Can we design the teensy alternative for keyboards?

[hasu]

Have you considered making one as small as possible for converters specifically?

Give it the minimum amount of pins and keep it small, ideally it would be cheaper as well.

This one already has almost the minimum components. The reset switch could perhaps be removed, as well as the led. The price would stay almost the same however, as the pcb area is ~free when compared to the p'n'p and chip + usb socket.

As for converter, we need voltage lever shifter because the mcu doesn't have any 5V-tolerant GPIO unfortunately.

Right, reset button can be otmitted, you can plugin with pressing BL(NMI/BOOTCFG0) button to start bootloader. I didn't see this from datasheet and I thought we had to have reset button. I may omit reset button on my next revision of breakout board. To have reset button is still useful when developping or debugging it, though.

[hasu]

I'm trying to follow 60% hand wiring keyboard scenario. Not finish yet but I found something on dimesion of connector.

The conector seems to be shifted a bit. I'm using Poker X(original one) case and PCB for reference.



Sticking out the connector a little more is preferable if possible. I think front legs of the connector may prevent it.

[hasu]

But is somehow cleaning up the d+ pin with a resistor to nowhere.. anyway still a mistery..! You see what I mean..

It has probably more to do with the USB controller on host side. On my PC, some usb 3 ports work flawlessly, while usb 2 do not.

Interesting. I don't have any USB3 port in hand yet.

pomk, this is my observation on false detection issue:
Due to errata e9457, need to pull-up `SPI0_SS_b`(or `SPI0_SCK`) to prevent false detection of SPI.
The errata doesn't refer UART though, `LPUART0_RX` also requires pull-up or down resisitor from experience.
Without resistor it easily false-detects activity on UART interface with finger touch on the pin.

Check the pins on your evironment. Those are numbered as 10 and 28 on ELF. I could cause the issue by touching the pins with finger or short wire.

[matt3o]

The conector seems to be shifted a bit. I'm using Poker X(original one) case and PCB for reference.

this is interesting and I think it should be addressed, after all poker cases are still the easiest to source

[pomk]

To be honest, I have never even seen a poker case.
The connector in the initial prototype is just aestethically placed in the middle.
There is room to move it by 1mm. Could you give me the specific amount it needs to be moved?
Also, be aware that if soldered on top of another pcb, the connector will drop by the pcb thickness of the motherboard.

[Nuum]

Can you use a Poker case with a handwired PCB? As far as I know the screws in those bolt the PCB to the case, not the plate.

[Findecanor]

You could if you used a custom-made plate and standoffs. Regular Poker plates have too large holes over the screw holes, sized to let a screwdriver through and it would not be feasible to fill those holes in.
Small cylinders that could be used for standoffs are available in stores that specialize in a good variety of nuts, bolts and fasteners.

As for converter, we need voltage lever shifter because the mcu doesn't have any 5V-tolerant GPIO unfortunately.

If you are going to revise the design, then I think that you should consider putting a voltage regulator on it. The internal regulator can drive only up to 120mA.

[matt3o]

I made some tests with various keyboard cases I have. Weirdly enough on the poker the USB in not exactly in the middle between the two switches. The whitefox for example has the usb slot in the perfect spot for the ELF board to work flawlessly.

This is a tough call. Should we make it compatible with a poker case? I don't think it is strictly necessary considering that you can re-route the USB port pretty easily, but I accept suggestions here. As others have said, the poker case is not exactly hand-wire friendly but it is also the easiest to find case around.

The internal regulator can drive only up to 120mA.

I don't have a strong feeling about this, what do you guys think?

[vvp]

I'm afraid that pomk will tell you that an external regulator which can handle 500 mA will not fit on the PCB (without enlarging the PCB)

[vvp]

bad post

[pomk]

I'm amazed of how much space is left at the moment, however I'd like to know the use case where more power output on the 3,3V line is needed, and where the board is used in a hand wired build. If the board is mounted on a motherboard, why not have the regulator there? The 5V line is available on one pin.

[Findecanor]

I was thinking of LEDs for backlighting, since the PCB has holes for in-switch LEDs maybe that was something that you had in mind. But there may not be enough pins to drive both a LED matrix and a keyboard matrix anyway. People might want to use LED strips for accent lighting though.

[pomk]

Many led strips can be driven with the 5V line. For example the adafruit ones are quite popular as far as I know.
https://www.adafruit.com/products/306

Besides, if someone is willing to solder 120+ components for lightning, then maybe soldering a trough hole regulator as well is not too much to ask for. I have yet to see a hand wired board with in switch lightning.

[hasu]

This is a tough call. Should we make it compatible with a poker case? I don't think it is strictly necessary considering that you can re-route the USB port pretty easily, but I accept suggestions here. As others have said, the poker case is not exactly hand-wire friendly but it is also the easiest to find case around.

I can understand ditching poker case support, we have already variety of 60% PCBs in the market and don't have to hand-wrire in many cases.

[matt3o]

I have yet to see a hand wired board with in switch lightning.

a guy in this very forum adapted hasu's firmware for backligting on a hand wired backlit keyboard. I can't find the thread right now, but it's there

that being said, why do we need more power on the 3v? Can't we use the 5v from the USB for backlight? I guess the 3v might be useful to connect a bluetooth module, I haven't checked but I guess a BLE module draws 20-25mA tops.

I can understand ditching poker case support, we have already variety of 60% PCBs in the market and don't have to hand-wrire in many cases.

okay then, let's keep it where it is now. We'll make a first batch and see what people think about it.

[pomk]

BLE is not a problem for power consumption. Besides, I'd rather make a separate bluetooth elf board, so that we could have battery life measured in years, rather than in hours as with adafruit adapters and non-optimized power designs.

And yes, the leds can be driven from the 5V. I'm not sure if it would be possible to somehow toggle them as well though. Maybe if the forward voltage is enough they could be connected to one pin on the elf, and toggle the pin from input low to output low. Effectively adding/removing 10k resistor from the circuit. You'd have to be careful though not to let the 5V run back to the mcu.

[matt3o]

Besides, I'd rather make a separate bluetooth elf board, so that we could have battery life measured in years, rather than in hours as with adafruit adapters and non-optimized power designs.

that I would really like to see happen!

[pomk]

I'm currently tinkering around multiple device support and wireless firmware updates. After these are done, all that is left is to port tmk functionality to it. For the last part any help would be appreciated. I'd like to just re use the way layers are handled, ie. have a function to which matrix state is given and which returns the keycodes to put to the hid report. The hid report is the same for bt and usb.
My current prototype yielded about six months from two aaa batteries. The low value was because mx swithces dont seem to like voltages below 2. With a new design using three batteries and a dcdc converter, the batterylife should reach at least one year.

[Ray]

I'm not sure if it would be possible to somehow toggle them as well though. Maybe if the forward voltage is enough they could be connected to one pin on the elf, and toggle the pin from input low to output low. Effectively adding/removing 10k resistor from the circuit. You'd have to be careful though not to let the 5V run back to the mcu.

I don't know for this one, but most MCUs can only sind current in the same order of magnitude as they can source. There may be more margin, but don't expect too much (or read the datasheet )

[tentator]

I also would like a BT extension and about leds supporr I think more than the 3 locks and an accent light below the keyboard I wouldn't do more.. not interested in Christmas tree keyboards

[pomk]

I'm not sure if it would be possible to somehow toggle them as well though. Maybe if the forward voltage is enough they could be connected to one pin on the elf, and toggle the pin from input low to output low. Effectively adding/removing 10k resistor from the circuit. You'd have to be careful though not to let the 5V run back to the mcu.

I don't know for this one, but most MCUs can only sind current in the same order of magnitude as they can source. There may be more margin, but don't expect too much (or read the datasheet )

Haha, that might be true

edit: 100mA rated max for GPIO sinking. This means that the chip cannot be used to drive in switch leds for an entire keyboard anyway without extra components, even if there was a more powerful regulator on board. Now if you add a transistor for controlling the output of an external regulator, thus achieving dimmable backlight, you could use the 5V line just as well.

[pomk]

I also would like a BT extension and about leds supporr I think more than the 3 locks and an accent light below the keyboard I wouldn't do more.. not interested in Christmas tree keyboards

I'm not talking about extension. That makes no sense if the purpose is in having a good battery life.

[pomk]

Adding the two resistors can be done with minimal modifications to the overall routing.

rev1_00.png (125.37 KiB) Viewed 4491 times

[matt3o]

do you think we need another round of prototypes or we could go straight to production?

[hasu]

do you think we need another round of prototypes or we could go straight to production?

what does 'production' mean, just run small group buy on DT for skilled users or early adapters?
I'd say you need more tests or iterations if it is not.

Did any of beta testers finish hand-wired keyboard or mother-daughter board scenario?

And I think we need to check kiibohd bootloader/firmware and Kinetis flash bootloader compatibility preferably.

[matt3o]

what does 'production' mean, just run small group buy on DT for skilled users or early adapters?
I'd say you need more tests or iterations if it is not.

well, if we produce just 25-50 of them the cost per board is pretty high (€25 each). I could make a small investment and produce more than we need to get a better per board cost, but of course the ELF must be spotless.

Did any of beta testers finish hand-wired keyboard or mother-daughter board scenario?

I'm working on a hand wired keyboard right now.

And I think we need to check kiibohd bootloader/firmware and Kinetis flash bootloader compatibility preferably.

this would be indeed very important. I sent some boards to haata, he is always super busy but I hope he could have a look at it.

Also I would probably need help setting up the website, documentation and all... ohmy...

[Wodan]

I can run sales/distribution for them in a GB style if that is appreciated. There is a _VERY_ attractive shipping method available from German Post that allows me to send small, slim packages insured&tracked worldwide for just 6€ which is a great option for these controllers.

Would run this in a similair fashion I did with the HADapter kit and similair to the planned Beamspring controller GB. Just round up my cost to the next full Euro during the GB phase and add a small margin once the items were ordered/received to compensate for my efforts of keeping stock/funding the production. Talking about something <10% here, mostly to make it more attractive for people to join during GB.

[pomk]

do you think we need another round of prototypes or we could go straight to production?

There are still some tests to be made. The two I had in mind are a shorting the regulator to see that no permanent damage is done and mechanical rigidity test of the USB socket. If the first test fails, then a fuse of some sort has to be added. The second one stops at either the USB connector failing (this is the preferred thing to happens and part of the micro-usb spec) or the socket failing (due to bad PCB or wrong amount of solder in the housings trough hole connections).

I can sacrifice one of my boards to this purpose

[tentator]

I'm doing a other handwired now and was waiting for a promicro so I can do it with this one instead..

About the bt capability then I did not understand : are we speaking of adding bt while still having same real estate size of two switches??

About leds then would 3 outputs be able to drive still 3 leds for caps scroll and num lock with just a resistor? (Need of 3v led?) For a led stripe instead I suppose a transistor on one output pin maybe pwm and 5v from usb should do right?

[tentator]

I think this one might deserve production of some multiple hunderts.. but demand could be not just a single shot gb but more something constant tough.. isn't there a DT shop section here?? I would offer also xwhatsit there.. I mean this stuff would always be demanded in time.. just not in peaks but more like flows

BTW I would test this one also to control an ibm trackpoint as usual in my kbds