HP 9825 Crowbar (Patreon)

I wanted to keep the crowbar design for the last episode and grand-finale of our HP 9825 repair series, but this is dragging on as we have been distracted by Apollo stuff and the re-emergence of Covid, and have not completely repaired the tape board yet (although it's mostly up now thanks to Eric). 

However, one of you requested the design so he could go ahead and power his units without fear, so I uploaded my crowbar design on my website. You can find it here:

https://www.curiousmarc.com/computing/hp-9825-scientific-computer#h.xrov16yr03br 

This is a reproduction of the original board that was developed by HP. I am not sure where and when this was developed, or if it was actually added to any production 9825. But obviously HP was aware of the problem and designed a fix. Fellow HP restorer Rik Bos found two examples of the board, not in an HP 9825, but in two of his HP 9835, both built in Germany. The later 9835 has the same form factor as the 9825, and an almost identical power supply. He sent a photo of his board, piggy backed on the output connector fingers of the original supply.

The part number on the board is 09825-66522, which suggests that the original intent was to use it on the 9825, although it has only be found on the 9835 so far. After a bit of perspective straightening in Photoshop, I got the picture below, from which I could extract the board dimensions:

You can pretty much guess the schematics by just looking at it and by where it is connecting on the fingers of the power supply. It has two Zener diodes that monitor the +5V and  +12V outputs of the power supply for over-voltage, which trigger a thyristor across the unregulated +20V, which is upstream of both regulators. That basically shorts the unregulated supply, which brings everything down downstream immediately, and blows the fuse upstream a few milliseconds later. The HP 9825 is saved.

Here are the inferred schematics:

On Rik's picture, you can barely discern the markings on the thyristor, it is a 2N4441. I happened to have the HP branded version of this thyristor in my collection of vintage components, which is a 1884-0213 in HP speak, as identified by my HP microfiches. That's what I put on my board, so it really looks like the original. You can find the 2N4441's on eBay, but I also provided a modern equivalent (BT258-500R) which will work just as well, is smaller, and widely available.

While I was at it, I went hunting for components that looked similar to the original ones, like the 1% 100 Ohms resistor with the 4 color bands (which I gather is unnecessary, a regular 5% would do just fine).  And also the unusual axial 0.01 uF cap, unfortunately the modern color is yellow, not teal. But I think it will still work, I'm told electrons are color blind ;-). I even reproduced the HP logo on the board, which is a trace underneath the solder mask. 

The original boards did not have a silk screen, so I made two versions of the board: one for purists without a silk screen, and one that has the silk screen. The one with the silk screen and is far friendlier to build, so it is the one I recommend. The ones in my 9825 won't have the screen, of course...

You can order the PCB directly from PCBWay using this link:

https://www.pcbway.com/project/shareproject/HP_09825_66552_Crowbar_Protecton_Circuit_for_the_HP_9825_Vintage_Calculator_a244bde6.html 

or use the Gerbers from my website and order it from your favorite supplier.

It installs by soldering the 4 pins and one wire without any modification of the original supply PCB: the pins go into via holes that are already present at the back of the contact fingers on the original supply. Now if you where wondering, like me, why they put an apparently unnecessary rectangular opening and a soft wire in the middle, look below at my installed board on my 9825 supply. You'll realize that they are needed to get around a capacitor that is on the 9825 supply, but was not present in the 9835 supply from Rik Bos. Mystery solved.

I of course tested the circuit, it correctly triggers below the maximum allowed TTL voltage, but never triggered in regular use over the very many power cycles we had during the repair, so everything seems fine. Hopefully all my 9825's are now protected from the nasty over-voltage failure.

Marc