Troubleshooting at porting keyberon to a keyboard about keyberon HOT 29 CLOSED

You can just swap the 2 coordinates before sending it to the layout.

from keyberon.

In which file I need to do it?

from keyberon.

I don't know your firmware, but something like that: https://github.com/TeXitoi/keyseebee/blob/4d29baf283a676035740669f2bfbe211ac492998/firmware/src/main.rs#L271

Transform would be something like |e| e.transform(|i, j| (j, i))

from keyberon.

It is the base of your ortho60-keyberon project.

I need to define my Cols like

pub struct Cols(
    pub PA0<Output<PushPull>>,
    pub PA1<Output<PushPull>>,
    pub PA2<Output<PushPull>>,
    pub PA3<Output<PushPull>>,
    pub PA4<Output<PushPull>>,
    pub PA5<Output<PushPull>>,
    pub PA6<Output<PushPull>>,
);

But I got this error:

let matrix = Matrix::new(
^^^^^^^^^^^ expected trait `embedded_hal::digital::v2::OutputPin`, found trait `embedded_hal::digital::v2::InputPin`

This error message looks for me that I olny can define Cols as InputPins and Rows as OutputPins

Sorry I'm a rust newbie.

from keyberon.

So if I switch my Cols and Rows I can compile the firmware. Also I transform the coordinates with your suggested function that.
But now I don't get any key output.
Do you have an idea how I can debug the software?

from keyberon.

That's the trick yeah.

Maybe diodes are in the wrong way? Try shorting some input and output without diode. Also with a debugger you can use gdb do go step by step.

from keyberon.

Ah well, I found the Problem. My hardware works with a PullDown instead of a PullUp for the Inputs. Maybe it is a good idea to specify which is the Input or Output and if the Input works with a PullUp or PullDown while create the matrix? I can try do to this and create a Pull Request. But this will take some time.

BTW: Thank you for your very fast support :)

from keyberon.

In this case, the default row is output should work.

from keyberon.

If I use the row as output I got sometime more than one keycode output.

Pin1 of the Cherry-MX key is connected to a Col.
Pin2 ot the Cherry-MX key is connected to the row with a diode.

So I think I can't use the rows as output. (I'm not a hardware expert at all)

from keyberon.

That's the standard row output active high, so that's implemented by default.

Don't you have any short or any diode in the wrong direction? If you have several keypress while pushing only one button, I'll check for shorts looking at the keys that activate.

from keyberon.

What's tour PCB? Does it works flawlessly with the default firmware? If that's some normal keyboard, maybe some keys activates several intersections, and that's by design, do you have the complete schema?

from keyberon.

It is a custom PCB. The matrix have 10 rows and 7 cols.
This here is the left side of the keyboard. I need here the output on the cols (at top in this image) and the input to the rows (left in this image).

All diodes are in the same direction. I checked this after soldering.

EDIT:
The cols are connected with a resistor to the controller.
The rows are connected directly to the controller.

from keyberon.

Are the resistor in series or is it a pull up (to 3.3v) or a pull down (to ground)?

from keyberon.

In series.

from keyberon.

Should not cause any problem.

from keyberon.

I'd check the keys that double register and try to find a pattern related to the col and row number. You might have a short between 2 cols, that would explain the double register.

from keyberon.

I will clarify this and get back to you. Thanks for your help.

from keyberon.

You were right, but there does not seem to be a short. Because when I swap the affected row with another, the problem shifts to that row.
Could it be because I don't have my rows from 1-10 as in the hardware, but 1,6,2,7,3,8,4,9,5,10? This is because the left side of the keyboard is row 1-5 and the right side is 6-10.
I have now adjusted this so that I have line 1-10 and it works fine.

If you are not interested in the feature, the issue can be closed.

from keyberon.

So, that's 2 keys from same colum, successive rows? or 2 rows whatever the order, always the same pins?

from keyberon.

I change only the pins on the controller side. The rows and cols on the matrix are fixed, because it is a PCB.
If I press my Col1(Pa1) Row3 switch and Col2(Pa2) Row4 the Col1(Pa1) Row 4 is recognized.
I change now the Pins of the Controller from
Pin on Controller => Pin on Matrix

• Pa1 => Col1
• Pa2 => Col2
• Pa3 => Col3
• Pa4 => Col4
• Pa5 => Col5
• Pa6 => Col6
• Pa7 => Col7
  to
  Pin on Controller => Pin on Matrix
• Pa3 => Col1
• Pa4 => Col2
• Pa5 => Col3
• Pa6 => Col4
• Pa7 => Col5
• Pa1 => Col6
• Pa2 => Col7

Then the double key press is recognized at Col 6 and 7. I also switch the pins to PB0 and PB1 and it is the same. Switching the rows has also the same effect.

I have found a workaround to "fix" it.
I activate with (Col1Row3) my layer(2) and on this layer I set Col1Row4 to NoOp.

from keyberon.

Do you use PB2 with a WeAct MiniF4? https://github.com/WeActTC/MiniSTM32F4x1

PB2 has a pull up on the board, and it can mess things. Better to not use this pin.

In your message, you said that Col1 and Col3 use Pa1, I suppose this is a typo. If the 2 columns use the same pin, it can't work properly.

from keyberon.

Yes it is a typo :D
I use the stm32f103c8t6 Bluepill

from keyberon.

It really looks like a short, a diode in the wrong direction or a faulty diode. The fact that it's more corrected to the board pin than to the button make me think of something on the board. Can you isolate the 2 cols and 2 rows that case the problem? do you reproduce the problem by emulating the button by touching the pins with 2 diodes in the good direction? can you give me the pin, I might find the time to test here on a bluepill.

from keyberon.

I will test it again with your tips and get back to you. Also I will try another bluepill.

from keyberon.

These are my pins for the matrix:

pub struct Cols(
    pub PA0<Input<PullUp>>,
    pub PA1<Input<PullUp>>,
    pub PA2<Input<PullUp>>,
    pub PA3<Input<PullUp>>,
    pub PA4<Input<PullUp>>,
    pub PA5<Input<PullUp>>,
    pub PA6<Input<PullUp>>,
);

pub struct Rows(
    pub PA9<Output<PushPull>>, // 1
    pub PA10<Output<PushPull>>, // 2
    pub PA15<Output<PushPull>>, // 3
    pub PB3<Output<PushPull>>, // 4
    pub PB4<Output<PushPull>>, // 5
    pub PB5<Output<PushPull>>, // 6
    pub PB6<Output<PushPull>>, // 7
    pub PB7<Output<PushPull>>, // 8
    pub PB8<Output<PushPull>>, // 9
    pub PB9<Output<PushPull>>, // 10
);

from keyberon.

PA15, PB3 and PB4? I suppose you deactivated the debugger functionalities on those pins? Maybe these is some bugs around pullups that are wrongly activated with the remapping? You may try to transform them in input pullup, input floating and finally in output push pull to be sure the configuration is correct.

from keyberon.

Any updates?

from keyberon.

Not really.
I add a dummy col that I have a workaround to use my keyboard.
The current situation also does not allow me to meet with my friends to examine the problem more closely.

from keyberon.

If you want you can take a look at my layout.rs and main.rs

from keyberon.