Let's play Where's Wally! Gutz's Diode Hunt in progress!

[berserkfan]

Hey Folks

As some of you may have realized, I want to turn a keyboard into a JIS to experiment with a different layout.

Looking carefully at the PCB of my CM QFR, I notice there are a few missing diodes.

Pictures are poor but I'll retake them again if anybody wants to join in the hunt. I notice specifically:

1) some diodes related to JIS keys are soldered in and some are missing. IE this is a rather inconsistent manufacturing process.

2) Some diodes relating to ANSI and JIS keys are missing.

For instance,

ONE
Ctrl-L, K14. It is missing a diode. I suspect the two holes on the bottom left of the PCB were meant for the diode until the manufacturers realized the keyboard can't be placed into its casing with the diode in place.

TWO
Alt-R, K36. It seems to have been mismarked as Alt-R when it is actually Win-R which is not marked on this PCB.

THREE
The real Alt-R is marked J133 like a JIS key, and it lacks a diode. I have been able to use this keyboard pressing Alt-R and two other keys though. (To run some macros.)

FOUR
Backspace is labelled K78 on the PCB and the corresponding diode, D78, has been soldered in. But the actual switch is located at K101 where the switch solder took place. So is there a diode for backspace or not? Was backspace operating without a diode and won't function with NKRO? Except that a user wouldn't have known, because you normally don't press other keys and backspace at the same time?

I also have a question. If I want to 'activate' the JIS keys reliably, do I need a diode in each of the holes associated with the JIS key? Using the 'bridging the PCB contacts method' that some geekhackers have advised, I have found that the JIS keys do register under autohotkey even when the diode associated with them is missing. But if I want to use them to run macros with more than two keypresses (eg JISkeyx-A-B) do I need a diode?

To be safer rather than sorry (don't want to assemble an entire keyboard and find that some keys don't work) can I just put one diode in each available hole, regardless of whether I want to use that key? I think it's better than hunting for the correct diode slots because the markings don't always seem reliable. But I am also concerned that may lead to problems not foreseen because I lack the knowledge.

I am asking all these questions because I'm tryin to figure out how diodes work and how they fit into the scheme of things. Thanks for your patience!

[Oobly]

Don't just stuff diodes in.

You don't show us the other side of the board, so we can't trace the contacts. You need to trace the contacts from the switch pads to the controller to work out the matrix. If the keys already activate a keycode in AHK, then they are making contact and adding diodes to empty spots won't help (and may just interfere with the functioning of other keys). If you trace them and find there is no diode in the circuit for a particular switch, then you may need to replace a wire jumper with a diode that is oriented correctly. Use the existing switches and diodes on that row or column (depending on whether the diodes are attached to the row contact or column contact) as a reference for which way to install the diode.

[berserkfan]

Don't just stuff diodes in.

You don't show us the other side of the board, so we can't trace the contacts. You need to trace the contacts from the switch pads to the controller to work out the matrix. If the keys already activate a keycode in AHK, then they are making contact and adding diodes to empty spots won't help (and may just interfere with the functioning of other keys). If you trace them and find there is no diode in the circuit for a particular switch, then you may need to replace a wire jumper with a diode that is oriented correctly. Use the existing switches and diodes on that row or column (depending on whether the diodes are attached to the row contact or column contact) as a reference for which way to install the diode.

OK, Oobly, your first line was the most important line in this entire thread.

That is THE most important thing for a noob like me.

Now I gotta digest what you just said...

[domoaligato]

plug the board in with no switches and short the pads for the switches you want to find if they work with tweezers. ahk or aqua key test should tell you want key it is.

[berserkfan]

plug the board in with no switches and short the pads for the switches you want to find if they work with tweezers. ahk or aqua key test should tell you want key it is.

Yup, I did that, and was able to identify that all the JIS keys actually work. But I don't know how to reliably do that shorting thing. Sometimes I need to reposition the tweezers many times before an actuation is recorded.

Ok, installing diodes the right way on CM PCB is not an issue. The direction is indicated.

I was more concerned with what may happen if there is no diode for a switch that can be actuated. No one has answered that yet. If I can avoid installing diodes, I WILL avoid installing diodes! Why do I want to spend hours trying to trace a pcb? It will drive me nuts.

[domoaligato]

plug the board in with no switches and short the pads for the switches you want to find if they work with tweezers. ahk or aqua key test should tell you want key it is.

Yup, I did that, and was able to identify that all the JIS keys actually work. But I don't know how to reliably do that shorting thing. Sometimes I need to reposition the tweezers many times before an actuation is recorded.

Ok, installing diodes the right way on CM PCB is not an issue. The direction is indicated.

I was more concerned with what may happen if there is no diode for a switch that can be actuated. No one has answered that yet. If I can avoid installing diodes, I WILL avoid installing diodes! Why do I want to spend hours trying to trace a pcb? It will drive me nuts.

if it works already with the tweezers then you do not need to install any diodes. just solder a switches in...

[berserkfan]

if it works already with the tweezers then you do not need to install any diodes. just solder a switches in...

So what about nkro? Aren't diodes for nkro?

I thought 2kro means only 2 keys can register at any time.

Then if I have a macro activated by a combination of 3 keys, that won't work with 2kro. The only way to make that work would be to install diodes?

Sorry for sounding so stupid...

[IvanIvanovich]

For mod keys it doesn't even remotely matter. KRO on those isn't really included. As for example on a 2KRO board you can press Z U + alt + ctrl and all of them register. Or same goes for a 6KRO where you could press 6 alpha numerics and X mods and they all register. The biggest problem with knowing if alternate switch layouts work is in the controller firmware which you already confirmed is a universal one so you should just go for it.

[domoaligato]

doesn't a qfr support nkro already?

[berserkfan]

For mod keys it doesn't even remotely matter. KRO on those isn't really included. As for example on a 2KRO board you can press Z U + alt + ctrl and all of them register. Or same goes for a 6KRO where you could press 6 alpha numerics and X mods and they all register. The biggest problem with knowing if alternate switch layouts work is in the controller firmware which you already confirmed is a universal one so you should just go for it.

Ivan that sounds awesome. Does it mean that I can press two alphanumerics and two mods, but CANNOT press one mod and three alphas?

EG I was hoping to configure an autohotkey script to run alt-pw1 as my password one, alt-pw2 as password two, etc.

Basically some government websites here are very infuriating. They force you to change passwords frequently. And they even keep records of old passwords, and refuse to let you use an old password. And of course each password must have many digits, upper and lower case, symbols, etc making it practically impossible to remember so I want to just use autohotkey to enter that password for me.

Also, what do you mean by modifiers? Are ctrl and alt the only modifiers for the purpose of having NKRO exception, or does shift, windows/GUI key count?

I don't fully understand how modifiers relate to the rest of the keys or how windows treats them so I am always worried that some autohotkey scripting I use will mess up windows, especially when it comes to ctrl-alt-something.

[IvanIvanovich]

Mods are technically any non alpha numeric really, well at least as it applies to the main typing block. So tab, caps, shift, ctrl, win, alt, menu, enter, backspace... all exempt to KRO. Or at least ways that is what I can see while testing in aquas. None of those keys should really absolutely require a diode since they get treated as a #KRO +1 anyway.

Hehehe. As far as passwords go, I always use the old leetspeak tricks, for example Ba$t4Rds! usually would meet all requirements and is easy to remember.

[berserkfan]

Mods are technically any non alpha numeric really, well at least as it applies to the main typing block. So tab, caps, shift, ctrl, win, alt, menu, enter, backspace... all exempt to KRO. Or at least ways that is what I can see while testing in aquas. None of those keys should really absolutely require a diode since they get treated as a #KRO +1 anyway.

Hehehe. As far as passwords go, I always use the old leetspeak tricks, for example Ba$t4Rds! usually would meet all requirements and is easy to remember.

HAHA
 
I used to have a password for all Singapore government websites called F@#$p4p (p4p being a reference to the bunch of scum who rule this country). Unfortunately as you can see, I was forced to change passwords after they increased the minimum number of letters in the passwords from 6 to 8.

But Ivan, what if I press some keys with no diodes + other keys with diodes? Does it mess up the computer?

I am thinking in terms of the JIS keys I'm trying to activate. I obviously don't know what Japanese consider a modifier. They seem to have many modifiers, if by that you mean keys that don't immediately send individual characters/ letters but instead instruct the computer to change something.

But Ivan, from your greater knowledge and experience, is it a problem in computers generally  if I map various mods in autohotkey that actually have preprogrammed system functions in windows that I do not know of? EG lets say Ctrl-Tab-1 does something in windows. I don't know that, so I use autohotkey to map Ctrl-Tab-1 to do something else. Will that have unexpected effects when under certain conditions autohotkey works, and under other conditions, the original Ctrl-Tab-1 function takes over and does something I wasn't expecting? I already have some confusion when sometimes windows key works and sometimes it doesn't, and I've never figured out the logic.

[dorkvader]

For mod keys it doesn't even remotely matter. KRO on those isn't really included. As for example on a 2KRO board you can press Z U + alt + ctrl and all of them register. Or same goes for a 6KRO where you could press 6 alpha numerics and X mods and they all register. The biggest problem with knowing if alternate switch layouts work is in the controller firmware which you already confirmed is a universal one so you should just go for it.

This is not always correct. Especially for a keyboard with 2kro and a very compact matrix, you very much run into blocking issues even with the modifiers. I have seen at least one vintage keyboard get around this by putting all the function keys on their own row AND column.

I say you solder switches up and test. What if "the worst" happens and you don't have NKRO on some of your needed keys? Well since you have photos of the "hard to access" side of the PCB, you can actually pre-clip a diode's leads short and solder it to the back side. It's a little harder, but still quite manageable.

as I read a bit more....

Mods are technically any non alpha numeric really, well at least as it applies to the main typing block. So tab, caps, shift, ctrl, win, alt, menu, enter, backspace... all exempt to KRO. Or at least ways that is what I can see while testing in aquas. None of those keys should really absolutely require a diode since they get treated as a #KRO +1 anyway.

Now that is absolutely not true. "menu" is not a modifier and is not treated as such. If you press 6 keys, and then press menu after, one of those 6 keys is getting unpressed (or menu will not actuate, depending on the behaviour of the KB controller)

I can do further testing with specific keyboards if you want proof, but I absolutely know that the G84-41** that I have has issues with "enter" and the function keys. Most modern keyboards are routed to avoid this, but on a 2kro keyboard, if you have any key on the same row as one modifier and the same column as another, it cannot be pressed in conjunction with those two modifiers.

[Oobly]

All keys switches in the matrix SHOULD have diodes if you want to retain full NKRO. Some keys may be wired directly to the controller by both pins and then they don't need diodes, but this is inefficient and controllers usually don't have enough free pins to do this. So you may find one or two switches wired like this, but usually they're all connected to the matrix lines with diodes. In order to work this out we need to trace the lines from those switches you suspect of not having diodes.

It would be less effort to trace those keys you think don't have diodes than what some of us have spent in answering your questions... It's not too hard. You just follow the copper tracks. If you post pics of the underside of the board we can do it for you and nobody has to speculate about it any more.

[dorkvader]

often the "program mode" key on a custom keyboard will be wired separately. For the KMAC it's capslock, for lightsaver it's the right key above backspace. On both these keyboards this key has no diode (nor does it need one)

about the underside of the board: it's a QFR. I don't have any good pictures of mine, but here's one I found:
http://imagizer.imageshack.us/a/img577/1105/dsc03031y.jpg

Note that this one has the "newer" internal revision, and has the cable that can unplug from the USB subboard. On mine (first / earliest revision / left alt issue) the cable is fixed to the usb subboard and there may-or-may-not be other differences.

[berserkfan]

All keys switches in the matrix SHOULD have diodes if you want to retain full NKRO. Some keys may be wired directly to the controller by both pins and then they don't need diodes, but this is inefficient and controllers usually don't have enough free pins to do this. So you may find one or two switches wired like this, but usually they're all connected to the matrix lines with diodes. In order to work this out we need to trace the lines from those switches you suspect of not having diodes.

It would be less effort to trace those keys you think don't have diodes than what some of us have spent in answering your questions... It's not too hard. You just follow the copper tracks. If you post pics of the underside of the board we can do it for you and nobody has to speculate about it any more.

Actually the PCB is no longer available because a friend wanted me to make a CM QFR TKL ANSI for him. So I'm just asking questions to build up my understanding of how these things work. I agree that some traces are easy to follow, but some I totally don't understand - eg on my Tipro they seem to all coalesce in a different part of the PCB.

I am, however, taking apart a Filco for the purpose of doing this JIS mod. I intend to sacrifice one Filco plate to my hacksaw just to get a few JIS keys inside for the sake of a modding experience. I will take photos and we can all join in that hunt when I'm done with the desoldering. Am not looking to build a perfect keyboard; am basically trying to build up understanding and experience.

[Oobly]

All keys switches in the matrix SHOULD have diodes if you want to retain full NKRO. Some keys may be wired directly to the controller by both pins and then they don't need diodes, but this is inefficient and controllers usually don't have enough free pins to do this. So you may find one or two switches wired like this, but usually they're all connected to the matrix lines with diodes. In order to work this out we need to trace the lines from those switches you suspect of not having diodes.

It would be less effort to trace those keys you think don't have diodes than what some of us have spent in answering your questions... It's not too hard. You just follow the copper tracks. If you post pics of the underside of the board we can do it for you and nobody has to speculate about it any more.

Actually the PCB is no longer available because a friend wanted me to make a CM QFR TKL ANSI for him. So I'm just asking questions to build up my understanding of how these things work. I agree that some traces are easy to follow, but some I totally don't understand - eg on my Tipro they seem to all coalesce in a different part of the PCB.

I am, however, taking apart a Filco for the purpose of doing this JIS mod. I intend to sacrifice one Filco plate to my hacksaw just to get a few JIS keys inside for the sake of a modding experience. I will take photos and we can all join in that hunt when I'm done with the desoldering. Am not looking to build a perfect keyboard; am basically trying to build up understanding and experience.

Sounds like a good and worthwhile project. Tinkering is a great way to learn. Dual sided PCB's can be a bit of a PITA to trace and sometimes you won't know you're actually dealing with a 4 layer PCB... you simply can't see the layers in the middle, so tracing those has to be done with tools. A single sided like the QFR board is easy enough, though, and fortunately I have yet to see a keyboard with more than dual sided PCB's.