Model MF - GB CLOSED

[Techno Trousers]

Well, this thread has been way too quiet lately. It definitely looks like CommonSense is going to win the poll, and I like the progress I've seen on DMA's CommonSense thread. Hopefully lot_lizard will be able to get back into the swing of things soon. We're coming into the home stretch now!

Also, I don't think it's been mentioned here, but rsbseb has rejoined us and is going gangbusters on his spherical buckling spring key cap project, code name SBS. He already has prototype molds and keycaps made, and has said he's going for full double shot plastic. I think those key caps are going to be an awesome companion to my Model MF keyboards. An MF keyboard with spherical keycaps in a beamspring colorway? Talk about back to the future!

[ekeppel]

I've been following that thread with interest too. Pretty exciting stuff!

It appears that several individual and group efforts have been coming together from different directions recently, and endgame could be finally on the horizon for some of us.

Of course it's always impossible to actually reach the horizon, isn't it? It always seems to stay ahead of us and there is a seemingly endless choice of routes we can take in our attempts to get there.

[andrewjoy]

i voted xwhatsit

just beacuse i already have 3 keyboards that use it so i want consistency

but i would not mind if the new one won

[Ray]

Does the choice of controller affect when this can ship?
If I can have this board two weeks sooner, I'll take that

[E TwentyNine]

This is close to an end game board. I'm fine with it taking a little longer.

[fohat]

I want something that is stable, durable, and future-proof as much as possible.

The xwhatsit that I have in my F-107 is quite nice and the mapping software works like a charm, but I voted for commonsense because it seems that simpler is almost always better.

[XMIT]

Hey lot_lizard, I haven't heard any updates here in a little while. How is manufacturing coming along?

[Phenix]

Voting for commensense, as I hope to see 'dual tap' feature from QMK on Model Fs
(tap esc, hold ctrl)

[Muirium]

Common sense. Active development is a killer feature. Xwhatsit's will take a lot of work to beat, but it's going nowhere so let's choose life.

[t!ng]

Hey Mu ...

[livingspeedbump]

This is close to an end game board. I'm fine with it taking a little longer.

All we need are some milled SSK cases and I can throw all my other boards away

[livingspeedbump]

I'm quite excited for the new controller actually.

[__red__]

... choose life.

Who's going to be the first to parody trainspotting

[DMA]

... choose life.

Who's going to be the first to parody trainspotting

I chose something else.

-=-
I'm pleased with 71% result - thanks for the trust, gentlemen.
Now I need to talk to someone on how actually produce the thing. Preferably using remote assembly services - reflowing 150 controllers is quite an undertaking (though certainly possible. I'll probably be legendary +5 PCB assembler by the end of it though, producing ☼Keyboard Controller☼s)

<b>Update</b> Produce = actually assemble. This thing came out to have SO MUCH safety margin in the sensing part that there's virtually no chance to fuck up PCB routing or be bitten by component quality. I mean, this iteration of PCB I forgot to route a damn POWER RAIL - and it still worked fine. And there's no part substitution possible - I already used cheapest resistors and caps I could find, and the chip doesn't have any alternatives.

[ekeppel]

This thing came out to have SO MUCH safety margin in the sensing part...[SNIP]

I'm sure very few here are concerned about this, but I wonder if you happen to know how the new controllers will handle strong RF interference? I have an amateur radio station and have to actually disconnect my F122 when I key up. If I don't, I get lots of random key presses sent that absolutely trash whatever I'm currently working on, lol.

It's probably no easy task to get a controller to reject that kind of noise. I would think that it's mostly being picked up by the sensors themselves, which can't really be shielded for obvious reasons, but perhaps whatever method you use to sift key-presses from noise will improve things. I will be interested to see how the final controller / sensor PCB compare to my current setup.

Thanks for all of your efforts!

[DMA]

This thing came out to have SO MUCH safety margin in the sensing part...[SNIP]

I'm sure very few here are concerned about this, but I wonder if you happen to know how the new controllers will handle strong RF interference? I have an amateur radio station and have to actually disconnect my F122 when I key up. If I don't, I get lots of random key presses sent that absolutely trash whatever I'm currently working on, lol.

It's probably no easy task to get a controller to reject that kind of noise. I would think that it's mostly being picked up by the sensors themselves, which can't really be shielded for obvious reasons, but perhaps whatever method you use to sift key-presses from noise will improve things. I will be interested to see how the final controller / sensor PCB compare to my current setup.

Thanks for all of your efforts!

So you are that legendary guy with 1200W transceiver? What's your current setup? xwhatsit or soarer? Any EMI shielding?

I have a couple of tricks in my sleeve made with you in mind.

Scanning machinery uses pass bands (so if transceiver induces MORE than the key usually outputs, the reading will be discarded) and IIR filter (so there must be just right amount of noise at just right amount of time several times in a row to trigger the keypress).
There's a small problem - ADC can't sense negative voltages. For model F it's much less of a problem, should be rock-stable (ADC readings must be in config.deadBandHi[row][col] - config.deadBandHi[row][col] + guardHi). For beamspring the problem is a bit bigger, because deadBandLo usually less than guardLo and readings are clipped at zero.
IIR is tunable - right now it's 3/4 (only 1/4 of the reading is produced now, 3/4 is historical average) - but changing one constant and recompiling can make it, i don't know, 15/16 - which will require probably 20ms of hold time for a key to register so you'll start worrying about not typing too fast or it misses the keypress In theory you can even go 127/128 safely - but that will be quite inconvenient in practice.

I would be more worried about transceiver inducing so much voltage it fries the circuitry. F122 makes for a pretty good Faraday cage - you'll probably need to construct paper/vinyl(to prevent shorts) + foil case for the controller section. I'm actually interested in what you have right now and what you tried for shielding if anything.

[Techno Trousers]

...foil case for the controller section.

That's great--a literal tinfoil hat for the controller!

[Madhias]

Due to my absence in the 2nd half of 2016 I missed that group buy =`(

[ekeppel]

So you are that legendary guy with 1200W transceiver? What's your current setup? xwhatsit or soarer? Any EMI shielding?

Well, not sure about the legendary part, but yeah, I was the guy asking for ideas for reducing RF interference on an XT board a while back. Same problem with this F122. Both were just simple Soarer's converter mods, with no additional shielding or anything. I have tried some large ferrites on the cables as a test, but the problem seems to be the controller or keyboard PCB picking up the RF rather than the cable.

My station puts out around 1300W on voice peaks and the antenna feedpoint is only about 30ft above my office, so I get a pretty decent amount of RF 'in the shack', as they say. This would most likely be a combination of direct feedpoint radiation along with RF travelling back down to the equipment on the shield of the coax. Radiation exposure falls within safety guidelines from a human health standpoint, but isn't so great with poorly shielded electronic devices, lol.

I could probably greatly reduce the RF on the coax by adding a nice big choke balun at the feedpoint, instead of trying to put 'band-aids' on affected devices (such as this keyboard). Until recently the cold weather made it a hassle to get the antenna down for servicing. Now that Spring is finally here, I may get to that and try to mitigate the source of the RF somewhat.

I have a couple of tricks in my sleeve made with you in mind.

IIR is tunable

This configurability is really excellent to have, and should allow me to tweak things to work the best for my situation. Thank you for this!

I would be more worried about transceiver inducing so much voltage it fries the circuitry. F122 makes for a pretty good Faraday cage - you'll probably need to construct paper/vinyl(to prevent shorts) + foil case for the controller section. I'm actually interested in what you have right now and what you tried for shielding if anything.

I honestly don't think there is a need to worry about induced voltage. I'm not setting off an EMP here, and in truth the RF fields are really quite weak by the time they reach the keyboard. It's just enough to disrupt the careful balance and trigger key presses.

My impression is that the F122 isn't particularly well shielded, though. The PCB hangs off the back and is fully open to RF through the plastic edges of the case. Each key sensor is also pretty much wide open. Every key / barrel is, from an RF perspective, an unshielded hole directly to the PCB. Yeah, I could probably put a tiny foil piece between each keycap and stem, but that's just getting crazy with the band-aids, and still leaves a lot of RF holes near the base of the barrel, rofl.

I think I will give the 'tin-foil hat on the controller' approach a try, and run some tests to see if it has any effect.

Thanks for all the nice input on this. RF is a tricky thing to deal with sometimes.

...which makes me wonder... Did IBM ever make a Tempest hardened F122? Hmm...

[ekeppel]

Due to my absence in the 2nd half of 2016 I missed that group buy =`(

You're definitely not alone. I managed to get in on this GB just before it closed only because I blindly stumbled upon it in Google's search results while researching a completely unrelated keyboard.

If you read back a ways in the thread, I believe Lot Lizard said he planned to have a few available for sale later on as a way to jump start funding for Phase 2, but quantities will be very limited and the price will obviously be higher.

[DMA]

Each key sensor is also pretty much wide open. Every key / barrel is, from an RF perspective, an unshielded hole directly to the PCB. Yeah, I could probably put a tiny foil piece between each keycap and stem, but that's just getting crazy with the band-aids, and still leaves a lot of RF holes near the base of the barrel, rofl.

Well, unless you're going higher than 1GHz (I would say "5cm band", but I'm being conservative here) those holes are too small for RF to penetrate the sense card cage. I would think improving controller (that PCB outside of the metal frame and wires coming to it) shielding will fix the problem. I don't have 1.3kW transmitter anywhere near me to test though.

The keyboard cable is shielded already if I'm not mistaken.

I think I will give the 'tin-foil hat on the controller' approach a try, and run some tests to see if it has any effect.

Don't forget to post the results! I would be very interested.

Thanks for all the nice input on this. RF is a tricky thing to deal with sometimes.

It also produces very painful burns which take long time to heal. Even 50W, as I found out in school. Careful over there.

[ekeppel]

TL;DR It's fixed. The RF was on the coax shield and was mitigated with an RF choke.

Well, unless you're going higher than 1GHz (I would say "5cm band", but I'm being conservative here) those holes are too small for RF to penetrate the sense card cage.
...
Don't forget to post the results! I would be very interested.

I imagine you are correct about the holes being too small, since we are only talking about 1-30MHz here, but as you know strong enough signals can cause harmonics to do all sorts of unexpected things. I wanted to rule that out, and I did, since no amount of shielding on the controller made any difference. I even wrapped the entire board in foil with no effect, which told me that the RF was coming in on the USB cable, but putting ferrites on that cable didn't fix it.

I suspected earlier that I was dealing with RF on the coax shield, and it turns out that was the problem. There was a lot of RF coming back from the antenna on the shield of the coax, and that coax then goes to various connections on my desk. The amplifier is just to the right of the keyboard, with the transceiver and antenna tuner to the left, so that creates opportunity for a lot of RF to radiate mere feet from everything.

I made a simple RF choke using seven large mix-43 ferrites in series on a short jumper of coax and inserted it in-line where the antenna coax comes through the wall near my desk. As soon as I did that, all signs of problems with the keyboard went away.

There is still some RF present, since I can hear it in my PC speakers, but it is greatly reduced from what it was. I believe it will go away nearly entirely when I move that choke up to the feedpoint above the house. Right now, the choke is temporarily installed only a few feet from the computer, so there is a good length of coax nearby that is still radiating RF, lol.

Thanks for the help. Nice to finally have this RF under control now.

[DMA]

While we're waiting for lot_lizard to surface.. Got that bright idea that I should minimize the changes and go with the existing form-factor. The resulting PCB is a bit barren

csmf-front.png (41.35 KiB) Viewed 5632 times

Getting the hang of that immersion gold printing

Unfortunately, MF logo was too hard to convert to DXF diptrace would understand (I suspect diptrace cannot into arcs in patterns - need to approximate by polylines). Also mounting holes are wrong - I'll make special components for those, with large shiny square pads.

PS: Also it looks to me the size is a bit off - 150 mil too narrow, to be exact. Will fix along with the mounting holes.

[__red__]

Thanks for the help. Nice to finally have this RF under control now.

Remember to test it on 160m with RTTY over AM.

I used to come over my neighbor's microwave with that setup at 400W

[looneyperson]

Aw man, I wish I hadn't been absent from deskthority for so long. I would have loved to convert my SSK into a model F. Oh well there is always next time.

[DMA]

csmf-front.png (40.42 KiB) Viewed 5529 times

Here it is, the mini-xwhatsit drop-in replacement version. 3.39x1.00 inches, $17 for 3 at oshpark.
I even managed to make the "CommonSense MF" decal - the MF logo was between C and S, 0.55" high - but oshpark throws internal error when I try to import those gerbers. So shown without.

I secretly hope someone will tell me why it threw the error - and I can order the prototypes in their full glory.
Also, since there's free PCB space - I wonder what other art to put there..

[ekeppel]

Thanks for the update - Looking good!

I wonder what other art to put there..

Looks like there is plenty of room still to mount one (or preferably more) 'mandolin' crystals on there somewhere.

[DMA]

Looks like there is plenty of room still to mount one (or preferably more) 'mandolin' crystals on there somewhere.

TIL "mandolin crystal". This shit is _weird_. Who even cares!
But if there will be no ideas I swear I'll put "SPARCO" and "A'PEXi" there.

[Techno Trousers]

You should put a URL to the GitHub repo where the CommonSense software project will live. Might as well use the space for something truly useful!

[DMA]

You should put a URL to the GitHub repo where the CommonSense software project will live. Might as well use the space for something truly useful!

In gold letters, no less. Might as well..