximum number of pins for all the logic functions.
• The 7805 voltage regulator is designed to generate a lot of heat, so has a relatively large metal tab that can be attached to a heatsink, therefore ensuring a good heat flow from the voltage regulator to the heatsink. In a standard unexpanded ZX81, the 7805 dissipates about 2.7 Watts of heat. Far more than the ULA. If the input voltage is lower and the output current is reduced, so it has to dissipate less power, then a 7805 can operate without a heatsink.
• Due to the very limited airflow through the case, as the heat from the components (primarily from the 7805) warm up the air inside the case, the heat transfer from the ULA epoxy plastic resin encapsulation to the air slows as the temperature difference falls. Exactly the same happens if a heatsink is fitted.

The best thing you can do to reduce the temperature of the ULA is to reduce the temperature of the air in the case, and the best way of doing this is to fit a modern DC/DC converter (such as a Recom, Traco Power or similar type) in place of the 7805.

2C000 ULA specifications:
ULA type/class: C - very high speed
Maximum clock speed: 20MHz
Gate delay: 8ns
Gate current: 210uA
Logic type: CML
Matrix size: 15 x 15 x 1
Matrix cells: 225
Peripheral cells: 40
Average no. of gates: 337
Matrix cell dissipation is 0.95mW per gate powered at +5V
Peripheral cell dissipation is 11mW powered at +5V

A 2C000 ULA powered at 5V with 300 active gates and 40 peripheral cells has a total power dissipation of 725mW.

An actual 2C184E ULA (fitted in a working ZX81) supply current is 133mA at +5V. Therefore the ULA power = 665mW.

This application note clearly shows that at an air temperature of 70°C the die temperature will be well below a conservative maximum die temperature of 125°C.

So in conclusion, the ULA does not need a heatsink.

Mark

[The Byte Attic]

Moggy, no intent to patronize you, far from me. I am too aware of how much I don't know. But you are the one who took the initiative to reproach my approach. And there is a thing or two I do know, particularly when it comes to the semiconductors/high-tech/computer industry, where I have been professionally for 24 years. In general, hot air has higher pressure than the surrounding atmospheric air. So hot air inside even a narrow case with few openings will always push its way out and create flow. Heat-sinks will almost always help, unless they are physically touching something above them, in which case they could indeed constrict air flow and make things slightly worse. But in the ZX81 the RF can is so high that you can be sure a heat-sink on the ULA will always help (unless it's a pretty high heat-sync; the one you see me using in the video you can be sure will help). The heat-sink I used more than doubles the total surface area of the ULA and significantly improves heat exchange. Cheers, TBA.

[The Byte Attic]

Mark, the point is not what the maximum operating temperature is. If you exceed that maximum the chip may fail on the spot. But the point here is the total life time of the chip, how long it will last, and that is influenced but the average temperature of operation, even if the latter falls within the limits defined in the datasheet. The cooler a chip runs, the longer it will last. You can count on it. Your point regarding the voltage regulator is a good one: indeed, replacing it with a buck converter will reduce the air temperature inside the case and improve the ability of the ULA to exchange heat with the surrounding air.

[1024MAK]

In my experience, the greatest causes of failures of electronic systems in general is the shock of powering on (or due to a rapid power off/on due to a power blip) or due to thermal cycling. And not due to the average temperature reached in a system when operating well within its normal operating specifications. Indeed, it has been known for PCs that are running continuously (24 hours a day, every day) to continue to operate for extended periods even when the case cooling fan or the PSU fan has failed.

The type of technology in the ULA is bipolar, and the CML matrix cells that make up the logic gates run from an internal regulated supply. It is this type of technology that makes it run warmer than more modern IC technology. Modern ICs are now made from CMOS technologies and hence unless run at very high clock speeds, mostly run cool.

Although in general, where possible, running semiconductors at lower temperatures is worthwhile in the hope of extending the the lifetime of said semiconductors, I agree with Moggy in that in a ZX81 (or indeed in a rubber key ZX Spectrum), the so called ‘benefit’ of fitting a heatsink to the ULA is marginal at best.

A lot of people get concerned when they touch a chip, transistor or diode and find that it (or part of it) is too hot to hold their finger on it. Just like there now exists this ‘thing’ where people feel the need to renew all the electrolytic capacitors in everything, without proper consideration or asking if it is needed.

In engineering, not everything is human finger friendly.

With home computers like the ZX81 and ZX Spectrum, a large number of failures are likely to be due to misuse of the expansion port/edge-connector (this in particular is known to kill ULA chips in ZX81 and ZX Spectrum computers). Some failures of home computers are due to faulty expansions. A large number of failures of various makes/models of home computers are due to a failure of one or more of the DRAM chips. There are of course other parts that also fail from time to time.

Mark