Retro Chip Tester (RCT)

I just recently bought a Commodore 4040.  Hopefully I don't need the RCT but if I do, thanks for the info.

I plan to try to hook it up to an Apple ][+ with an IEEE-488 card...

... insider. I myself was interested in this gadget because I happen to sell Apple-1 kits so a lot of vintage ICs go through my hands and need to be tested. Alas, a significant fraction of the "bad" ICs I encounter are not plain duds, but they have parametric faults due to age (I know all the prevalent fault mechanisms for all mainstream process technologies, so this is no surprise to me). You may be able to fish out many "dead" ICs with the capabilities of this tester, but there will be enough "limping" ones left which it can't detect. You literally have to check them in the final application, let them run for a while ("burn-in") and maybe challenge them a bit with a heat gun and cold spray to see if they are weaklings. Since you have to do this work anyways, testing the IC in the application, why would you buy / build such a "tester" ?

I can see it may have some merits to quickly probe if a suspect IC really is a basket case, if it says the IC is bad, it is so (if the test program has no bug). So sure, you can weed out some bad ICs. But the problem is, if this "tester" tells you the IC is good, this may be a lie, and then you put it back into the application and look for other candidates to be the culprit. Which will be a waste of time if the "tested good" part was the real culprit.

What I recommend for anyone repairing vintage electronics with socketed ICs is the following, super cheap method for dummies (like me ;-) . . . get / borrow / steal a "good" machine of the same kind.  Get some large piece of ESD foam ( if you can't, use a sheet of styrofoam wrapped in aluminum foil). Depending on your skill level understanding electronics, start with the most suspect ICs first (depending on the symptoms). But even if you know absolutely nothing, and have no instruments, just pull out a small group (6 or so ?) ICs from the bad machine and replace them with the ICs from the same PCB location of the good machine. If the "bad" machine works afterwards, you know the bad IC is in that group and one-by-one back substitution can pinpoint the one who is the culprit.  During the whole process, arrange the pulled ICs from the "bad"  machine neatly on the ESD foam such that you know from which IC socket they came from. I typically put a sheet of paper on the ESD foam and write down the IC # or location on the PCB (if there is a location grid) on the place where I put the "bad" IC down. These pins pierce the paper easily. They also will pierce your fingers and draw blood as easily. See. This is how evil they are !

With this method, even a kid can repair vintage electronics even without schematics nor instruments. Of course, there are embellishments. If you have instruments, such as that super expensive $7 multimeter, check the regulated supply voltages to be in spec first. Use the AC voltage setting to see if the clock runs. If you have an oscilloscope, the better. If you have a schematic and can read it, the more power to you !

I do not want to indulge into the more esoteric aspects of repairing vintage computers such as the "magic touch" just using the fingers of your hands. I learned this technique in the 1980s from a guy who worked as a service technician for DEC. It's unbelievable but it finds a lot of faults within a minute and you don't even need to set up your instruments. It's "magick" ! (No I won't tell you this secret).

The time-honored way of testing ICs for parametric faults is using signature and curve analyzers with known-working reference ICs.

Examples are the HP 10529A and the Huntron Tracker. But these are really time-saving convenience instruments, you can do the same thing with a signal generator and oscilloscope; it just takes longer to set up.

... not to be confused with the HP10529A !

Do NOT buy a "Signature Analyzer". Unless you have the service manual of the machine to be repaired calling it out. You may need a diagnostics floppy disk or EPROM, too.

Signature Analysis was a 1970s fad. It was somewhat useful, though. The idea was, when you run a test program provided by the manufacturer of the machine to be repaired, the "Signature Analyzer" could catch the known signal pattern the test program produced, and any deviation from the "Signature" hex code at any given probe location would mean the fault sits upstream of this location. This method was very efficient to find the offending ICs (or modules).

But here is the problem: unless you have the  test program and the service manual, the Signature Analyzer is useless.

About the "IC compare gadgets":

I myself got burned with the whole IC:IC comparision idea by a product I designed in the 1980s called "PLDdiag". It compared two PLDs. But it did not work when the PLD contained certain nefarious things like ring oscillators or asynchronous state machines. Still, for any "honest" PLD it did a good job. But we sold only a few and bought them all back (except for the customers who refused to give them back, same story as with the Apple-1).

To find a "final solution" (bad vibes, yeah, but listen to the rest of the story) to the PLD testing problem I then wrote a CAD suite with an ATVG. I made my 2nd million with that software. It was great and could do things which even today, 30 years later, the known "science" can't do.  Here is the irony. One customer who bought that CAD suite from our company testified that in the first year this ATVG and my "ambigous delay" timing simulator saved them more than 2 million of rework costs. Using my CAD software, they weeded out hundreds of dubious PLD designs and generated test vectors for all their PLDs which brought the rework down to virtually nil.

Lesson learned. Always demand a cut on the saved money / increased profit your software tools bring to the customer. Do not just take the crumbs they pay for the software. Alas, this would only work in a honest world not run by psychopaths and "demon possessed" people. (not that I'm overy religious - I despise organized religions - but that "demon" metaphor is the most lazy and energy efficient explanation I have why these monstrous individuals even exist - one level worse, above psychopathology - they need a bullet to the head to let the "demon" out and stop them to turn our lives to hell. In a better world the government (Of The People And By The People) would lock those monsters up into some sort of "Arkham Asylum" but - just as a thought experiment - imagine what happens if these monsters pull the strings of "their" government puppets --- I leave it to the reader to decide if this is the world we live in now. No further comments wanted. I just had the urge to give you the "spark").

Just got me one of these from Germany!

Can wait to build and test it!

In post #8, Denis1973 wrote:

"My plan is now to replace some of my newer Apple-1 ICs by tested good ones from the 70s one by one."

Uncle Bernie comments:

You can do that even if you have no "Retro Chip Tester". Just replace one TTL in the Apple-1 after another (power turned off, of course) and then let it run for a while.

If all works fine even after warm up, turn power off, and replace another TTL. Keep those known good ones on some strips (ESD foam not needed for TTL but mandatory for MOS) organized in the same order as they came off the Apple-1. So if anything fails, such as "infant mortality" during burn-in, you always can replace the known good ICs back to the exact place they came from. By using this method, you can make your own "100% tested and burned-in" IC sets for the Apple-1, which I highly recommend for anyone who has a known good and functional build. I see my kits as a process to "settle seeds" worldwide but when I stop selling them, the Apple-1 community must be ready to make their own "known good" IC sets using my proven procedure. You can speed the process up by replacing groups of 6 ICs at a time, or whole banks for 8 DRAMs. In all my burn-in rigs I have the diagnostics page enabled in the A1, A2 PROMs so finding bad DRAMs and other types of misbehavior  is quick.

It has been my experience that in about every 3rd such "harvest" process I end up with an Apple-1 that had at least one faulty IC. So if the failure rate in the wild is the same (which is likely as the "quality" of the 1970s TTL was like that), 1 out of 3 Apple-1 builds based on untested ICs would fail. Unless that builder has access to a known good IC set, chances are slim to find the bad one(s) in a useful amount of time.

Not all ICs who fail in the Apple-1 are duds. I have found certain tubes of ICs where the whole tube would not work in the Apple-1, but when plugged into some other machine, like an Apple-II, they would work just fine. Of course this is a bush league technique but unless the "tester" also tests the logic levels with accurate signal sources and comparators under the specified loads and tests the timing to be correct (like real IC testers costing millions of US$ do) you are not much better off with that "tester" than being bush league.

The reason why so many TTL ICs from the 1960s and 1970s are bad is lack of testing by the manufacturers. While every IC got a crude go/nogo test for the basic functions, parametric tests were only done on a sample basis, because the testers who could do that were either heinously expensive or too few were available to run each and every IC produced through the full parametric test suite. So even in a lot declared "good" by evil statistics (AQL = Acceptable Quality Level), some could fail parametric performance in the application.It's all a question of money. The "military grade" ICs were all fully tested to exacting standards but would cost much, much more than the "commercial grade" ones. Quality is just a matter of costs, you know. Nobody can make a quality product at the cheapest price.

The consequences for the electronics industry were profound.  With such insufficiently tested ICs, you can't build complex boards with more than a few dozen ICs on them. Too many boards would fail and need rework. So OEMs (especially computer manufacturers) set up their own in-house incoming inspection using fully parametric testers. These are making the mysterious dots or stripes which you can sometimes find on surplus ICs. Tested good ones got such a mark. So these are not signs of low quality, as some suspect, but actually a sign of passing the incoming inspection of some large OEM. But of course, after 30-50 years living in a tube somewhere in a warehouse, all bets are off.

The more dire consequences of this testing and AQL level skullduggery are twofold. In many cases, the rejected lots of the semiconductor manufacturers (which did not reach the required AQL level) ended up on the "grey market" and they are still lurking in the warehouses of the IC brokers worldwide, waiting for unsuspecting buyers. And unscrupolous OEMs also put the rejects from their incoming inspection back into the original manufacturer tubes and sold them to surplus peddlers aka IC brokers.

I suspect that most of the super lousy quality ICs I encountered for my Apple-1 project  came from such rejected lots. Otherwise I could not explain why certain 1977 date code Signetics 2519N have up to 25% failure rate, some 1975, 1976 data code Signetics 2504V have 30-50% failure rate, and some National Semiconductors MM1404 had 50-70% failure rate. No such problem with any AMD IC I saw - Jerry Sanders had a quality obsession and so AMD tested their products much more thoroughly than the rest of the pack. This is why I use AM1404 or AM2804 in my kits. But Signetics also could produce quality products - I just bought a lot of excellent 2519B with date code 7627 and there were only 1.4% duds in the whole lot. The first ones are still in burn-in, though.

TTL failure rates I've seen are much lower, but still nasty. If you have 3% bad TTLs, Pgood = 0.97 per IC, but if you have 34 of those in the Apple-1, the probability that all are good is only 0.355 (0.97^34), so 35.5% of the builds have all good TTL, and 64.5% of the builds would fail. This is the power of the exponential function which also bites unsuspecting debtors by compound interest. The human brain obviously is not well equipped to understand exponential functions. Yeah, you can always do the math. But you can't grasp it intuitively. So you can't "feel" the danger. And thus, after 30 years, the poor mortgaged homeowner has already paid back the originally borrowed amount twice or trice, and still does not own the house. Same bad surprise for those who venture out to build Apple-1 with untested ICs: the money's lost. Disappeared into a sinkhole, never to come back. Don't do that. Use known good ICs, always.

Comments invited !

Off topic,  but the Wickuler Pils is making me thirsty...  alas I'm on antibiotics so no beer for me for another week...  :-(

I will probably get the RCT in the future as i read good reviews.

But for "normal" TTL IC's family 74xx i use this one and i love it. I like very much that is terminal based. But more than that, the ic test output gives you great informations like Vss mA  and much more. 

Have a look here: http://www.tauntek.com/LogICTester-low-cost-logic-chip-tester.htm

Dennis....

How did you manage to find that blow mold case to fit that retro chip tester so perfectly?

I really didnt know what to search for besides "blow mold case"  I found your case out of UK for reasonable price and ordered one.  Amazing that thing fits so perfect!

I dont use it much and most of the 3d printed ones were either "too fancy" or too plain.  This one is perfect for storing it on the shelf when not in use.

Thanks again!