A replacement for the HP-41 zebra connector

[mfleming]

Hi everyone,

I've been working on a replacement for the zebra connector between the 41 CPU board and the keyboard PCB. I've had no luck finding a source of replacement elastomer conductive strips, at least not in hobbyist volumes. I've also been unsatisfied with alternatives to the stock connector. After trying a couple of ideas I've hit on something that works for the calculators in my collection. I'm using a flex circuit to bring signals from the keyboard PCB contacts to the CPU PCB contacts.

       

The first picture shows the flex circuit on the trace side next to a pair of zebra connectors for scale. Note the flex contacts are gold plated and the traces between them are covered with a plastic layer to prevent shorts. The second picture shows a pair of rubber cords, held in place with adhesive tape, that serve to press the flex contacts against their PCB contact counterparts when a case is closed and screwed tight.

       

These other two pictures show the flex circuit after it has been folded over to fit the lower posts in the calculator body. The flex can take a crease to keep it folded so long as the crease is not so sharp as to damage traces. The final photo shows the flex in place before placing the CPU board on top. Closing the case puts pressure on the CPU board and presses the flex contacts against the boards, just like the stock zebra connector.

I experimented with a number of material types and shapes before settling on what works very consistently for me. I only have three calculators to work with though, all the same CV model and all produced within a two year span. There could be some difference in models or production method changes that might affect the working of the replacement.

What I'd like to do is get a few people to try the replacement on a calculator that has a missing or flaky zebra connector. The idea would be for me to get some feedback on ease of installation, any fitment problems you might have, and most important any connectivity problems. The calculator you repair should be exercised about once a day to see if any problems pop up. I usually run through all the alpha characters and do a few calculations while holding a calculator in hand on or the desk each day. With a module plugged in, a CAT 2 exercises the I/O bus. That's all I would ask of any volunteer.

Hopefully this sort of "field trial" would prove the viability of this replacement and serve as input for a final design update. I have enough remaining material for eight people. I'd also like to get a idea of the level of interest from everyone in a zebra replacement. More volume obviously would mean lower cost.

PM me if you'd like to give the replacement a try and are willing to provide some feedback on your results. Let me know the model and year of the calculator you would use. I'm quite sure this will prove to be a successful substitute for an aging part that is no longer produced!

~Mark

[Monte Dalrymple]

I can try one on four machines 2005A, 2025A, 2205S, and 2506S. I'll probably be using a CL board though, if that's okay.

Monte

[anetzer]

(03-17-2017 01:18 AM)mfleming Wrote:  I experimented with a number of material types and shapes before settling on what works very consistently for me.

It will be very helpful if people post their experiences to the thread, including calculator (and board) details, description of material used and sources to obtain it (hopefully for more than one continent...).

And then we sit back and wait for the posts to crack open!

[Sylvain Cote]

If you are willing to ship to Canada, I can try one.
I have several fullnut HP-41 (C/CV/CX/CL) that I can test it with.
I am using my HP-41's daily.
Sylvain

[mfleming]

(03-17-2017 01:13 PM)Sylvain Cote Wrote:  If you are willing to ship to Canada, I can try one.
I have several fullnut HP-41 (C/CV/CX/CL) that I can test it with.
I am using my HP-41's daily.
Sylvain

First, in answer to another member's question, this trial replacement part is free and is not limited to the U.S. I checked International First Class rates for a letter and it is really not all that bad.

As anetzer suggests, this thread would be a good place to post hardware used, results and so on for others to see. I'm hoping that with successful trial results this can be another offering to members like Michael's keyboard overlays.

There are a couple of areas I'd like to validate with wider testing that I'll include in the instructions. For example, in the initial design I left the area between rows of contacts empty. I thought I might need to remove some of the flex material to accommodate protrusions like the heat stake on the bottom PCB and solder lumps on the CPU PCB. It wasn't necessary for me, but before I make a design change to move traces and reduce flex circuit size, I'd like to be sure it is not a problem for anyone else.

Lastly, I'd love to see it tested on a CL board! When I first got mine last summer (thanks!) I had some intermittent contact problems that finally cleared up after a liberal cleaning of the zebra with DeoxIT Red, then a coat of Gold. Eventually though, that old part (2113A) will fail...

~Mark

[Nate Martin]

I have several salvage 41's I would be willing to try on. This sounds like a great idea.

[mfleming]

I've posted letters today to those who've agreed to test the flex circuit as a zebra connector replacement. The letter includes all material needed to assemble the zebra replacement, including two types of the cord material. Much to my surprise (though not surprising) a bumpy letter that can't go through an automatic sorting machine requires an extra twenty-one cent stamp. Ah well!

Just an update to those receiving the letter, the gray spline cord material seems to deform under pressure with time, and doesn't immediately spring back. As a result, opening and then re-closing the case will result in a contact failure. So, scratch that choice of material from the list of supplies...

~ Mark

[Sylvain Cote]

Hello Mark,

I have received the letter Friday afternoon.
I will install it somewhere this week.

Best regards,

Sylvain

[mfleming]

(03-26-2017 08:08 PM)Sylvain Cote Wrote:  Hello Mark,

I have received the letter Friday afternoon.
I will install it somewhere this week.

Best regards,

Sylvain

I'm quite surprised how quickly the letters were received by others as well. I really expected at least ten days transit time. I can't complain though!

I do have an update regarding the cord material included for standoffs. The hollow core of the cord allows for compression, but it does not spring back very quickly after the case is opened. As a result, the flex connector seems to fail when the case is reclosed unless the cord is squeezed back to a round shape or replaced with fresh material. The cord will rebound to normal shape though after a couple of days. The photo below shows compressed gray cords that have returned to shape and a freshly removed black cord placed next to them for comparison.
   

Another member, Hans Brueggemann, suggested a solid silicone cord found here:

Quote: have a look at these, 3mm dia. should be way to go:
http://www.silex.co.uk/products/solid-silicone-cord/

I've found a large selection of such material from McMaster-Carr here in the U.S.
https://www.mcmaster.com/#silicone-o-rin...k/=16wtrk6

Just to test the limits of material hardness, I tried 16 gauge stranded copper wire with an insulation jacket, a round wire you might find in appliance wiring. With a bit of reluctance (garbled display, singing buzzer) the calculator actually started up after a couple of battery pulls and seems to work just fine. I definitely don't plan to leave it this way, but I will check on it daily for a week to see how the connector holds up.

More soon I hope,
~ Mark

[nfmax]

Just seen this!

I have a 41CV, s/n 2227S... which originally had the gold wire connection system, eaten by battery neglect. I found a supplier of salvaged ZebraStrip connectors (from scrap laptop LCD displays) and ordered a few of what looked like the right size, but they didn't work reliably. I ended up wrapping loops of silver plated wire-wrap wire around the ZS rubber, spaced on a jig to be 0.1" apart, and painstakingly aligned with the PCB contacts. It did work, but not reliably. Your flex circuit approach looks much more promising. Seems I've missed the boat this time, but if you do go into production, please let me know!

[anetzer]

So...

Silicone tubing seems to be what most think to be a good choice.

Amazon in Germany seems to be a simple source for tubes of different measures, e.g.

https://www.amazon.de/gp/aw/d/B00K5TQG1I...1evynPOAcL

where 1x2mm means 1mm inner and 2mm outer diameter.

Question to those who tried: What size would you go for in terms of total, compacted silicone thickness? Is 2 mm enough or should one look to 4, 5mm even?

Thanks!

[AndiGer]

Anybody thinking to reuse the black plastic piece the original zebra is put on/between?
So you have to cut the slices to 40mm and place them accurately.
Didn't try the flex yet.
Andi

[Geoff Quickfall]

BIG REMEMBER (reminder) HERE

About bottom screw posts. The screws have a very short thread count, or if factory replaced at some point a slightly longer thread count. There are two types of case backs, early versions that has hex nuts hold the PCA without encasing the screw post and later versions that press down and do encase the screw posts. (Pictures tomorrow). If you are using an early version 41C with nuts holding the PCA in place you will definitely have the early case back.

The early mechanics had the nuts hold the PCA in place. The later version has the case back hold the PCA in place. The later versions ar more susceptible to overs stressing and cracking. So be judicious on the diameter of silicon tubing realizing that only two screws are holding the PCA and zebra

The Silicon tubing should be an optimal size, enough to ensure contact even during a short distortion of the case but not so large they add stress to the WEAKEST point of the 41C series except for the halfnuts. I will measure the existing tube diameter used in the most robust connector tomorrow. The least robust being the strips with wire wrap and a foam insert. Speaking of foam inserts this is the main reason for the battery/module block failure. They too, need the foam replaced due to,compression fatigue.

I just went through my 41c parts box here and reassembled 3 pre 1980 41Cs, a CV, a halfnut 41CX this weekend, three card readers and a RAM Box. I noticed in the restorations I had to reinforce and repair a multitude of screw posts.


Geoff

This connector type has a silicon tube in the middle of the curved edges:



This system has the foam strips:



These are the hex nuts that hold the early version PCA onto the contact strips (the balls and springs were for the non existent external power supply contact points):

[Dave Frederickson]

FWIW, Fujipoly is a "golden zebra" manufacturer.
http://www.fujipoly.com/usa/products/zeb...ctors.html

I tried to sample a replacement "golden zebra" for a 71B, but despite calls to the company I'm afraid my request didn't measure up to their minimum requirements.

Dave

[Geoff Quickfall]

Those are almost identical to the foam strip version except silicone versus foam. To bad one cannot purchase a length for 'testing' :-).

Geoff

[mfleming]

(03-29-2017 03:39 AM)AndiGer Wrote:  Anybody thinking to reuse the black plastic piece the original zebra is put on/between?
So you have to cut the slices to 40mm and place them accurately.
Didn't try the flex yet.
Andi

First thing I tried! Unfortunately, a zebra connector that is compressed too far will not hold the flex contacts against the corresponding PCB contacts. So far, a round support strip shape works best, and a hollow core cord (or tubing) allows some give to the material as the case is screwed closed. The drawback to the material I'm using is that it takes about 48 hours to recover its shape should you wish to open the case and adjust the connector. If you don't wait for recovery, the connector suddenly doesn't work anymore!

I've ordered solid cord from McMaster, which should arrive tomorrow. Solid cord may do a better job of recovering from compression, but the diameter is likely to be more critical as far as post stress is concerned. Hopefully the right arrangement that works for all the variants Geoff listed will be determined soon.

~Mark

[Sylvain Cote]

Hello Mark,

Test subject: HP-41C fullnut S/N:2340S40422 (fully working)

I have successfully replaced the original connector with your flex and the calculator is working perfectly.

The 0.140" black spline is too tick, I cannot close the shell with it.
The 0.125" grey spline is still too tick but I can close the shell although not completely. (see pictures)

Regards,

Sylvain

[Nate Martin]

Tested on a couple of 41's had some initial success using just a foam insert as both of the splines were too thick, and very difficult to close the 41's...they worked with the thinner spline but took some maneuvering and extra hands. I also used the black spacer from the original zebra connector with the splines. The foam insert was just thin packing foam from a mailer padding that I cut out.

using an older motherboard that had the nuts to keep it in place on the posts, the backside of the motherboard was too rough with the solder connections and two penetrated one of the line traces. I will try to post a pic soon. The trace still has continuity but it looks like it hit the trace dead on which could cause a short with those solder connections if not a broken trace. I could probably resolve this issue with a little klapton tape on the motherboard before installation.

The divots did not penetrate the flexcircuit, and were in the area with the old black zebra connector holder.

[mfleming]

(03-31-2017 02:44 AM)Nate Martin Wrote:  using an older motherboard that had the nuts to keep it in place on the posts, the backside of the motherboard was too rough with the solder connections and two penetrated one of the line traces. I will try to post a pic soon. The trace still has continuity but it looks like it hit the trace dead on which could cause a short with those solder connections if not a broken trace. I could probably resolve this issue with a little klapton tape on the motherboard before installation.

I would definitely like to see a picture! I've had the flex material distort and dimple from solder bumps that lay along a line drawn between the two support posts, but I've not seen a CPU board with protrusions outside of the contact footprint. A final design of the flex connector replacement should not extend past the left/right boundaries of that footprint. Was the flex material actually punctured, or just severely distorted?

Based on the thickness of the zebra connector spacer frame, I think a support material diameter of 3/32" (2.5mm metric, or about one seven-hundredth of a fathom ) would be about the right size. Both splines I supplied were 25% - 50% larger, which would account for the poor fit.

Thanks for testing this on a calculator I don't have!
~Mark

[james summers]

Successfully installed (second attempt) on a 41CX SN 2406S.

On my first attempt there was no sign of life on replacing the batteries - I think this may have been due to one of the spacing splines shifting slightly as when I replaced the zebra connector as a check I noticed that one of the soldered component leg projections on the motherboard had dented one of the traces on the connector (circled on the picture). Looking at the projection on the motherboard, it did not look particularly severe (circled) and there were no marks on the connector for some of the more severe projections so I assume that the spline had shifted slightly so that it was under the offending projection.

I did find that the splines had a natural bow which tended to cause them to come unstuck form the doubled sided tape holding them in position. On my second attempt, I spent some time flexing the splines to try and remove the bow before reassembly and was rewarded with MEMORY LOST on reinserting the batteries. All keys work, CAT 2 lists the contents of an inserted module and a few trial calculations are all satisfactory.

I used the smaller grey splines which still seem slightly too large but there is only a small gap where the two case halves join at the front (picture with old zebra connector in front). When I first dismantled the 41 I noticed that one front and one rear post were cracked but as I did not have any suitable glue to hand I had to leave them as is for the moment and therefore did not want to tighten the screws any further.

I'll see how the CX performs over the next week and will then try the connector in another 41 and see how it goes.