My new Apple1 Replicas (TTL / LS-TTL) and cards

In post #1, 'vossi' wrote:

" ... I have found that the 74123 must not be LS – I have tried various ones and the pulses are much longer than 480ns ... "

Uncle Bernie comments:

My own LSTTL based Apple-1 uses a 74LS123 with no problems. But I had to adjust both timing resistors to get the correct oneshot pulse width. But ... I never found a 74123 which would produce the right pulse width with the R/C values on the motherboard. With a 47pF capacitor, I typically needed ~22k resistors, and not the 27k specified in the original design, to arrive at 480ns.

This timing is not super critical but 480ns is at the optimum "sweet spot" according to theoretical calculations based on the 6502 and the DRAM datasheets. So Woz knew what he was doing. The irony is that he could have used a fully digital solution and could have avoided the 74123 completely. But even for his choice of using a 74123 there is a logical explanation: more flexibility on which DRAMs can be used.

Still, use of any such oneshot of any kind is frowned upon in the electronics industry, and for a reason: it can't be used for tighter timing specifications without adding a trim pot. And trimming of anything must be avoided in mass production . Unless your product is so expensive and highly profitable that you can afford to crank trim pots (and trim C's and L's, too).

- Uncle Bernie

In post #3, 'vossi' wrote:

" Maybe you had luck with your 74LS123. From what company and with what datecode is yours ? "

Uncle Bernie comments:

In my professional realm, "luck" does not matter as I never rely on "luck" to make something work. In any field of engineering,  relying on "luck" for a positive outcome is futile and may even be deadly. Or do you want to walk over a bridge which was designed betting on "luck" instead on thorough structural calculations ? Or fly in an airplane which was designed by rules of thumb ?

There are certain projects even a hobbyist not having a well equipped lab with all the usual instrumentation can bring to a successful end, but building an Apple-1 using LSTTL ICs is not among these easy hobby projects. You need at least a good oscilloscope with more than 20 MHz bandwidth to be able to measure and dial in the timing of the 74123 oneshots.

I've used a National Semiconductor DM74LS123N with date code B8145, but it's a futile exercise if you try to hunt for the same date code. This is not how the world works ! Use a suitable oscilloscope and some trim pots ... put 5k in series with the trim pot to avoid killing the 74123 on the minimum setting of the trim pot.

Using an oscilloscope I arrived at 15k/47pF and 8.2k/1nF in my case, for said DM74LS123. Yours may need somewhat different resistor values. But in general, it has been my observation that LSTTL oneshots typically require lower RC time constants to arrive at the same pulse width as a TTL family oneshot.

In my famous Apple-1 IC kits, I provided hand selected R's and C's for a given specimen of 74123, and even within ICs from one original manufacturer's tube and with the same date code, I saw large enough variations that required  a set of (binned) R's and C's from 38pF to 47pF and from 20kOhm to 24 kOhm.

This is why such analog / RC based oneshots are frowned upon in professionaly designed electronics needing tight timing margins. But they are good to use for wider, noncritical timing margins, or, if equipped with a trim pot, to compensate for unavoidable variations in some other component or subassembly. But once something needs to be trimmed on the assembly line, then this increases costs, not only for the person who does the trimming work, but also for the instrumentation on that station.

- Uncle Bernie