UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > Specific Vintage Equipment > Vintage Computers

Notices

Vintage Computers Any vintage computer systems, calculators, video games etc., but with an emphasis on 1980s and earlier equipment.

Closed Thread
 
Thread Tools
Old 17th May 2020, 11:51 pm   #61
electroanorak
Hexode
 
electroanorak's Avatar
 
Join Date: Apr 2007
Location: Tonyrefail, Rhondda, South Wales, UK.
Posts: 337
Default Re: How long can a computer last?

I have a comaq Armada E500. It was given to me in 1996 and served me until 2009. I do still use it at times. The batteries are still good for seven hours. It runs XP, SP3 and boots in seconds.

Daniel.
electroanorak is offline  
Old 18th May 2020, 10:18 am   #62
GMB
Dekatron
 
GMB's Avatar
 
Join Date: Aug 2003
Location: near Reading (and sometimes Torquay)
Posts: 3,086
Default Re: How long can a computer last?

I would not reflash memories, but just make sure you have a copy of the data and that you can make a new one when needed (test this by doing it). EPROM memory can be accessed by examining the performance at different voltages (the basis of the verify-check which is not just reading them in the normal way). By doing the verify in an analogue fashion you can get an idea of how it is doing. (I used to design EPROM programmers).

SSD life depends a lot of usage. The more it is written the deader it becomes because the voltage levels of the 0s and 1s are not cleanly reset by writing. What annoys me greatly is that modern bloated operating systems are endlessly making unnecessary writes to their main disc thus wearing them out.

You must take the quoted life expectancies with a cup-full of salt. They are for ideal situations. It assumes the programming was perfect and that you do not get unlucky with cosmic rays. The snag with Flash and EPROM memory is that it can be dead at any instant. In fact it may already be corrupt - it is surprising how many bits you can often change before anything goes wrong unless there is a start-up check done. The worst kind of these is where it refuses to run if corrupt. I have a Tektronix scope that has corrupt firmware, but still works - so long as I remember the secret button combination to press to make it run in spite of the detected damage.

It is very clear to me that most computers are actually making a lot more errors than people are aware of. The software of modern systems is very poor...
Decades ago I was involved with a mainframe operating system. Discs were a constant source of errors - as reported by the disc's own error detecting. We wondered what they did not report so modified our filing system to put a header on every sector on the disc with detail about where it was, how it linked to others and what file it was part of. When this system went live there was a shocking increase in disc errors reported by our software! Writing valid sectors to the wrong place was a popular sport of moving head discs, for example.

We also seem to tolerate mystery errors. When I was a junior systems programmer, every morning I would get a stack of paper listing the entire contents of the memory in hex when the mainframe system had crashed. I would examine this to determine which bit caused the error. Double bit memory errors (undetected by parity) turned out to be more common than the maths would suggest! Every crash was seen as a dreadful failure of the service, that required complete explanation. How times have changed!
GMB is offline  
Old 18th May 2020, 3:18 pm   #63
audion_1908
Hexode
 
Join Date: Nov 2009
Location: Chertsey, Surrey, UK.
Posts: 456
Default Re: How long can a computer last?

Quote:
Originally Posted by SiriusHardware View Post
Quote:
why are modern EPROM programmers that support the old proms so expensive?
I think because only a tiny minority of people now need to program things like Bipolar PROMs, so to get that functionality you have to buy the top of the range programmer which can programme absolutely everything else as well. Creating the programming profiles for each and every single device that the programmer can handle takes time, and the man-hours involved are reflected in the price.

My high-end device programmer also dates from the mid-nineties and I bought that one specifically because it programmed Bipolar PROMs, which, even then, only the high end models still did. The support software is DOS, so I keep not just one, but two DOS computers working to ensure that I will be able to keep using it for the forseeable future.

I've noticed that some of the newest, cheap 'universal' EPROM programmers can not generate a VPP (programming voltage) of 25V. The designers are assuming that virtually no-one wants to be able to programme first and second generation EPROMs any more.
Yes I was wrong the 2 highest programming voltages where 21v and 25v not 24v
audion_1908 is offline  
Old 18th May 2020, 3:39 pm   #64
audion_1908
Hexode
 
Join Date: Nov 2009
Location: Chertsey, Surrey, UK.
Posts: 456
Default Re: How long can a computer last?

Quote:
Originally Posted by electroanorak View Post
I have a comaq Armada E500. It was given to me in 1996 and served me until 2009. I do still use it at times. The batteries are still good for seven hours. It runs XP, SP3 and boots in seconds.

Daniel.
The e500 looks like a cool computer, it uses a fast p3 and can take 512mb of ram total and having a 100mhz FSB can take most of the more common pc133 ram and at least one person has upgraded their e500 to 1 GB of RAM using Mac RAM! Apparently the CPU is socketed but won't boot with any other CPU that's a shame as you can get up to 1.4 gigahertz socket compatible CPUs

How much RAM did you upgraded to and what do you use it for now? It's good to see an old computers still in use.
How are the old batteries still so good, did it do 7h when new? My 7-year old laptop struggles to do half an hour flat out, I run Rosetta at home 24/7 and use to also run seti at home
audion_1908 is offline  
Old 18th May 2020, 4:01 pm   #65
audion_1908
Hexode
 
Join Date: Nov 2009
Location: Chertsey, Surrey, UK.
Posts: 456
Default Re: How long can a computer last?

Quote:
Originally Posted by GMB View Post
I would not reflash memories, but just make sure you have a copy of the data and that you can make a new one when needed (test this by doing it). EPROM memory can be accessed by examining the performance at different voltages (the basis of the verify-check which is not just reading them in the normal way). By doing the verify in an analogue fashion you can get an idea of how it is doing. (I used to design EPROM programmers).

SSD life depends a lot of usage. The more it is written the deader it becomes because the voltage levels of the 0s and 1s are not cleanly reset by writing. What annoys me greatly is that modern bloated operating systems are endlessly making unnecessary writes to their main disc thus wearing them out.

You must take the quoted life expectancies with a cup-full of salt. They are for ideal situations. It assumes the programming was perfect and that you do not get unlucky with cosmic rays. The snag with Flash and EPROM memory is that it can be dead at any instant. In fact it may already be corrupt - it is surprising how many bits you can often change before anything goes wrong unless there is a start-up check done. The worst kind of these is where it refuses to run if corrupt. I have a Tektronix scope that has corrupt firmware, but still works - so long as I remember the secret button combination to press to make it run in spite of the detected damage.

It is very clear to me that most computers are actually making a lot more errors than people are aware of. The software of modern systems is very poor...
Decades ago I was involved with a mainframe operating system. Discs were a constant source of errors - as reported by the disc's own error detecting. We wondered what they did not report so modified our filing system to put a header on every sector on the disc with detail about where it was, how it linked to others and what file it was part of. When this system went live there was a shocking increase in disc errors reported by our software! Writing valid sectors to the wrong place was a popular sport of moving head discs, for example.

We also seem to tolerate mystery errors. When I was a junior systems programmer, every morning I would get a stack of paper listing the entire contents of the memory in hex when the mainframe system had crashed. I would examine this to determine which bit caused the error. Double bit memory errors (undetected by parity) turned out to be more common than the maths would suggest! Every crash was seen as a dreadful failure of the service, that required complete explanation. How times have changed!
I did not know EPROM programmer would test at reduced voltage, that makes sense to shake out the flaky chips,

I have read that computers can operate with slightly corrupt RAM or eproms, but a test to show even one bits has gone flaky tells you the rest probably on it's way out and replace or fix it would be useful, I would like to know more about how my eeprom programmer works. Do eproms get slower after multiple rights and age?

I doubt a pc mb can to an advanced firmware read test, and taking the EPROM out has a certain amount of risk, so probably better to reflash on mb, but I like the idea of an advanced test every so often, or when you get the computer

When you say not to reflash, do you mean ssds or flash BIOS? an SSD is not going to be significantly worn by a complete rewrite every 10 or 15 years or so, and do computer motherboard flash BIOS automatically rewrite their code, I doubt it, I would think that would have the write endurance to take a complete flash every decade or so, as I would not expect it to have wear leveling firmware and some people are forever changing their BIOS settings.

There's few ways around the operating system chipping your SSD is to death, use an OS that doesn't continuously write to the ssd, ie not windows! Leave the computer off most of the time, have two ssds, one for OS one for data, that would simplify backup and data restoration.
I wonder how will Linux treats an SSD?

Why did the mainframe hard drive write sectors in the wrong place? What is a heat expansion alignment issue? What did the hard drive have a badly designed synchronisation system, it sounds like the hard disk made a real mess of your data!

Last edited by audion_1908; 18th May 2020 at 4:08 pm. Reason: Mixed
audion_1908 is offline  
Old 18th May 2020, 6:05 pm   #66
GMB
Dekatron
 
GMB's Avatar
 
Join Date: Aug 2003
Location: near Reading (and sometimes Torquay)
Posts: 3,086
Default Re: How long can a computer last?

Quote:
I did not know EPROM programmer would test at reduced voltage
Actually it is done at raised voltage, which means you can't do this in circuit. This does mean that a corrupt EPROM can sometimes be read correctly by varying the voltages.
EPROMs do not get slower as far as I am aware. Their expected failure mode is for zeros to become ones as they leak. But I have also seen catastrophic failures, e.g. on the firmware of an HP logic analyser - the data was wrecked so I guess something broke in the address decoding logic.

Now take care about systems programming their own flash memory. This is quite unlike how EPROM programmers would do it. Also a great warning: inbuilt Flash memory on microprocessors sometimes has a very short life for rewrites before dying. Like 100 times or less, not the huge number that general Flash memory usually can do. I suspect that things like BIOS Flash could well come into this category. Some older PIC processors had both kinds of Flash.

Long term with SSDs is an interesting problem. They have a life of their own so writing new data may not just overwrite the old but could go somewhere else in the store. This means that writing a complete copy could end up pessimising it - it might be better to leave it alone.

Operating systems write all kinds of logs and general crap to disc all the time, and Windows has its stupid registry which is a disaster waiting to get you. When your system is SSD aware it just means it doesn't bother optimising the sector layout or trying to defragment them - as that is worse than a waste of time!

Discs write sectors to the wrong place due to mechanical error, i.e. they seek to find the right track but something causes the head to go out of position just as it starts to write. I bet this problem is getting worse as modern discs have to go to extraordinary lengths to keep their heads on track as the data is so dense but I do not see the Windows filestore going to the lengths we used to do. Mind you, when I started, disc heads worked hydraulically - we had pools of oil on the floor on bad days!
GMB is offline  
Old 18th May 2020, 6:51 pm   #67
audion_1908
Hexode
 
Join Date: Nov 2009
Location: Chertsey, Surrey, UK.
Posts: 456
Default Re: How long can a computer last?

Quote:
Originally Posted by GMB View Post
Quote:
I did not know EPROM programmer would test at reduced voltage
Actually it is done at raised voltage, which means you can't do this in circuit. This does mean that a corrupt EPROM can sometimes be read correctly by varying the voltages.
EPROMs do not get slower as far as I am aware. Their expected failure mode is for zeros to become ones as they leak. But I have also seen catastrophic failures, e.g. on the firmware of an HP logic analyser - the data was wrecked so I guess something broke in the address decoding logic.

Now take care about systems programming their own flash memory. This is quite unlike how EPROM programmers would do it. Also a great warning: inbuilt Flash memory on microprocessors sometimes has a very short life for rewrites before dying. Like 100 times or less, not the huge number that general Flash memory usually can do. I suspect that things like BIOS Flash could well come into this category. Some older PIC processors had both kinds of Flash.

Long term with SSDs is an interesting problem. They have a life of their own so writing new data may not just overwrite the old but could go somewhere else in the store. This means that writing a complete copy could end up pessimising it - it might be better to leave it alone.

Operating systems write all kinds of logs and general crap to disc all the time, and Windows has its stupid registry which is a disaster waiting to get you. When your system is SSD aware it just means it doesn't bother optimising the sector layout or trying to defragment them - as that is worse than a waste of time!

Discs write sectors to the wrong place due to mechanical error, i.e. they seek to find the right track but something causes the head to go out of position just as it starts to write. I bet this problem is getting worse as modern discs have to go to extraordinary lengths to keep their heads on track as the data is so dense but I do not see the Windows filestore going to the lengths we used to do. Mind you, when I started, disc heads worked hydraulically - we had pools of oil on the floor on bad days!
I know EPROM programmer uses a raised voltage to burn the data into the chip,and the older design of eprom the higher the voltage but had presumed normal TTL logical level of 5v was used to read verify.

I think of the flash in embedded microcontroller or motherboard firmware only has a write endurance of 100, I still don't see the problem of rewriting it every 20 or 30 years or so? if the data retention and is about 40 years.

In normal use a computer some data is stored to the hard drive/ SSD is barely or never accessed, ssds have very clever on board controllers, do they automatically rewrite all data, could you force them to rewrite old data just by doing a full disk read instead of trying rewrite the data, as you're right, wear leveling firmware will put the new data a different place to the old.

pessimising?

Yes old mechanical hard drives always needed regular defragmenting and you could often revive a painfully slow computer that way, I did defragmented lots of computers the 90s! But most users didn't know about dfrag, that's why it was integrated into later versions of windows to run automatically in the background, also if you have over a few gigabyte hard disk, a full defragment takes a long time, longer than overnight!

Modern hard disks are amazing with the density that date has packed and I'm surprised they don't fail catastrophically more often.

hydraulic actuator! I have never heard of that before! no wonder you had track alignment problems, you have fluid dynamic problems, water hammer and the like, I think hydraulic fluid is also corrosive so when you had a bad day you had a really bad day! As I bet it eat circuit boards chips and cables and skin alike!

I suppose you could fix the problem with the mainframe hard disks in the the hard drive controller level, as long as it has some form of processor or you added some extra custom hardwired logic, a read before write system to verify alignment could work, I presume the problem was the heads being on the wrong track not in the wrong sector, as if it is on the wrong sector, I would think the motor is not synchronized well enough, a read before write system would give you a bad access speed penalty of atleast one hard disk revolution, unless you had some sort of data caching system, but with a well defragmented file system might not be that bad, as read before right would favour sequential file access.

Also keeping the hydraulic fluid at the same temperature would help a lot, I bet you got a lot of data errors when the thing was first turned on, or was the computer never turned off?

Last edited by audion_1908; 18th May 2020 at 7:05 pm.
audion_1908 is offline  
Old 20th May 2020, 1:53 pm   #68
audion_1908
Hexode
 
Join Date: Nov 2009
Location: Chertsey, Surrey, UK.
Posts: 456
Default Re: How long can a computer last?

What about the PCB? What is the expected lifetime of early modern 1990 or later PCBs? And what kills them slight damp temperature fluctuations, high temp storage?
audion_1908 is offline  
Old 20th May 2020, 3:17 pm   #69
Guest
Guest
 
Posts: n/a
Default Re: How long can a computer last?

Quote:
What about the PCB?
I have PCBs made in the late 50's that still work and they are SRPB (paper and resin) types.
 
Old 20th May 2020, 3:42 pm   #70
audion_1908
Hexode
 
Join Date: Nov 2009
Location: Chertsey, Surrey, UK.
Posts: 456
Default Re: How long can a computer last?

Quote:
Originally Posted by merlinmaxwell View Post
Quote:
What about the PCB?
I have PCBs made in the late 50's that still work and they are SRPB (paper and resin) types.
Is paper and resin PCB susceptible to overheating when improperly soldered? Or if a cheap grade?
audion_1908 is offline  
Old 20th May 2020, 5:05 pm   #71
Malcolm G6ANZ
Octode
 
Malcolm G6ANZ's Avatar
 
Join Date: Feb 2006
Location: Bristol, UK.
Posts: 1,030
Default Re: How long can a computer last?

Years ago I was looking after a Data General system. It had a 50MB (I think) hard disc the size of a dishwasher. Anyway at some point the heads crashed. We still needed a disc drive and they were no longer available except for really silly money. So myself and a colleague set about repairing it. We stripped it down and removed the platters and the damaged head. The platters were OK with no marks on them so they were cleaned with IPA and allowed to dry. There was only one damaged head so we took one from another scrap drive we had and replaced it. The drive was then reassembled and it was initialised and formatted. Amazingly it still worked and continued to work for several years until the system was replaced
Malcolm G6ANZ is offline  
Old 20th May 2020, 6:06 pm   #72
G6Tanuki
Dekatron
 
G6Tanuki's Avatar
 
Join Date: Apr 2012
Location: Wiltshire, UK.
Posts: 13,953
Default Re: How long can a computer last?

Quote:
Originally Posted by audion_1908 View Post
Quote:
Originally Posted by G6Tanuki View Post
Problem solved!
But sadly the modems dumped.
That's how a lot is fixed now, dump and replace, as it's cheaper but more wasteful, and not very green
Back then, just as now, I always had a policy of not running any gear that did not have current full manufacturer-support.

TBH I've always been happy to see the back of 'kit that kept me awake at night wondering just when it would inevitably fail and some guy in Wyoming would be phoning me at 03:00z saying his network was down".

I've never let greenie-stuff be an impediment to my contracted uptime/availability statistics.
G6Tanuki is offline  
Old 22nd May 2020, 9:59 am   #73
phildil
Pentode
 
Join Date: Sep 2014
Location: London, UK.
Posts: 113
Default Re: How long can a computer last?

Quote:
Originally Posted by GMB View Post
Writing valid sectors to the wrong place was a popular sport of moving head discs, for example.

We also seem to tolerate mystery errors. When I was a junior systems programmer, every morning I would get a stack of paper listing the entire contents of the memory in hex when the mainframe system had crashed. I would examine this to determine which bit caused the error. Double bit memory errors (undetected by parity) turned out to be more common than the maths would suggest! Every crash was seen as a dreadful failure of the service, that required complete explanation. How times have changed!
I can confirm about the disk drive sectors being written incorrectly. In the mid-80s I worked at a large old-school (big iron in climate-controlled computer room) 24/7 installation with a lot of rather old-fashioned 1970s kit, what we'd now call "legacy systems". One of our regular ops jobs on these machines was to patch certain disks at byte level to replace data after an erroneous write. These would usually be membership records, and the first we'd know about it was when a customer would ring up saying "I can't log in". Sometimes on night shifts we'd have a dozen or so of these patching repairs to sort out.

Also there was an early relational database system running on a separate system which would regularly fall over and lose the database structure. Rebuilding the database from backups took a couple of days: it would grind away and every so often we'd go down to the computer room "to see how the Admin system is getting on".

happy days!
Phil
phildil is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 9:09 am.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.