IBM 5251 restoration (rebuilding phase)

User avatar
snacksthecat
✶✶✶✶

05 Jun 2019, 23:59

adamcobabe wrote:
05 Jun 2019, 11:26
Interesting. Seems a little risky to me given there is minimal stock to replace them if you mess up (from Orihalcon at $10 each). Did you bend the fly plate in the image?
I can't recall if I bent that one or if it was like that.

I decided to just do it because the keys felt uneven. To be honest, about 90% of them were perfectly fine. Most of the remaining 10% were only slightly off from the 1/2" sweet spot. A sad few had bad bends.

Image

Probably the more worthwhile activity for someone to do would be testing each key for feel, then doing the spring shaping only on the ones that felt off.

So now I have to put all these fly plates back on. The IBM manual again has some details on how to do this efficiently.

Image

I've been doing pretty much how the document says to. I used a small allen wrench because it's fatter than a paper clip so it doesn't slip away as easily. I mounted it vertically using a clamp so that I could use two hands to maneuver the switch around. Here's a visual:

User avatar
snacksthecat
✶✶✶✶

07 Jun 2019, 06:01

I'll have to do a deep cleaning of the edge connector. It is in pretty bad shape. Luckily this junk comes off pretty easily with a little bit of alcohol and something pokey.

Image

Image

User avatar
SneakyRobb
THINK

07 Jun 2019, 16:36

snacksthecat wrote:
07 Jun 2019, 06:01
I'll have to do a deep cleaning of the edge connector. It is in pretty bad shape. Luckily this junk comes off pretty easily with a little bit of alcohol and something pokey.

Image

Image
Isn't the silver there the solder finish to protect the copper trace layer under from oxidation? I wonder if this will affect long term connectivity.

Regardless your rebuild looks very good. Your team work to remove the switch stem is pretty great. I have removed those on my own and accidentally sliced my fingers open a few times. By far and away the hardest and only real dangerous part of the operation.

User avatar
snacksthecat
✶✶✶✶

07 Jun 2019, 17:42

SneakyRobb wrote:
07 Jun 2019, 16:36
Isn't the silver there the solder finish to protect the copper trace layer under from oxidation? I wonder if this will affect long term connectivity.
Yeah, unfortunately it still doesn't work with xwhatsit. Not sure what to do from here.

User avatar
SneakyRobb
THINK

07 Jun 2019, 19:26

snacksthecat wrote:
07 Jun 2019, 17:42
SneakyRobb wrote:
07 Jun 2019, 16:36
Isn't the silver there the solder finish to protect the copper trace layer under from oxidation? I wonder if this will affect long term connectivity.
Yeah, unfortunately it still doesn't work with xwhatsit. Not sure what to do from here.

Hi, that is most unfortunate. It sort of looks like the traces above your last shot corroded away? I would imagine if that is so they would not return signal.

It would cost you perhaps $50 +import/etc but you could always make a new PCB for it. I have had success making new beamspring PCBs with DMA and Wcass help.

User avatar
snacksthecat
✶✶✶✶

07 Jun 2019, 21:39

SneakyRobb wrote:
07 Jun 2019, 19:26
Hi, that is most unfortunate. It sort of looks like the traces above your last shot corroded away? I would imagine if that is so they would not return signal.

It would cost you perhaps $50 +import/etc but you could always make a new PCB for it. I have had success making new beamspring PCBs with DMA and Wcass help.
I'm hoping it's still salvagable in some way. Either fixing the connector or using different firmware. In xwhatsit, it seemed like all of the rows of keys worked, but at different voltage thresholds. As far as I know, you can't set a different threshold for each key with that firmware. I know in CommonSense you can do that. I will probably explore that route before looking into designing the PCB. Though, are there any footprint libraries you know of?

User avatar
SneakyRobb
THINK

07 Jun 2019, 22:03

snacksthecat wrote:
07 Jun 2019, 21:39
SneakyRobb wrote:
07 Jun 2019, 19:26
Hi, that is most unfortunate. It sort of looks like the traces above your last shot corroded away? I would imagine if that is so they would not return signal.

It would cost you perhaps $50 +import/etc but you could always make a new PCB for it. I have had success making new beamspring PCBs with DMA and Wcass help.
I'm hoping it's still salvagable in some way. Either fixing the connector or using different firmware. In xwhatsit, it seemed like all of the rows of keys worked, but at different voltage thresholds. As far as I know, you can't set a different threshold for each key with that firmware. I know in CommonSense you can do that. I will probably explore that route before looking into designing the PCB. Though, are there any footprint libraries you know of?
Hi, There is a series of videos from user "red'

viewtopic.php?f=7&t=20375

https://youtu.be/CNPMcrKGtd4
With his repository in the description
github.com/redvers/modelf

User avatar
adamcobabe

08 Jun 2019, 16:14

I had an issue with the 5251 where the Xwhatsit would slide side to side and contact the wrong trace pads, though that was more of a columns issue. Long shot, but I am wondering if maybe the plating is affecting the capacitance as nickel is ferromagnetic. I'm not an expert in physics though so, maybe just ignore me. ;) Nickel is used in electronics all the time, right? You could also just test without the foam to make sure the extra thickness isn't pushing differently for certain rows.

User avatar
snacksthecat
✶✶✶✶

10 Jun 2019, 03:09

Holy cats. It's the nickel plating.

I have a second 5251 that I finally brought out of storage. I swapped different parts in and outbetween the two boards. For instance, I tried the "new" pcb with the old plate/switches. It kind of worked, but the bottom row was dead and the voltage threshold didn't make sense. So then I tried the new keyboard on it's own, without swapping any parts between the two. It worked perfectly I'm 99% sure this is the problem.

User avatar
snacksthecat
✶✶✶✶

13 Jul 2019, 23:48

Okay I may have jumped the gun with that last conclusion. The jury is still out but I suspect it may be a problem.

Anyways, I'm moving forward using the other keyboard as a stand-in.

The edge connector on the PCB had a bunch of black spots. Some sort of corrosion/oxidation (?). I removed this with rubbing alcohol and a some gentle scraping with the probe from my multimeter (it actually worked really well)

ImageImage

But there was now a lot of exposed copper. I'm not sure the function of the silver colored metal that covers the copper. Is this maybe the cause of my issue?

So I'd gotten frustrated trying to get the xwhatsit working with this board and gave up. I wanted to try with CommonSense and see if I could get better results. I didn't have the connector piece so I soldered directly to the PCB.

ImageImage

Unfortunately it does not work at all. I even went so far as to swap in a different cypress board, thinking that might be the problem. When I load up Flight Controller, it always flags as insane = true. I tried reducing the number of pins to zero in on the problem. If I remember correctly, it worked with 1 row and 1 column. Maybe I was able to take it a little further than that.

Any thoughts on what the issue with this PCB might be? Is it savable?

kmnov2017

14 Jul 2019, 23:58

snacksthecat wrote:
13 Jul 2019, 23:48
But there was now a lot of exposed copper. I'm not sure the function of the silver colored metal that covers the copper. Is this maybe the cause of my issue?
As long you are getting a positive continuity test with a multi-meter, the exposed copper isn't a problem.

kmnov2017

15 Jul 2019, 00:43

SneakyRobb wrote:
07 Jun 2019, 19:26

It would cost you perhaps $50 +import/etc but you could always make a new PCB for it. I have had success making new beamspring PCBs with DMA and Wcass help.
Do you have kicad files for the 3101 PCB? I am assuming the PCB for 3101 is the same for 3278 and 3727.

User avatar
PlacaFromHell

15 Jul 2019, 01:32

snacksthecat wrote:
13 Jul 2019, 23:48
But there was now a lot of exposed copper. I'm not sure the function of the silver colored metal that covers the copper. Is this maybe the cause of my issue?
In think the silver plating is used to ensure a good continuity between the PCB and the controller, no more than that. Something like the javelin of your home ground cable, if it were entirely copper you would have to pay a fortune for it, so they are made of iron galvanized in copper. Shouldn't be a big problem.

User avatar
SneakyRobb
THINK

15 Jul 2019, 02:32

kmnov2017 wrote:
15 Jul 2019, 00:43
SneakyRobb wrote:
07 Jun 2019, 19:26

It would cost you perhaps $50 +import/etc but you could always make a new PCB for it. I have had success making new beamspring PCBs with DMA and Wcass help.
Do you have kicad files for the 3101 PCB? I am assuming the PCB for 3101 is the same for 3278 and 3727.
Hi I dont have any files for these pcbs. I dont know if they are the same or not versus 3278 3727 etc.

Regardless they would not be too hard to create. I have made multiple beamspring pcbs at this point using ideas gathered from other forum users. Multiple forum members would be able to create these files for you as well.

PancakeMSTR

16 Jul 2019, 18:38

How do you know how to wire up the commonsense controller?

User avatar
SneakyRobb
THINK

16 Jul 2019, 18:56

Spoiler:
snacksthecat wrote:
13 Jul 2019, 23:48
Okay I may have jumped the gun with that last conclusion. The jury is still out but I suspect it may be a problem.

Anyways, I'm moving forward using the other keyboard as a stand-in.

The edge connector on the PCB had a bunch of black spots. Some sort of corrosion/oxidation (?). I removed this with rubbing alcohol and a some gentle scraping with the probe from my multimeter (it actually worked really well)

ImageImage

But there was now a lot of exposed copper. I'm not sure the function of the silver colored metal that covers the copper. Is this maybe the cause of my issue?

So I'd gotten frustrated trying to get the xwhatsit working with this board and gave up. I wanted to try with CommonSense and see if I could get better results. I didn't have the connector piece so I soldered directly to the PCB.

ImageImage

Unfortunately it does not work at all. I even went so far as to swap in a different cypress board, thinking that might be the problem. When I load up Flight Controller, it always flags as insane = true. I tried reducing the number of pins to zero in on the problem. If I remember correctly, it worked with 1 row and 1 column. Maybe I was able to take it a little further than that.

Any thoughts on what the issue with this PCB might be? Is it savable?

Hi it still looks to me like the traces are physically corroded/destroyed in multiple places breaking the connectivity. Those dark areas along the lengths. Do you have any close up photos of those areas?

I don't know if you are brave enough to slice a through the mask to probe test but from your photos it looks like almost every column has some areas where the trace is destroyed. If you are brave I would try that. If you find that the dark areas do indeed have no connectivity then the board might be toast.

Like this looks to me to be totally broken
Spoiler:
WAuTzPM.jpg
WAuTzPM.jpg (249.64 KiB) Viewed 999 times

User avatar
snacksthecat
✶✶✶✶

16 Jul 2019, 19:00

Thanks Robb. I'm going to check into that.

Also sorry I haven't gotten back to you yet.

User avatar
snacksthecat
✶✶✶✶

16 Jul 2019, 19:02

PancakeMSTR wrote:
16 Jul 2019, 18:38
How do you know how to wire up the commonsense controller?
It's described in the github repo but it's a bit confusing. I tried putting together a simpler guide.

viewtopic.php?f=7&t=21899

PancakeMSTR

16 Jul 2019, 19:13

snacksthecat wrote:
16 Jul 2019, 19:02
PancakeMSTR wrote:
16 Jul 2019, 18:38
How do you know how to wire up the commonsense controller?
It's described in the github repo but it's a bit confusing. I tried putting together a simpler guide.

viewtopic.php?f=7&t=21899
Thank you I will take a look at this.

User avatar
snacksthecat
✶✶✶✶

25 Aug 2019, 01:24

Well this is another project that I've really let go. I got back around to trying out the keyboard with CommonSense this time, since I figured it would give me more control than xwhatsit. I ran into so many issues. Believe it or not, I rebuilt the converter from scratch 4 times before I finally convinced myself to try something else.

Image

Whenever plugged into the keyboard into the PC, FlightController would complain that it's insane and stop scanning.

Finally I decided to ask DMA what the deal was and he pointed me to a section in the readme that has been there all along. This is what it says:

If something is wrong - you can still use matrix monitor using the following trick:
  • open matrix monitor,
  • make sure the mode is "Now"
  • click "Start!"
  • go to main window
  • click "Scan"

So essentially, this enables you to start the matrix monitor and see which rows/columns are having the problem. In my case, there was one row where all the keys were showing as pressed.

Now I swear I went through and tested all the traces before, but somehow I missed this. This time when I went back and retested, the multimeter confirmed what I was reading on the screen.

So I followed the trace from its beginning to its end looking for the problem spot. This was challenging since there are so many spots where it looks like the trace is cut/corroded but it's really fine. I did eventually locate the issue; a spot on the PCB where I don't know exactly what happened.

Image

I did my best to repair this little cut using a thin strand of wire.

Image

I covered this repair with a small piece of electrical tape.

Now when I plug it into flight controller, I have no issues.

But it does beg the question: what happened to that trace? It didn't look like it had been slashed or cut. It looked to me like it wore away for some reason. There are so many spots on the PCB that look identical. Are those eventually going to wear away as well? Is my poor keyboard on borrowed time?

User avatar
Polecat

25 Aug 2019, 02:06

snacksthecat wrote:
25 Aug 2019, 01:24

Now I swear I went through and tested all the traces before, but somehow I missed this. This time when I went back and retested, the multimeter confirmed what I was reading on the screen.

So I followed the trace from its beginning to its end looking for the problem spot. This was challenging since there are so many spots where it looks like the trace is cut/corroded but it's really fine. I did eventually locate the issue; a spot on the PCB where I don't know exactly what happened.

I did my best to repair this little cut using a thin strand of wire.

I covered this repair with a small piece of electrical tape.

Now when I plug it into flight controller, I have no issues.

But it does beg the question: what happened to that trace? It didn't look like it had been slashed or cut. It looked to me like it wore away for some reason. There are so many spots on the PCB that look identical. Are those eventually going to wear away as well? Is my poor keyboard on borrowed time?
The green corrosion tells the story. The mask on the copper at that spot got damaged or was applied too lightly. Bare copper is only slightly less prone to corrosion than bare steel, so board manufacturers do one of two things. They either tin (or gold $$$) plate the copper, or they apply a thin coat of mask or paint. Your repair will outlast the rest of the board, but any other corroded spots will have the same problem unless you clean and protect them. Some 600 grit sandpaper or a fine wire brush can be used to clean, and some cheap paint or nail polish to coat once cleaned. Check each trace with a meter after cleaning to be sure you didn't damage the remaining copper.

Post Reply

Return to “Workshop”