The y have used Fairchild for their source. If so, that explains the high rate of failure.
Years ago when at Intel, we disqualified Fairchild as a source for parts because of the poor testing and high failure rates. Dwight ________________________________ From: cctalk <cctalk-boun...@classiccmp.org> on behalf of curiousma...@gmail.com <curiousma...@gmail.com> Sent: Thursday, August 11, 2016 12:12:14 AM To: General Discussion: On-Topic and Off-Topic Posts Subject: Re: the value of old test and repair equipment For some reason the 7474's have a higher failure rate than other TTL IC's in HP equipment. I don't know if it's true in general for 1970's TTL. Signetics MSI chips (counters and stuff) seem to be prone to failure too. Marc > > All this talk of older test equipment reminds me of the HP 4261A LCR > bridge I repaired a while back, last winter I think. > > > My dad found the 4261A in the garbage years ago, and it seemed to work > fine, until one day he powered it up and the display showed garbage. > He decided to open it up, and noticed some uncovered windowed EPROMs. > Knowing that EPROMs sometimes flip bits in their old age, we decided > that was the first place we would look. We were also able to locate > the full HP service manual in PDF form for the instrument which helped > tremendously. In typical HP fashion, it had full theory of operation, > schematics, state diagrams, etc. > > > Now, I have an EPROM burner that does your typical JEDEC pinout parts, > 27 series and such. The issue is that these were Intel i1702A's from > the early 70's I think. Not only are 1702's a totally different > pinout, but they run on 14V (a +5V, and a -9V rail, with no connected > ground, this is how intel got TTL levels on a MOS chip at the time). > The 4261A has a total of 4 1702's, two of which form a finite state > machine which controls the instrument, while the other two perform > display decoding. > > > I had to pull out my dad's DeVry Console 80, which has adjustable > positive and negative supplies, and I manually clocked out the data > and compared the contents to a dump I found online. I started with the > state machine EPROMs, and compared the data. I did find a few > discrepancies, but there was too much difference to have been bit rot. > Given the sudden nature of the issue, I would have expected one, at > most a couple bit flips, or something much more drastic (like total > chip failure). Upon reading through the state diagrams in the HP > manual, I noticed that there was a change noted in the state diagram > between certain minor revisions of the 4261A. I looked at what the > changes were, and deduced that my ROMs were in fact correct for the > serial number prefix. > > > At a dead end with the EPROMs, I decided to see if the state machine > was even running at all. I used a DVM in DC mode, and measured perfect > TTL ones and zeroes on all the state number outputs, which means those > outputs weren't changing: the state machine was stuck. I wrote down > the state it was stuck in and referred to the state diagram. I noticed > something interesting. The state machine in the 4261A is able to > evaluate simple conditions and control flow based on those. The state > path to get to the state that the FSM was stuck on meant the FSM was > always taking one of the conditional paths (always true, or always > false, I don't remember which). At that point, I started looking into > the condition circuitry, tracing out the path, checking IC's as I > worked my way back, until I made it back to 1/2 of a 7474 which had a > set input that was stuck active (low). This pin went to a pullup > resistor, and nothing else in our unit (certain options used this pin, > but not ours). We desoldered the IC, and sure enough, that pin was > shorted to ground internal to the chip. We replaced it with a 74LS74, > and the 4261A has been working great ever since, even with the > original 40 year old 1702's. > > > Also, on the topic of interesting HP products, and perhaps my personal > favorite so far, is the HP dynamic signal analyzer 35670A. This > instrument can perform all sorts of cool measurements. It can produce > a test signal, and measure two different points in the circuit being > measured. The measurement input channels give you a complex number > phasor of the measured signal, which means you can do all sorts of > cool measurements of networks, especially since you can do complex > number math with the equation support of the instrument. The signal > generator will perform sweeps too, of course. This was very useful > determining whether the speaker crossovers my dad built were working > as intended (actually they weren't, and this instrument helped us > uncover a problem). We also used this to do inductor and capacitor > characterization. There are all sorts of applications this instrument > is good for. > > > Joe Zatarski
