What's the issue with LSTTL in Apple-1 ?

5 posts / 0 new
Last post
Last seen: 10 hours 1 min ago
Joined: Apr 1 2020 - 16:46
Posts: 493
What's the issue with LSTTL in Apple-1 ?

I have just completed another little experiment and replaced all the 74xxx ICs in my newest Apple-1 build (lucky number 7) with LSTTL. The Schottky muxes I did leave alone. Power drain on the +5V rail went down from 1.4A to just 0.95A, which is good, so I can use a weaker TO-3 regulator rated for 1.5A (and able to do 2A max) and a smaller low profile heat sink. These weaker regulators also happen to have the cherished "golden" baseplate that appears to be unobtainium for the LM323K, but was spotted in some original Apple-1 (note that the stamp on the regulator always can be altered, but turning  a "steel" baseplate into a "golden" baseplate is much more tricky).


This said, my question for which the search function on this forum did not yield results: I seem to remember that somebody did claim somewhere that in the Apple-1, the 74161 can't be substituted by 74LS161 because the former had a bug (or quirk) the Apple-1 allegedly needs to function properly, and the 74LS161 don't have that quirk/bug/feature and hence, would not work in the Apple-1.


But my Apple-1 seems to work with all the 74161 replaced by 74LS161 (from Texas Instruments, if that helps).


So what is wrong ? Is that hidden feature really there ? Can anyone enlighten me / give links to more info ?


Last seen: 2 days 9 hours ago
Joined: Dec 17 2020 - 09:16
Posts: 21
Thats an interesting result

Thats an interesting result on the current drop! Does it help to run the 323K a bit cooler? Thats my biggest safety concern on the Apple-1 design.


Regarding the 74161 you can have a look in Mike Willegal Blog 11 years ago when he did a deeper investigation on that topic:


Last seen: 10 hours 1 min ago
Joined: Apr 1 2020 - 16:46
Posts: 493
About the LM323K temperature ...

... under normal conditions there is no reason to worry, but alas, things can and do go wrong. I recommend you to get one of these cheap contactless thermometers (available online and in many car parts shops) which work by looking at the infrared. I paid less than $10 for mine and it works fine and it amazingly accurate, although it does not work very well on shiny objects, where it produces misreadings. To measure the temperature of the LM323K, don't point it at the shiny TO-3 case but at the matte black cooling fins.


I found out that a typical Apple-1 build running off an 8Vac transformer for the +5V supply has a temperature elevation of 50 deg Celsius (or Kelvin as for the delta K) over ambient temperature. So with 18 deg Celsius room temperature, the LM323K runs at 68 deg Celsius, which is OK for it. These regulators in the commercial temperature range (0...70 deg Celsisus) are far below the temperature levels they really can take. They could run much hotter without problems. BTW, this data applies to Apple-1 with 10.2Vdc on the 5300uF/15V filter cap. If you have a higher voltage there, then the LM323K will run hotter.


These regulators, unless defective, are self-protecting, so they can't suffer from thermal runaway an unprotected NPN pass transistor (such as the 2N3055) could die from. I have seen 2N3055 glowing dark red in linear power supplies. (Oooops).

This won't happen with any of the Bob Widlar designed linear regulators and all the knockoffs made by other manufacturers than National Semiconductors, as all include the thermal protection circuits. This explains why these regulator ICs (78XX, 79XX, LM105, 205, 305, 323 etc. etc.) became immensly popular and are still in production half a century after they were released on the marketplace. They are virtually indestructible (although I know how to do that within their specified input voltages, part of the vast amounts of useless knowledge I have, such as designing two-phase NMOS dynamic logic, which was used by the original 6502.)

The caveat here is that once the thermal protection kicks in, the regulator will reduce its output voltage until the power dissipation is manageable by the heat sink. Which means your "+5V" supply may go down to 3V or less. And the TTLs won't work there anymore.

Last seen: 1 year 5 months ago
Joined: Apr 7 2021 - 19:16
Posts: 4

Dare I ask if anyone has tried HC or HCT parts?

I'm just starting my first Apple 1 clone build and thought I'd give it a shot.


P.S. I've been burned as a heretic before, so I'll probably survive : )


Full disclosure, I'm planning to use a switcher, too.

Last seen: 10 hours 1 min ago
Joined: Apr 1 2020 - 16:46
Posts: 493
I recommend to avoid HC and HCT ...

... in your first Apple-1 build. And based on my own LSTTL build I must issue a warning against LSTTL, too: for me it worked fine until a few MM1404 shift registers started to act erratically. They were from the pile of "questionable" ones which were fall-outs from my production of 100% tested and burned-in IC kits, so this was no surprise. The bad surprise came when I changed all the dubious MM1404 (which I should have thrown into the trash instead of trying to reuse them) against Am1404. Which work fine in any TTL based Apple-1. But they did not work in my LSTTL based build at all. No video. And the AM1404 were known good. When put back into the burn-in Apple-1 they came from they worked again.


Conclusions / Warnings / Advice:


- 74HCxxx has the wrong logic level thresholds on its inputs if running from a 5V supply.

- 74HCTxxx has the right (TTL) logic level thresholds but it will whack the too thin and too weak Apple-1 power and ground lines badly.

- all CMOS ICs able to see the wicked negative logic swings of the PMOS shift registers may latch up, get hot, and start smoking.

- 74LSxxx does not want to cooperate with AM1404 shift registers. But it works with MM1404. The Signetics 2504 I don't have so I didn't try them.


I think the best course of action for a 1st build is to use a known good, TTL based, IC set. You can get one from me, or take it out from a known working Apple-1 clone. When your first build works, you can start to experiment and slowly change ICs one by one to morph it into the kind of machine you want.

I think that the TTL -> LSTTL change is the one with the best chances of success and less frustrations. LSTTL was specifically designed as a drop-in replacement for TTL and so the designers of these 74LSxxx tried to match the speed of the 74xxx counterparts as closely as possible.

All other logic families have vastly different speeds and power supply effects and in the Apple-1 context this spells: T-R-O-U-B-L-E.


But some people like to get themselves into trouble. Such as debugging of marginally stable hardware.


Good luck !



Log in or register to post comments