Ellipse Model F62 only rarely comes alive

[enthdegree]

Hello, I recently bought one of Ellipse's ultracompact F62's secondhand. Trying repeatedly to plug/unplug the keyboard with a known-working USB-C data cable, it only appears in dmesg maybe once every 20 minutes. Once the keyboard comes alive it works fine for at least several hours with no particular sensitivity to cable orientation. Does anyone know what might be causing this?

I can only guess at some possibilities, though none jump out to me as particularly likely:

• There are floating grounds somewhere among the case, controller, and computer.
• The computer's USB port isn't able to consistently provide the keyboard enough power to boot up.
• There are shorts or dead solder joints somewhere inside the case.
• The USB-C connector on the xwhatsit controller is worn out or damaged.

I doubt it is a bad USB connection since the keyboard works so reliably once alive. While it's working, unplugging & replugging from the computer makes it go dead again. Are there any test points to check and common components to replace or reflow?

Thanks!

[AndyJ]

Try it on another computer.

Try it on a powered USB hub.

Ask the manufacturer how much an out-of-warranty repair might cost.

[Muirium]

Use its shocking heft to beat up the scumbag who sold it to you as working!

No kidding though: Ellipse will skin you alive. I would not pursue that route!

[pandrew]

Is this a USB-C to USB-C cable? If yes, try a USB-A to USB-C cable.
The (old version of) the controller that comes with most F62s does not support C-to-C cables. (and if that's what you have, then I'm surprised it even works sometimes, I expect C-to-C to not work at all)
I think there may be a new version that fixes C-to-C support but I don't remember if it's compatible with the F62.

If it's not an C-to-C cable:
Then just try other computer, try other cable, try without hub is good advice.
If you have solenoid driver connected, then try without it.

What OS do you use?
What happens when you plug it in and doesn't work? If on Linux, anything shows up in dmesg? if on Windows, anything shows up in device manager?
Short list of things to check: USB connection to +5V, ground, D+ and D-, keep in mind that you may get ground from two sources: the actual ground wire in the USB cable, or from the shield, if your usb cable has a shield, because the wcass board itself connects shield to gnd. So if you have trouble with gnd , you might notice a shielded cable work better. Then check the reset signal, it should be high when you measure it. The capacitor on the reset signal could also become bad (although I think the controller would still work in that case, cause atmega32u2 has power-on-reset). Then check the oscillator circuit. XTAL2 is the output, measure it with an oscilloscope, or if you don't have one, you can also often check these with a multimeter to with DC voltage measurement, I'm measuring 0.5 .. 0.8V on the crystal pins of a working board. If you're measuring much less than that or much more than that on XTAL2, then there may be something wrong with the oscillator circuit.

[enthdegree]

Is this a USB-C to USB-C cable? If yes, try a USB-A to USB-C cable.

I was indeed a C-to-C cable and now it works reliably with an A-to-C. Thank you.
The other information is helpful too. Is the controller hardware documented anywhere? (guessing, is it a plain version of this: https://github.com/sethbc/xwhatsit-capsense-controller ?)

[Ellipse]

Glad you got it working! The new Model F's are originally supplied with an A to C cable so I'm guessing there was an issue with the other cable. I have plenty of the original custom made A to C cables in the shop.

Also I have a bunch of presoldered controller+ribbon cable+large capacitive PCBs that can serve as drop in replacements for these keyboards in case they are needed. Anyone who needs one or wants a spare please PM or email me for details.

Some helpful documentation links from xwhatsit:
https://static.wongcornall.com/ibm-capsense-usb/
https://static.wongcornall.com/ibm-caps ... e-usb.html

[Muirium]

Well spotted, Pandrew. I ran into this myself when I installed a USB C port inside my IBM AT. I simply bought a short female C to male Mini USB dongle to extend from my Xwhatsit controller to the original cable opening. Then when I plugged in a USB C to C cable and hooked it up to my Mac, it was nowhere to be seen. Huh? Tried a USB A to C cable and it worked perfectly. So I shrugged it off as a USB C backwards compatibility issue.

I believe a resistor needs installed between some of the C pins to indicate this is a USB A connection, riding on a USB C cable. Sounds like that’s what was subsequently done to later versions of the controller.

[ed.avis]

Hi, I discovered the same thing -- connecting the F62 to my laptop using a USB-C data cable did not work but to a desktop with USB-C to USB-A is fine. What are the options, then, for connecting it to a device that only has USB-C support? Can I use the USB-C to USB-A cable together with an adaptor, just as I might plug in an older USB device to the laptop?

[modelf]

What are the options, then, for connecting it to a device that only has USB-C support?

I have a Dell laptop that's USB-C only, but it came with a Dell OEM C-to-A adapter, and it works great. My Model F/AT keyboard (IBM original) with a USB-A Soarer's converter, and a 1986 Model M with the same Soarer's, both work on the laptop with the Dell adapter (and also with this 3rd-party adapter from Amazon).

[ed.avis]

Thanks. I believe with the right C-to-A adaptor the board will work. It's ugly, though, that you have to use the C-to-A cable and then the opposite adaptor.

For the record, here's the thread where I asked on the New Model F Q&A. The admin does not so far admit that this problem exists, contending that it must be an issue with the USB-C cable(s) I tried. https://www.modelfkeyboards.com/questio ... b-c-cable/

(I was hoping for a response from the project like "oh yes, boards manufactured before such-and-such a date use an older version of the controller PCB, but you can purchase a newer one and replace it".)

[mattaw]

For the record, here's the thread where I asked on the New Model F Q&A. The admin does not so far admit that this problem exists, contending that it must be an issue with the USB-C cable(s) I tried. https://www.modelfkeyboards.com/questio ... b-c-cable/

(I was hoping for a response from the project like "oh yes, boards manufactured before such-and-such a date use an older version of the controller PCB, but you can purchase a newer one and replace it".)

I do believe Ellipse, above in the thread, is the project founder (and probably the admin too!) and has offered to help already.

Another data point, my new B104 beam-spring round 1 keyboard does not work with my X570 AMD motherboard USB-C port with an expensive C-to-C cable, but I have not tested the cable extensively so I do not know if it is compliant. I did not photo the controller when I was inside adjusting beam tension earlier today, but I will do and report back next time I crack it open.

(On the lack of a recommendation for a USB-C to USB-C cable I am sympathetic to the admin sticking to what they sell and ship. As it impossible to know if a USB-C cable is any good without testing it (and even high cost ones can still be out of spec. see the testing by Benson Leung) they risk giving bad advice and wasting your money and time. In general it is just not easy to make a general long-term recommendation, especially on cables considering how many manufacturers constantly rebrand and change cable designs without updating the part numbers.)

[Ellipse]

Yes confirmed, I am the admin on the project web site. As discussed elsewhere, the recommendations are to use the provided cable with an adapter plug or use a "dumb" USB-C to USB-C cable that is not active.

[Muirium]

The “dumbest” USB C to C cables I have—which surely have no smarts at all—are all too smart for this. How dumb must it be?

The only way I can think of would be to hack the cable’s innards, which while very dumb, isn’t the kind of dumb you’re on about.

[ed.avis]

Thanks for getting in touch, Ellipse. My question at the moment is not "how can I use the keyboard with a USB-C laptop" -- as you note, it is possible to use an adaptor. Rather, I wanted to reproduce the issue and narrow it down. I don't believe that it is limited to smart cables or USB hubs.

Until now I've been trying the cheapest USB-C to USB-C cables I could find, and none of them worked. But I tried Apple's 60W USB cable, which has "Apple A2795 China" written in tiny letters on one of the connectors with a serial number below. As far as I know, you can walk into any Apple Store and buy another one just like it. That one works fine to connect the F62 to the Dell 5750 laptop.

So to summarize what I found so far:

- cheap USB-C to USB-C cable: does not work (tried various)
- cheap USB-C to USB-C cable, then C-to-A adaptor, then A-to-C adaptor: works
- Apple USB-C to USB-C cable: works
- Apple USB-C to USB-C cable, then C-to-A adaptor, then A-to-C adaptor: does not work. (that surprised me)

I don't think the cheap cables I tried were active or smart ones; there is some other kind of magic going on. I wonder whether the others on this thread who've seen a similar issue will be able to try the Apple cable.

I understand that replacement controllers are available. What's not entirely clear to me is whether these controllers are a newer revision that has some kind of fix to USB support, letting them work with a wider variety of USB-C cables (although still excluding smart and active cables).

[Muirium]

I'll check my iPhone's USB C cable to see if it's an "Apple A2795 China" as you said. (The pics I’m seeing in a web search suggest it could be.) It might work with my USB C modded AT, in which case it should work with all Xwhatsit-based controllers.

[Ellipse]

I confirmed that an old Samsung (?) USB-C to USB-C charger cable worked fine in my testing. Interesting that it did not work in your testing and that you got it working with the Apple C to C cable. Maybe certain C to C cables have different pins wired, meaning some pins are left unwired? Maybe a USB 2.0 speed C to C cable would be better (the A to C cables for the project are USB 2.0 type).

The issue with the USB-C to USB-C cables was that the controllers needed to add 5.1k pulldown resistors to certain pins of the USB-C connector, as pandrew noticed when helping to update the wcass controller in 2022. I don't remember the specifics but this was updated for the latest production 33 pin xwhatsit controller being used on many of the keyboards finishing up production this year (except for the ones with extra columns that use Rico's Leyden Jar controller which did not need a correction). The 30 pin controllers were made before this update and will not be made again as they are the older version. The 33 pin controller version allows 3 LED lock lights and a solenoid to be used at the same time, with the 3 extra pins for the 3 LEDs now that boards such as the F104 have LEDs built in.

https://hackaday.com/2023/01/04/all-abo ... -emarkers/

[ed.avis]

Thanks for the info. To fully resolve this question I'd need to update my keyboard to the latest controller (I assume it has an older controller, having been made in May 2021) and then retest the various cables. If one day I should happen to buy another New Model F, I will revisit this. For the moment I will take the Apple cable when I need to use it with a laptop.