• Please review our updated Terms and Rules here

TTL Parts

E

Erik

Site Administrator
Staff member
Joined
Apr 27, 2003
Messages
3,588
Location
San Jose, CA
I have several machines in my collection that have been repaired by simply replacing TTL components on various boards.

One Altair 8800 was fixed when a friend helped me by tracing through the TTL on the front panel to isolate and replace a bad component or two.

Another Altair 8800 that I recently recieved had bad TTL on both the processor board and the front panel board. Fortunately both of those were socketed so I was able to isolate the bad parts fairly quickly.

The same was true of an Altos in my collection. One bad TTL part kept the machine from booting.

With that said, I have two questions:

1. What causes these parts to go bad? On the second Altair and on the Altos there was no reason for the parts to fail other then pure age. No spikes, static or other abuse could/should have caused it since there was none. The other Altair may have been poked the wrong way at some point.

2. Having replaced these busted components with newer ones, where can one find older date-coded parts without tearing up other old machines? I'd like to keep my collections as close to pristine as possible which, to me, means keeping the date codes for the parts within reason.

Any ideas or comments?

Erik
 
re: TTL Parts

Erik,

Older TTL (before mid-1980s) was *extremely* voltage sensitive. While you say there were no "spikes", a warm PSU that allowed the 5V line to drift over about 5.25v could easily have killed them.

If you really want to stick with original parts, you're going to have to do the same thing NASA do when they want ceramic-packaged 286 and 386 chips for the shuttle program - go on eBay and buy old equipment you can strip them out of! :)

Luckily, as TTL was both flaky and expensive in those days, pretty much all of it was socketed, making for easy stripping down of old boards. For bulk TTL, I'd suggest keeping an eye out for old telephone exchange boards circa 1979-1982.

If you're willing to compromise with newer devices, most machines will accept LS series chips (Low-power Schottky(sp?)) which have a greater voltage latitude, but being based on CMOS technology need to be handled with anti-static precautions. Also, they're no good for circuits where a single output line drives three or more inputs; in that case you need the original part.
 
"LS series chips (Low-power Schottky(sp?)) which have a greater voltage latitude, but being based on CMOS technology ........"

Sorry Charlie, LS TTL chips are not CMOS, nor based on CMOS technology.

CMOS = Complimentary Metal Oxide Semiconductor technology.

LS series is still BTL technology.
 
It is hard to know why some TTL chips go bad. Most of the very old ones that fail spontaneously are not due to electrical abuse or over-voltage and they are generally ones in epoxy packages not ceramic.

TTL's in general are fairly tolerant of brief over voltage events, if it is very brief and not above 7.5V provided the dissipation is not too high.

In general though, it doesn't often happen because of the protections added inside analog regulators like the LM309k, or the over-voltage & over-current detectors and crowbar systems used in computer PSU's.

Often you see modern generic switching psu's re-tasked to run logic boards and these special protections are not in the psu, this is asking for trouble in my opinion, doesn't seem to stop people from doing it though. Be cautious if you see switching units to replace analog regulators, look at the circuit and think of failure modes, can they fail in a way where they can over-voltage the output ? What protections for over-voltage, over-current and over-heating are built into the design? Compare that to the analog regulator.

When SMP's started to dominate computer power supplies, these protections were excellent, have a look at the supply in the 5155 computer and the lengths they went to, to make sure the supply would not damage the computer(IC's), have a look at the voltage window circuit on page 20 that would shut down the supply if the voltages were out of spec:

www.worldphaco.com/uploads/The_IBM_5155_POWER_SUPPLY.pdf

Of course a bad supply could potentially damage any logic chip, but its like the old saying, don't blame the player, blame the game, its not the logic chip's fault.

I have yet to find a single ceramic body mil spec 54 series TTL that is dead out of its box or had one fail in use, in boards now counting the many many hundreds of parts.

The epoxy case ones I think fail from time to time due to some process inside the package that cuts the connections from the pin to the die. Or a destructive process in the die. For example a gate input can go spontaneously open circuit. This is interpreted by the rest of the die as a logic high. It is just a fact that plastic is not as physically stable as ceramic or glass.

It was said by "someone" populating large boards at the game makers like Atari in the 1970's that one in 1/100 new 74 series TTL's, of the epoxy package, were defective. This has not been my experience, for NOS TTL's of the 74 series in epoxy cases, it is less at probably around 1/300 or less .

The 54 mil qualified parts (plain or LS) are phenomenally reliable and very hardy IC's.

A study by Tektronix on ESD resistance of parts indicated that TTL required orders of magnitude more electrostatic discharge energy to damage it than cmos.

In short, out of all the logic IC's ever made, I would rate the 54 mil spec IC's the winner by far for reliability (and I have been searching for a dud for 50 years now). They do fail of course, and I have read a report of this, it is just very very rare and they are very hardy devices.

The only way you can go a step more reliable is to get the RNS series radiation hardened mil spec versions that were Space rated, that were subject to even more test & inspection protocols than the 54 series, these are hard to find now, I really like them.

For the space rated parts, I always image the scenario of a Satellite coming home and the remains of it of it crash lands in the garden. You find a 54 series TTL deep in the guts of the burnt and mangled wreckage, remove it and it is still working to make a useful project !

The good news is, if you really want the 54 parts, they are often just as easy to get as the 74 parts, and they are a very good choice . Generally though, I'm happy with plain 74 and 54 and LS parts. These IC's you will find are also much more reliable than memory product IC's based on mos technology in vintage computers. But still, in a vintage computer it is not uncommon now to find the occasional dud epoxy cased 74 series IC in them.
 
Hugo Holden said:
It is hard to know why some TTL chips go bad. Most of the very old ones that fail spontaneously are not due to electrical abuse or over-voltage and they are generally ones in epoxy packages not ceramic.
Click to expand...
^^This. "Plastic" encapsulation technology was somewhat chancy in the earlier days. Ceramic was standard for military-industrial applications. Of course, that's not to say that ceramic didn't have its own growing pains, particularly when high-density RAM started to appear.
 
There are all sorts of reasons.

This is one of the reasons the JEDEC Standards were produced.

Their website has some interesting reports (although you may have to join to access them).

Dave
 
Peet42 said:
<snip>

Older TTL (before mid-1980s) was *extremely* voltage sensitive. While you say there were no "spikes", a warm PSU that allowed the 5V line to drift over about 5.25v could easily have killed them.

<snip>
Click to expand...

That's not true. Original 7400-series TTL is spec'd to run correctly (forever) with a 5.25V supply, and has a 7V absolute maximum rating.
 
One thing that helps confirm that some logic IC's fail on the shelf (specifically epoxy cased ones) is testing large numbers of NOS IC's, that have never been used. One could argue that they could have been dud from new, but, I don't think so because the rate of duds would be unacceptable.

For example I acquired a good number of 2114 memory IC's for testing. The reason is these IC's feature a lot in PET computer repairs.

On testing a large stock of NOS epoxy cased MM2114's from National with 1970's date codes I found that about 1 in 10 was faulty. They can't have been like this new. So some sort of degradation has gone on inside the package, without any power applied.

On the other hand testing over 60 ceramic body types, the D2114 from Intel and some other ceramic body equivalents made by AMD, of 1980's vintage, not a single dud to be found.

I think if you start to find duds in 1/10 of some brand and age of epoxy package chip, probably the writing is on the wall for failures in the similar units because the same processes will be going on in them. I have fitted my PET and AIM-65 with the Intel's ceramic 2114's, and I'm not expecting any more trouble because they come from a much more reliable cohort of parts.
 
I seem to remember that a lot failure in early ICs was moisture intrusion in plastic packages.
That was one of the reasons for ceramic packages.
 
jlang said:
I seem to remember that a lot failure in early ICs was moisture intrusion in plastic packages.
That was one of the reasons for ceramic packages.
Click to expand...
This is one reason I'm nervous about washing old circuit boards.
 
The TTL chips in all of our vintage tech. is 20 years older and, therefore, more failure modes will appear!

Dave
 
You must log in or register to reply here.
Back
Top