Hard to source components as of late 2023

For the 2504 you will have to give about 18-20$ a piece, unfortunately this is the new reality. There are replacements for the 2513 and 2519 on modern components from the esteemed P-Lab. When I was a seller on eBay I used to build them, it works great.

Of the rare/expensive ones, there's also a DS0025 and a 16xMK4096/MK4027. If you want I can give you the contact of the seller of the 2504 and 4027, he has everything tested but he's in the EU.

About the 74ls... I wouldn't recommend to use them at all, some of them will work, some of them won't...

DS0025 = MH0025 = К170АП3 (USSR)

Macintosh_nik wrote:

DS0025 = MH0025 = К170АП3 (USSR)

Isn't the АП3 actually a DS0026, requiring some level mods? Source: https://www.applefritter.com/comment/87269#comment-87269

ebruchez wrote:

...

So far I don't see a trace of that chip online.

Check eBay: click!

I am not a big connoisseur of Soviet microcircuits, in my case К170АП3 worked without any modification on the board, but DS0026 did not.

No need to be one, there are lots of online catalogs indexed by Google, like this one: https://files.elektroda.pl/372530,odpowiedniki2.html

In post #4, macintosh_nik wrote:

DS0025 = MH0025 = К170АП3 (USSR)

Uncle Bernie comments:

I know the pain with these DS0025 which once were available below $1 each, and now the prices went stratospheric (I've seen them for $50 each).

You can always use the DS0026, which I had in my first kits, and it worked, but since it's faster and produces steeper edges than the DS0025, you really need all the 'reliability' mods I proposed (and published here on Applefritter long ago). With the DS0025 and robust DRAMs like the Intersil ones I had in my last kits, you could get away with just adding the six damping resistors, and sometimes, even those were not needed and Apple-1 worked fine without any 'reliability mods'. How close these are to the 'unreliability cliff' I don't know (and don't want to know ;-)

Be aware that Chinese counterfeiters lurking on Ebay, Alibaba and elsewhere sell fake 'DS0025' and 'DS0026' which are NOT the real thing, and I got scammed, too (but got my money back, after a time expenditure for complaints, clawback via credit card company, etc.) of my precious RQLT which may have been worth 1000 times more than the fraudulent merchandize. But I did learn something interesting from that, too:

These counterfeiters re-stamp industry standard MOS gate drivers as 'DS0025' or 'DS0026'. These industry standard parts are made in CMOS technologies are used by the billions in the automotive industry and elsewhere, whereever the gate of a beefy MOSFET power switch needs to be driven by a small microcontroller or the like.  These ICs have the same pinout as the 'DS0025' or 'DS0026' but there is a catch (actually, two):

1. lack of the pulldown resistors on the inputs

2. extremely high driving strength (they need to drive several nanofarads of MOS gate capacitance in just a few 10's of ns)

The pulldown resistors can be added (maybe 2kOhm to the most negative supply, not sure, need to look it up), it's trivial, but the current spikes produced by these little 'monsters' would immediately crash any Apple-1 with its weak power and ground rails and the lack of adequate bypass capacitors. But if you put two 39 Ohm ... 75 Ohm series resistors (one each)  into the outputs of these drivers to PHI3 and PHI4 on the Apple-1 motherboard, then they work fine !

I did that to find out what the mystery parts of the counterfeiters were. And I once knew the manufacturer and type number of these parts, but somehow lost my notes on them, or can't find them.

There is another small problem, and that is that these CMOS drivers typically are not specified for operation at 17 volts supply voltage (in the Apple-1, -12V and +5V), but only for 15V. One example is the Renesas EL7212, which has an abs max supply voltage spec of 16.5V ... altough it is not recommended by the manufacturer to operate the device at the absolute maximums, I did not succeed to blow one up in the Apple-1.

So what I did learn from the Chinese scammers was that plenty of MOS drivers with the same footprint as the DS0025/0026 do exist, so there is no need to pay moon prices for real DS0025/0026 (or fake ones).

- Uncle Bernie

UncleBernie wrote:

In post #4, macintosh_nik wrote:

 

DS0025 = MH0025 = К170АП3 (USSR)

 

Uncle Bernie comments:

 

I know the pain with these DS0025 which once were available below $1 each, and now the prices went stratospheric (I've seen them for $50 each).

 

You can always use the DS0026, which I had in my first kits, and it worked,

So would it be possible to build a small adapter board using the Chinese CMOS parts and necessary discreet components that would plug into the Apple-1 board?

In post #9, softwarejanitor wrote:

"So would it be possible to build a small adapter board using the Chinese CMOS parts and necessary discreet components that would plug into the Apple-1 board ?"

Uncle Bernie answers:

this would be even better because then you could use the SMD versions of these drivers which are more abundant (and cheaper) than the SMD versions. Mouser wants $6.97 for the DIL-8 @ qty 1 ( Mouser #968-EL7212CNZ )  while the SMD costs $2.99 at qty 1 (Mouser # 968-EL7212CSZ ).

The  ICL7667 was advertised as a DS0026 substitute and is specified for operation at 17V (which fits the Apple-1 exactly).

The MAX626 is specified for up to 18V.

So once you start substituting, you have plenty of choices. Some applications engineering may be required (how to bias the inputs / size the pulldown resistor, how much series resistance in the outputs, how much  good  bypass capacitor at the device), but I see no obstacles that could not be overcome. As I said, these MOS drivers with the same footprint as the DS0025/0026 were made by the billions. Practically every beefy power MOSFET in any application needs such a driver.

So no need to pay moon prices.

- Uncle Bernie

(P.S.: when I have some time, I think I have a photo, or can snap one, of an Apple-1 which uses those).

If you use an SMD version you can probably fit everything onto a carrier which isn't too much different in size than the original chip, so that would probably be a bonus actually.  Enough work-around like this and it may be possible for people to continue Apple-1 builds a while longer even though we're getting down to where a lot of the original prices as you mentioned are as far as the moon.

Here is the promised photo, snapped it fresh, since I could not find the older photo:

DS0025mod.jpg

You can see that the pulldown resistors at the inputs are 2k Ohm, and the series resistors are 47 Ohms, all SMDs.

The bypass capacitor is 100nF in a glass package. Note the shortest possible wires to the +V / -V supplies.

This modified Apple-1 is one of my 'lab rats' and it ran fine with the Chinese counterfeits.

Currently I have plugged  in a DS0025. No changes necessary. All the candidates work in the same socket.

- Uncle Bernie

Nice!  The advantage to doing it that way is on the top side of the board it probably looks more correct.  A plug in adapter board would be nice in not having to mod the board like that.  If you can find your notes or recall or otherwise again deduce the chip the Chinese counterfieters are uisng and document it here that would probably be awesome for posterity.

In post #13, softwarejanitor wrote:

"If you can find your notes or recall or otherwise again deduce the chip the Chinese counterfieters are uisng and document it here that would probably be awesome for posterity. "

Uncle Bernie comments:

I don't think it's necessary to buy these IC from the Chinese counterfeiters. You can get any of the types I listed above cheaper, as originals, so why buy dubious counterfeits ?

I understand that people may be scared it wouldn't work, if they just go ahead and try them out, but as long as you don't put the 2504 shift registers in before you have confirmed that the two phase clock at PHI3 and PHI4 is in spec (swings from -12V to +5V with no overlap of the -12V zone) then nothing can go wrong. All you need is an oscilloscope. Once you know the clocks are good, you can put in the 2504 again.

I would not try this with the 2504 in the sockets from the beginning. If the PHI3, PHI4 clocks are wrong (i.e. the negative voltage portion overlaps), the dynamic logic in the shift registers will have some shootthrough current. Which does not kill them instantly, but it is a stress that can be avoided by verifying the clocks first.

- Uncle Bernie

I was just being dense...  I see now that you have provided equivalent part numbers.  I wouldn't buy them from the counterfieters either.  No need to reward them for what they do.

In post #16, ebruchetz wrote:

" The rest will come together as well, I hope. I am going the eBay route for the 7 x 2504, 2513, and the 16 x MK4027. Crossing fingers.  "

Uncle Bernie comments:

For the 2504, there are plenty of drop-in substitutes, such as the AM2804. These are also used  in U.S. weapons systems and have a NSN (National Stockpile Number). From time to time, a batch is made for the defense industry and the excess ends up at IC brokers. So from time to time, a few thousand of them become available at the IC brokers.

I also know that there is one IC broker who sits on thousands of 2504/1404 in the metal can, and the price was below $5 each. But I did not buy any as the minimum order would have left me with far too many I could not resell (you need to build a small DIL-16 carrier with two of them, as the pinout is a bit different than the DIL-8 version). There are some photos around of Apple-1 with these TO cans.

The 2513, be careful. No each 2513 is the same thing. The problem is that "2513" is a family of mask programmable character generator ROMs, and Signetics offered several different character sets, uppercase, lowercase, etc. The "CM" number identifies the character set. CM2140 is the one you want. Just ~2 years ago I found more than 1000 Signetics 2513 at an IC broker in Mexico City. Alas, they turned out to have the wrong "CM" number, so they were useless for the Apple-1.  Beginning with 1977 or 1978, Signetics tried to avoid the confusion and changed the type number to "CM2140".

I would search only for "CM2140".

General Instrument made a drop-in replacement, the RO-3-2513 CGR001, sometimes also stamped RO3-2513 CGR0001. Again, the "CGR" number is key. All other numbers other than 001 and 0001 have a wrong character set.

For the DRAM, you can use any 1st and 2nd generation 4k x 1 DRAM. With 1st generation, you need to watch out for the speed grade, some are slow (often, due to a metal gate process). Most 2nd generation 4k x 1 DRAMs used silicon gate processes and were much faster (I never encountered one that would have been too slow for the Apple-1).

So in the end, it's only the 2519 which is troublesome (there never were 2nd sources), but a solution with SMD CMOS 64 bit shift registers (programmed for 40 bits length) exists.

- Uncle Bernie

I've built completely on SMD components and hybrid version, in my opinion the first one is cooler. About the Lithuanian seller it is a good choice. I've known him for a long time, he has a 2504 tester of his own design. And in general he is a very reliable seller. Who is interested I can give his email, so it will get a little cheaper. It's just a pity that now you can send only postcards from Lithuania to Russia....

IMG_20231114_081533.jpg

02.jpg

Funny to see my name on that.

I created the big design just to test if that works.

I also created a SMD-version: https://oshwlab.com/szillat/2519-fixed-smd

But I like the P-LAB version more as it is smaller and more elegant.

The design files are open source you can mod them to match your foot print.

Thank you Natas for the appreciation.. :-)

Of course it is all open sourced, I've just uploaded the EasyEDA export of the PCB in case someone wants to change anything.

The schematic is still the one on Natas's repo linked from my page https://p-l4b.github.io/2519/

BTW, it seems that now 4557 is available only in SOIC-16-Wide case, instead of the SOIC-16 I used... <facepalm>

@ebruchez:

Apparently DIGIKEY has a few hundreds of SOIC-16 (regular, not wide) HEF4557BT that should match my original PCB:

https://www.digikey.it/en/products/detail/rochester-electronics-llc/HEF4557BT-652/15637775

Enjoy your build! :-)

Regards,

C.

Ops!

I didn't notice that... small detail.

Sorry about that!   ;-)

I buy here, I don't have much more options.

https://sl.aliexpress.ru/p?key=bzvoUGa