Largest Keyboard using Arduino Pro Micro as Keyboard Controller?

[clickclack123]

Probably soarer or hasu could answer this question best:



How many is the most keys that you could connect to an Arduino Pro Micro when using it as a controller for a keyboard?

From the specs here, it says it has:

4 x 10-bit ADC pins
12 x Digital I/Os (5 are PWM capable)

So am I correct in assuming that the biggest keyboard you could do without external expander IC's is a 6x6=36 key matrix? Are there any other tricks to get the most out of such a small board? Any relevant links someone can point me to about creating a matrix that would work with hasu's or soarers code?

I'd like to use this board because they are so cheap on ebay, only $5 or so delivered! That's cheaper than just the bare ATMega32u4 by itself!

[Pac Caps]

I've read that these can handle the Model M 122 key models. So at least 122 )

[clickclack123]

I've read that these can handle the Model M 122 key models. So at least 122 )

But how do they do it? There doesn't seem to be enough pins. Maybe using an IO Expander chip (like on the ergodox)?

Can you find a link where you read that?

[Pac Caps]

I've read that these can handle the Model M 122 key models. So at least 122 )

But how do they do it? There doesn't seem to be enough pins. Maybe using an IO Expander chip (like on the ergodox)?

Can you find a link where you read that?

It was somewhere in the Soarer's converter thread.

The M uses a 5 pin terminal connector, so you just need to solder the 5 pins to the controller, and load soarers

[regack]

I've read that these can handle the Model M 122 key models. So at least 122 )

But how do they do it? There doesn't seem to be enough pins. Maybe using an IO Expander chip (like on the ergodox)?

Can you find a link where you read that?

That's probably when they're used as a converter.  In that case you're just hooking up a couple of wires for it to sit between the normal output of the keyboard and the computer.  It's just an atmega32u4, with what looks like 18 pins usable... so... I guess you could do some kind of weird 9x9 and have 81 keys?

[JPG]

I've read that these can handle the Model M 122 key models. So at least 122 )

But how do they do it? There doesn't seem to be enough pins. Maybe using an IO Expander chip (like on the ergodox)?

Can you find a link where you read that?

It was somewhere in the Soarer's converter thread.

The M uses a 5 pin terminal connector, so you just need to solder the 5 pins to the controller, and load soarers

When used as a converter, id does need many pins. Only enough to connect the cable, so 4-5 as far as I know.

When used as a controller, you typically plug your rows and columns to the pins of the controller. So depending on how you design the pcb, at most you could do something like (number of pins/2)square. But most pcb's don't optimise the controller pins usually so it's more like real number or rows times real number of columns.

And the pro micro has less pins than a teensy so you could not use it for a tkl. Maybe for a 60%, but I am not even sure. I'll try to find the place where Soarer mentioned the number of pins available on the pro micro.

[Pac Caps]

I've read that these can handle the Model M 122 key models. So at least 122 )

But how do they do it? There doesn't seem to be enough pins. Maybe using an IO Expander chip (like on the ergodox)?

Can you find a link where you read that?

It was somewhere in the Soarer's converter thread.

The M uses a 5 pin terminal connector, so you just need to solder the 5 pins to the controller, and load soarers

When used as a converter, id does need many pins. Only enough to connect the cable, so 4-5 as far as I know.

When used as a controller, you typically plug your rows and columns to the pins of the controller. So depending on how you design the pcb, at most you could do something like (number of pins/2)square. But most pcb's don't optimise the controller pins usually so it's more like real number or rows times real number of columns.

And the pro micro has less pins than a teensy so you could not use it for a tkl. Maybe for a 60%, but I am not even sure. I'll try to find the place where Soarer mentioned the number of pins available on the pro micro.

Ah I see. Because the M 122 already has a controller inside of the case, the Pro Micro just uses the 5 pins to convert the code to USB

[Soarer]

Yep, as a controller it can handle 9 x 9 = 81, since the analogue pins, TX and RX can be used as digital pins as well. There are some LEDs on board on a couple of pins, and they might just flicker with the scanning (not a big deal, but they couldn't easily be used as indicators at the same time).

As it happens, the 122 M has a 4-to-16 line decoder driving 16 of its 20 strobes. So 20 x 8 = 160 keys (sparse to avoid ghosting on shifts etc), and would only use 4 + 4 + 8 = 16 pins, if you used the decoder So Pro Micro would be enough to replace the CPU on a 122 M's controller PCB (probably irrelevant here, but hey!).

The Pro Micro schematic is what you need to see the 'real' pin names rather than the Arduino ones.

[clickclack123]

Thanks everyone.

Damn.

So you would need an IO expander to do a full 104-key keyboard. Then you'd need a pcb so you might as well just put an atmega on the pcb yourself and break all the pins out so you don't need the IO expander!

When are the Arduino Micro (not pro micro) clones going to show up on ebay? That could replace the teensy easily. They'd end up selling a lot of them to keyboard junkies. Oh well.

[jorgenslee]

Yep, as a controller it can handle 9 x 9 = 81, since the analogue pins, TX and RX can be used as digital pins as well. There are some LEDs on board on a couple of pins, and they might just flicker with the scanning (not a big deal, but they couldn't easily be used as indicators at the same time).

As it happens, the 122 M has a 4-to-16 line decoder driving 16 of its 20 strobes. So 20 x 8 = 160 keys (sparse to avoid ghosting on shifts etc), and would only use 4 + 4 + 8 = 16 pins, if you used the decoder So Pro Micro would be enough to replace the CPU on a 122 M's controller PCB (probably irrelevant here, but hey!).

The Pro Micro schematic is what you need to see the 'real' pin names rather than the Arduino ones.

Thank you for this Soarer. I recently got 2 Pro Micro for $5/each. I plan to hardwire a keyboard and luckily the plate has 79 keys.

[kolonelkadat]

theoretically, couldnt you get like i dunno 240 keys with the 16 io pins using the magic of charlieplexed diodes? you may not get nkro out of this arrangement though.

[bcg]

If you're good you can try soldering some wrap wire directly to the unused pins.  I'm sure I couldn't pull that off though, my hands are way too shaky

I would probably ruin at least one in the attempt so I would maybe just go with this instead:

http://www.ebay.com/itm/Micro-ATmega32u4-5V-16MHZ-100-compatible-Arduino-Mirco-Replace-pro-mini-/181181371949?pt=LH_DefaultDomain_0&hash=item2a2f40562d

That has the same bootloader as the Pro Micro but you get the 5 unused pins shown on the Pro Micro schematic, plus an onboard reset (the lack of a reset button on the Pro Micro is extremely annoying) plus a 3V regulator and output pin if thats useful to you.  The Arduino Micro is actually a really nice dev board, and the price on the knockoff is really not bad for what you get

[yasuo]

Pro micro is very small ) but micro usb i think better than USB regular,not easily separated
i will plan also for handwired just 64keys

[Soarer]

theoretically, couldnt you get like i dunno 240 keys with the 16 io pins using the magic of charlieplexed diodes? you may not get nkro out of this arrangement though.

1KRO. Unless... I dunno if this would work, but using diodes with just the right voltage drop across them might enable it to be more thn 1KRO. AVR inputs are LVTTL - a 'low' is below 0.2 * VCC - 0.1V and a high is above 0.2 * VCC + 0.9V. So at VCC = 5V, low < 0.9V and high > 1.9V. You'd have to arrange it so that 2 diode drops read differently to a single drop. That means using active high strobes and (external) pull-down resistors on all matrix IO pins, because 0.9 < 1.9 / 2. Diodes with a drop of between 0.5 * (5 - 0.9) = 2.05V and 5 - 1.9 = 3.1V could be worth trying, if you can find them. It would be tricky to find the right zener though, because the voltage drop depends on current and the current flowing through a switch matrix is low (but could be tuned by the pull-down resistor value used). Also, with multiple keys pressed, parallel paths would reduce the current through each path. A starting point for tests might be 3V zener diodes with pull-down resistors of 3k3, which would flow about 0.7mA and give a zener voltage drop of about 2.7V, so 2.3V on the input pin for a single drop path.

edit: zeners won't work, their voltage is when reverse biased

edit2: LEDs might though! Simple red ones, with a rated Vf of about 2V, would actually have a Vf at low current (<1mA) of say 2.5V which would be ideal. Minor side effect that the LED would glow slightly under pressed keys... oh well

[doub]

4 x 10-bit ADC pins
12 x Digital I/Os (5 are PWM capable)

So am I correct in assuming that the biggest keyboard you could do without external expander IC's is a 6x6=36 key matrix? Are there any other tricks to get the most out of such a small board? Any relevant links someone can point me to about creating a matrix that would work with hasu's or soarers code?

One trick you can use is using the ADC pins as analog inputs, in order to get more than one bit per pin. You'd need a more complex wiring, with more than one line (column or row) connected tot he pin through different resistor values. By sampling the ADC, the analog value would tell you how each of these lines are energized. You won't get 10 bits (ie. 10 connected lines) per ADC, because of the resistor precision (if you want them cheap, they will be 10% or even 20% tolerance), but you can probably connect 4 lines per ADC reliably (maybe more with calibration if your resistors are stable over time/temperature).

That would give you a total of 12 (digital out) x 4 (analog in) x 4 (analog levels) so 192 keys. The cost is more complex wiring and programming.

[Soarer]

Seems Semiconwell make some R-2R resistor ladders that could be used for that. Think they're 1 or 2%.

[Grendel]

Parallel-to-serial shift registers.

[Soarer]

Decoders... multiplexers... so many ways to skin this rat.

[clickclack123]

Just a small follow-up question,

Since I've established that the pro micro won't suit my situation, can someone confirm that it's not too hard to get soarer's or hasu's firmware working on a normal Arduino micro before I order one? Like the ebay like posted by bcg.

[bcg]

Just a small follow-up question,

Since I've established that the pro micro won't suit my situation, can someone confirm that it's not too hard to get soarer's or hasu's firmware working on a normal Arduino micro before I order one? Like the ebay like posted by bcg.

Just tested on a genuine Arduino Micro, and these steps will work:  http://geekhack.org/index.php?topic=48650.msg1060586#msg1060586

With the Micro though instead of shorting the reset to ground you can just push the reset button and the indicator LED "breaths" for eight seconds giving you time to load the hex

That is assuming the Chinese board is using the same bootloader as the genuine board, which it probably is

[clickclack123]

Great, thanks for testing it out.

[clickclack123]

OK, another follow-up question,

What's the largest keyboard that can be made using the Arduino Micro without external devices or other complicated shenanigans?



From the specs:
20 Digital I/O Pins
12 Analog Input Channels

But I think they're shared, ie a pin can be either a digital or an analog input.
I'm unsure if you can use MOSI/MISO as normal digital inputs. The schematic is here.

Sorry if I'm sounding like a n00b, I have had a play with the basics of the Arduino, but not for building a keyboard, so I've never had the problem of needing to repurpose pins for a keyboard matrix yet...

Plus I'm a cheapskate, so if the Micro for $15 delivered will do the same job for me as a Teensy for $26 delivered, then I'll go that way... I might even learn something in the process of making it work!

[doub]

The rule is simple, the number of keys is the number of columns multiplied by the number or rows. You need one I/O per column and one per row. To achieve the largest number of keys, you need to balance the two (but that makes wiring more complicated since it won't match physical columns/rows).

So with 20 digital I/Os, you can at best get 10 columns x 10 rows, or 100 keys. If MOSI and MISO can be used as digital I/Os, that's 11x11 or 121 keys.

[clickclack123]

The rule is simple, the number of keys is the number of columns multiplied by the number or rows. You need one I/O per column and one per row. To achieve the largest number of keys, you need to balance the two (but that makes wiring more complicated since it won't match physical columns/rows).

So with 20 digital I/Os, you can at best get 10 columns x 10 rows, or 100 keys. If MOSI and MISO can be used as digital I/Os, that's 11x11 or 121 keys.

Can MOSI and MISO be used as digital I/O then? If they can, it will hopefully simplify the wiring necessary as I approach the maximum number of keys.

[doub]

From the schematic they look like they can be used as GPIO, but I can't say for sure. The info is probably somewhere in the MCU datasheet, and maybe someone with Arduino experience will be able to give a definite answer.

[regack]

You can use the MOSI, MISO, SCK pins, yes.  You should be able to use 26 pins on there, which would  at max, be a 13x13 matrix of 169 keys.  The pins with LEDs might be weird. 

[clickclack123]

Great! Just ordered one!

[obra]

Can MOSI and MISO be used as digital I/O then? If they can, it will hopefully simplify the wiring necessary as I approach the maximum number of keys.

Yes. I do that regularly with the keyboards I build with micros

[clickclack123]

You can use the MOSI, MISO, SCK pins, yes.  You should be able to use 26 pins on there, which would  at max, be a 13x13 matrix of 169 keys.  The pins with LEDs might be weird.

Wouldn't it be 24 pins? I've marked the pins that can't be used on this image in blue. The ICSP header has no pins that can be used because they've been duplicated on the outside pins.

Seems wasteful of real-estate to have two NC pins on the outside, and duplicated pins on the ICSP header as well. They could have fit four more IO's on the board.

I actually only need 18 pins for a keyboard matrix, I want to use the extra pins for led's.

[mkawa]

ideas:

1) charlieplex? strobe the switches?

2) run the alphanums and arrows on a traditional key matrix and then pulse the rest of the keys via serial interface?

afaik the only reason you'd go with a micro pro over a teensy 2.0 for a keyboard controller though is price, and this is a lot of design to go through to save 10 bucks.

[MOZ]

How many switches do you intend to use? And LEDs?

[clickclack123]

How many switches do you intend to use? And LEDs?

An 8x10 matrix with as many led's as possible!

[clickclack123]

ideas:

1) charlieplex? strobe the switches?

2) run the alphanums and arrows on a traditional key matrix and then pulse the rest of the keys via serial interface?

afaik the only reason you'd go with a micro pro over a teensy 2.0 for a keyboard controller though is price, and this is a lot of design to go through to save 10 bucks.

I wouldn't go through all the effort of those suggestions, too much effort really. I have thought about using one Pro Micro for each hand though, and programming them to speak I2C or something between the hands. I have been watching the OneHand project on Deskthority wrt this, but haven't tried myself.

Although I think the teensy is a great product, I'd rather learn to use the open (and cheaper) version TBH.

For me in Australia, a teensy is $26 delivered (25 IO), a micro is $15 delivered (24 IO, I think), and a pro micro is $5 delivered (18 IO). So it's a significant cost saving in percentage terms for equivalent hardware IMO. I've been testing the micro and it seems to do exactly the same job as the teensy would.

I have thought about using a micro for the matrix, and talking i2c to a pro micro for a lot of leds. The pro micro is a very flexible board for $5. It would take a bit more programming skill than I currently have, but not unachievably so, and the journey of learning how to get it working might be fun for me. I could probably even repurpose a lot of the Ergodox TMK firmware code for this, since it talks i2c to the IO expander.

[mkawa]

what's not open about the teensy? they're really just simple platforms for the exact same chip. the micro pro has fewer GPIOs broken out.

also, for the LEDs, you're going to need a big fat driver for that network anyway, so you might as well do a second chip..

if you're doing a custom PCB, i would just put the atmega32u2 or u4 directly on your board and use it for the key matrix, and then source a large relatively high current LED network driver; there was one we were looking at quite a while ago from a german company whose name i can't remember (__red__ found the chip for us). it was difficult to get large quantities, but really easy to grab a couple samples.

all that information is in "the light" project thread, which is somewhere around here. alaric went as far as grabbing samples and doing a small LED board via OSH with the chip. it was quite capable and could be run on high-ish USB current (1A) or normal (500ma).

[clickclack123]

what's not open about the teensy? they're really just simple platforms for the exact same chip. the micro pro has fewer GPIOs broken out.

also, for the LEDs, you're going to need a big fat driver for that network anyway, so you might as well do a second chip..

if you're doing a custom PCB, i would just put the atmega32u2 or u4 directly on your board and use it for the key matrix, and then source a large relatively high current LED network driver; there was one we were looking at quite a while ago from a german company whose name i can't remember (__red__ found the chip for us). it was difficult to get large quantities, but really easy to grab a couple samples.

all that information is in "the light" project thread, which is somewhere around here. alaric went as far as grabbing samples and doing a small LED board via OSH with the chip. it was quite capable and could be run on high-ish USB current (1A) or normal (500ma).

Problem is that from anywhere I've checked, the bare 32u4 costs more than the pro micro! Plus I'd have to do smd IC soldering, which I'm slightly scared of, although I'm fine with smd resistors/diodes.

I'm hardwiring a prototype atm also, so the bare chip would not be good for my situation. I have been investigating pcb design, just got a bit stuck over the weekend in kicad (actually that reminds me, I might start a new thread to see if anyone can help me with that).

How much current could a 32u4 supply by itself? A single led uses about 15mA, I believe. I just want a few to show what layer I'm on, etc.

Edit: Datasheet says 200mA total, 40mA per pin absolute maximum. So 16 led's at 12mA would be doable. I don't know how much those led drivers cost or how easy they are to get, but I like the cost and accessibility of the pro micros. Plus I can repurpose and use them for different things.

Also if I remove or connect thin wires to two of the leds that are already on the pro micro board, I can get two more IO's, making it 20 total. That's an unbeatable 4 to 1 IO to dollar ratio!! *cheapskate mode off* 

Any chance you could find me a link to that "the light" thread? Did a bit of a search then, unfortunately couldn't find it.

[clintwn]

Surface mount ic is usually easier. Paint on paste, get your George foreman to the right temp, watch it flow, lift off the board. No need to mask, because solder will be attracted to the traces.

Oh yeah, and when you're done, light rubbing of your iron against wick against traces makes short work of shorts.

[clickclack123]

Yes I would have to agree with you now actually, just changed the regulator on my pro micro for a 3.3V one using a hot air soldering station, and it was super easy and looks totally professional.

Pro micros are still cheaper than the bare chip though!