1970s era CMOS TI 7400 chips - REPLACE or LEAVE?

[jaunty]

hi all - i'm working on a mid 70s digital delay (audio piece) that has lots of old 7400 CMOS chips in it - I'm wondering if these are as failure prone and static sensitive as other CMOS chips of the period (like the 4000 series). I'm redoing all the soldering on many boards as it's all gone rather tatty looking and grey and krinkled and cracked. As a matter of course - if there were only a few i might try replacing but there are many ... I've had no luck at all finding anything on the topic via search engine  ... so here i am asking the experts who may have dealt with some of these (?)

the packages in question are specifically SN74LS1 and LS2 series.

[RandallMcRee]

Well they're not CMOS, then...LS TTL is entirely different technology.

[DaJMasta]

Probably not necessary, 74LS is not particularly susceptible to things and doesn't have a ton of big disadvantages over newer parts, reasonably low power consumption, decent speed, etc.  Just needs to be 5V.

Also when installed in a board, generally, chips are much less static sensitive than when they're out, so just leaving them on the board and handling with care should be enough for hobby kinds of use.

[wraper]

74LS series are more robust to ESD than almost any CMOS, old and modern because there are no FET gates to blow up.

[wraper]

Also if replacing with something else, CMOS ICs likely will have lower output drive capability, different input threshold voltages and propagation delay.

[james_s]

Why would you replace them if they're not bad?

[SeanB]

Only replace the failed ones. In my experience generally the only TTL that fails is that which had inputs and outputs leading off the board, they get killed by either being shorted to Vcc or Gnd for long periods, or from having voltages under 0V or over 7V applied to them.  The other failure is edge triggered logic, like 7474 and other counters/ latches, failing to trigger reliably, but that is pretty rare unless it too is either driving off board or being used as an input device. Generally they are pretty reliable provided you do not have a batch which had QC issues, and after a half century those pretty much will all have failed long ago anyway.

[Gyro]

A number of 74Cxx series ICs have been found to have failed in old Datron 1041/1051 meters. The 74xx, 74LSxx and 4000 series parts in the same meters have all been fine.

It doesn't sound as if you're actually dealing with CMOS parts, just standard bipolar TTL, so I'd leave them alone but if you find any 74C then treat them with suspicion and maybe replace with 74HC (assuming that doesn't cause timing issues).

[Synthtech]

I make my living servicing 1970’s electronics and it’s very rare that I ever see a failed 74 series ic, even the HC series. Less than one per year. The chips that I see failing are 4000 series CMOS, especially multiplexers/analog switches, 4051’s, 4053’s along with RAM and some op-amps, usually 4558’s and 1458’s.

I would not consider ever swapping out 74 series IC’s that are working.

[digsys]

I've had slightly different experiences with 7400 series. Mainly in 1st generation motherboards and early emulators etc. Had heaps of overheated / dead ones.
The 74HCT series are a direct replacement, but as pointed out, the only time they may be an issue is if there is a "race condition" happening.
In those cases, a small pF cap was enough to fix it :-)

[drussell]

Original 74 and 74LS are not CMOS.  They are also far more robust than CMOS logic and rarely fail.  I've salvaged hundreds and hundreds of 74/74LS series logic from 70s and 80s gear over the years and re-used them for a multitude of purposes.  They're still just fine, regardless of manufacturing date.  You really have to try pretty hard to kill those babies. 

I wouldn't change any of them unless you find any that are actually defective when you test out all operations of the device they're in.

[jaunty]

Why would you replace them if they're not bad?

in case they are very prone to go bad ... it's generally recommended with 4000 series logic chips for example

[jaunty]

NOT CMOS. got it ... i could have sworn I'd read that they were .. as the 4000 series are supposed to be - and I know they had problems with aging etc. anyway ... not that it really matters to the project at hand. thanks everyone - i was kind of hoping for encouragement not to have to swap them out

[oPossum]

NOT CMOS. got it ... i could have sworn I'd read that they were

Some of them are. The letters following the 74 determine the family.

https://en.wikipedia.org/wiki/7400-series_integrated_circuits#Families

All 4000 series are CMOS.

[james_s]

Only replace the failed ones. In my experience generally the only TTL that fails is that which had inputs and outputs leading off the board, they get killed by either being shorted to Vcc or Gnd for long periods, or from having voltages under 0V or over 7V applied to them.  The other failure is edge triggered logic, like 7474 and other counters/ latches, failing to trigger reliably, but that is pretty rare unless it too is either driving off board or being used as an input device. Generally they are pretty reliable provided you do not have a batch which had QC issues, and after a half century those pretty much will all have failed long ago anyway.

Earlier this year I repaired an Asteroids Deluxe board for somebody that had I think 5 bad 74153 TTL chips, and there was another that had obviously been replaced previously. I was surprised to see this happen but it must have been a bad batch. The failed ICs were all the same brand, the really odd thing is they all failed at the same time.

[SeanB]

Not surprising, they all were probably from the same slice of silicon, and it probably was contaminated somewhere along the line in processing, or they had improper alignment of the masks and diffusions. Worked on test, but long term they all had enough stress to fail from some reason, probably either the metal migrate on the top of the chip in thin areas, or the internal gates punched through.

[james_s]

That's the only thing I can think of, but I still find it pretty surprising that they all lasted almost 40 years and then failed at the same time. I suppose it's possible that they were on the edge of failing and a momentary overvoltage pushed them over. A bad connection on the sense lines to the 5V regulator can cause that.

[drussell]

That's the only thing I can think of, but I still find it pretty surprising that they all lasted almost 40 years and then failed at the same time. I suppose it's possible that they were on the edge of failing and a momentary overvoltage pushed them over. A bad connection on the sense lines to the 5V regulator can cause that.

Yeah, I would suspect that they were murdered my some kind of power spike which finally pushed them over the edge.

[jaunty]

NOT CMOS. got it ... i could have sworn I'd read that they were

Some of them are. The letters following the 74 determine the family.

https://en.wikipedia.org/wiki/7400-series_integrated_circuits#Families

All 4000 series are CMOS.

yes  - that's where i got the idea they were CMOS

[drussell]

yes  - that's where i got the idea they were CMOS

The original, standard 74 series (7400, 7401, etc.) as well as the 74F, H, L, S, AS, LS, and ALS variants are bipolar TTL.  The 54xx are the same gates, however, they are the military temperature range and 64xx are industrial temperature range variants.

LS was probably the most commonly used (low-power, Schottky) in most places in most circuits, at least through into the early 1990s for most circuits, though often still with a smattering of plain 74 due to the increased current output capability, thus greater fanout.  74F (fast) was commonly used once speeds (especially glue logic on later computer motherboards, etc.) went up above the 16 MHz or so that most of the parts in the earlier series could handle.

I remember doing lots of circuit designs as a kid where an LS couldn't quite drive all the gates you needed, so you'd grab a regular 74 version from the drawer of salvaged chips and swap it in place of the LS.  LS was nice because when you had dozens and dozens of chips in your design, you could still run from a 7805 regulator.    Rows and rows of regular 74 made for a power hungry beast!

[glarsson]

Odd fact: The 74LS family is not TTL, it is DTL.

[Gyro]

   No, it's Low power Schottky TTL. The Schottky diode junctions prevent the transistors from saturating, thus speeding up the switching - or more precisely, achieving the same switching speeds as standard TTL at much lower supply current.

[EDIT: To make full use of the Schottky speedup you'd use 74S which achieves higher speeds while taking about the same power as standard TTL]

[wraper]

Odd fact: The 74LS family is not TTL, it is DTL.

No it's not. It's TTL with Schottky diode clamps. DTL is ancient technology prior to TTL.

[drussell]

Odd fact: The 74LS family is not TTL, it is DTL.

Ummm, no.  Take a look at the internal schematic diagrams of an LS gate.

DTL was essentially the direct predecessor to TTL, where the actual logic function in the gate was "computed" by diodes then the output buffered by a transistor, (hence the name Diode - Transistor Logic.)

[MK14]

Odd fact: The 74LS family is not TTL, it is DTL.

The INPUTS can be considered, as being like DTL (But I'd still say they are TTL, really). But NOT the (rest of the circuitry or the) totem-pole outputs!

Please see page 10
http://ecee.colorado.edu/~mcclurel/ON_Semiconductor_LSTTL_Data_DL121-D.pdf

Most LS elements use a DTL type input circuit with
Schottky diodes to perform the AND function