MIDIbox SEQ new frontpanel idea

[latigid on]

This last order had 11 different PCBs in it and includes everything needed for a SEQ, like the MIDIO/I2C/line transmitter and also a new design for a Core that doesn't rely on the Discovery board. The USB entry module had some fab issues, because it's not clear how to call out plated slots:

(the thing to do is to give the actual cutout of 0.8mm min. width in a separate file, not just the router path as I've seen before)

But hopefully they'll be on the way soon.

The subsequent steps will be to build and test to make sure it's all sound, then send a set to AdrianH, so he can work on the case once all of the heights/dimensions are known. Of equal importance is to get TK. on board, as this is three different BLMs that will need updated programming and HW config files. He is extremely busy this year but if the past is an indication, he will give a bit of his "vacation" to the project. Otherwise I'll take the train to Munich and sleep on his doorstep :P.

Hopefully things become clearer in the next month or so. If all is well I can start to take orders, although I may have to look further as the forum is quite quiet at the moment.

@gerald.wert 
No problem! I spend the time on these projects when I can because I want something nice for myself, and I also want to do good by those who want to take part. But this is MIDIbox, and if you have the time and motivation, everything is there to build your own custom unit.

Best regards,
Andy

 

 

[u-link]

Awesome, 1$ per switch is really way cheaper than I'd have expected! Great to hear you are already at that stage. Keep us updated!

[borfo]

8 hours ago, latigid on said:

This last order had...

I'm in the process of designing myself my perfect self-contained MIDI jam case, and I've been planning to make a new SEQ anyway.  I've already got a bunch of these panel switches and they seem pretty ok...  I think I may as well try a panel mount job with a few customizations.  (I had actually also settled on using rows of 8 instead of the Wilba design.) Great timing that this new design is finalized though - wouldn't want to build a new seq only to find that everyone's using something totally different now.

Anyway, great work on the design.  Looks great, and the modular design will be totally useful.  Same with the consistend 8 button grids.  Everything's 8 in Midi, and in a lot of music.  Having that consistency in the button rows will naturally lead the interface to become more useful and intuitive, I think.

 

Edited May 11, 2017 by borfo

[borfo]

I've been looking at these mechanical keyboard switches - there really are a lot of options with these things, and a lot of keyboard nerds working on various projects.  And knockoff keys and caps can be had pretty cheap for those of use who like to cheap out on components a bit where possible.

You may want tio take a look at some of the matrix keyboard (straight grid-style layout) designs out there.  There are some PCBs available - often pinned to just add an Arduino pro micro as a controller - this guy has released schematics and gerbers for many of his designs.  http://www.40percent.club/2017/03/ordering-pcb.html

His designs may be of particular interest to us, since he uses cut down keycaps and makes extremely compact keyboards.  The "Gherkin" http://www.40percent.club/2016/11/gherkin.html is a backlit 3x10 key matrix with 16mm key spacing - measuring about 5cm x 16cm.  PCBs can be had very cheap - he says an order of 5 sets of the 3 PCBs that make up this keyboard cost about $55 including shipping.

This "Gnap" keyboard   is larger and has a more standard staggered keyboard layout, but it has individual LED control, and the gerbers are on github.  It's 12x4.  It's pinned out for two arduino Pro Micros, but the schematic looks like we might just be able to wire these pcbs directly up to DIN/DOUT modules.

http://www.40percent.club/2016/10/gnap-20-plateless.html -

https://github.com/di0ib/tmk_keyboard/tree/master/keyboard/gnap/pcb

 

Here are 6x5 gamepad PCBs from another supplier  for $8 - not sure how they're wired though, and I don't think they have LEDs.  http://www.switchtop.com/product/gamepad-macropad-pcbs

 

Edited May 26, 2017 by borfo

[latigid on]

Of course, easy mode would be to use an existing PCB and MX-type switches. But those keyboards are built for a specific purpose, as are mine. What will you do with the extra four columns and two rows? Arduino-compatible, but using an SPI-controlled matrix driver. SRIO on a MIDIbox simply updates with the clock tick and requires almost no overhead. (Plus, the former doesn't fit with TK.'s plans of derivative HW that's simple to integrate.) Are the LEDs addressable or always on?

The biggest challenge is to get LEDs to illuminate the entire switch. Hopefully the PCBs come tomorrow so it will become clearer to explain, and I'll probably start a new thread for it.

MX-type have the LED as THT close to the key, or an aperture on the north edge. This is okay if you have a small glyph to illuminate, but for blank keys it looks bad. Enter Matias, where the entire key is transparent. But there's no room for LEDs, except on the underside as reverse mount. Now try to find a bicolour LED that's bright enough. I did some testing with 2x lensed 1206-ish sized parts. Maybe it could have worked, but it would've required 4 LEDs per switch. This increases the price and still isn't perfect. 

Here it is. It's not the intended use for these LEDs, but it works well. You can choose R, G or B, but the BLMs support only two of the colours. You could control the third separately, or all three with a different driver.

 

This is the route I'm gonna take, you're of course welcome to try your own methods with existing hardware.

Best,
Andy

[borfo]

I can imagine lots of uses for button layouts that aren't identical to the SEQ's 16 columns - for NG controllers in particular.  But I'm not at all suggesting that those boards are a replacement for anything anyone's doing here, just thought I'd share a link to a site with a lot of open-licensed keyboard designs with gerber files.  In particular, I thought that the work he's done on making the key spacing as small as possible might be interesting to people here. 

Re: addressable LEDs - some of his designs have per-key addressable leds.  Others have a few addressable LEDs somewhere on the keyboard, but not on each key, or backlighting LEDs on each key that can be addressed as a group, but not individually.

Edited May 27, 2017 by borfo

[latigid on]

It's true that some of these layouts could be useful for other projects; I suppose they're somewhat related.

It's also a different story doing "bespoke" builds rather than a repeatable project. For instance, the closer spacing means the caps must be cut down (or those custom clear caps, which don't look great IMO), and not everyone is prepared with the right tools.

Another thing to consider is how the button layouts work with other hardware. 19mm standard spacing is actually perfect for a 40x2 LCD.

Sorry, I'm not trying to be too grumpy. The PCBs were designed over a period of months with the whole build in mind, and the order was placed weeks ago, so it's not changing unless there's a problem with the prototypes 

[borfo]

Dude, I'm literally just saying (to nobody in particular) "These keyboard switches are neat.  Hey, look at this site with a variety of open-licensed keyboard designs with gerber files."

Nobody is criticizing you or asking you to change anything.

 

Obviously, this thread isn't the place to post anything not directly related to your PCBs though, so my apologies for the intrusion.

Edited May 27, 2017 by borfo

[latigid on]

It's okay, I'm not offended. It's just unclear from your wording whether you were suggesting "You may want tio take a look at some of the matrix keyboard..." in a general sense or for this project. A few weeks ago you were asking for extra headers to use panel-mounted switches, so my initial sense was for the latter.

Fact is they _are_ really cool hardware items, and they should definitely be incorporated into more projects. You might have seen Wilba's Cherry Jammer keyboard with a clever velocity sensing mechanism, or I think it's antichambre who was looking at reflectometric light sensors. 

No DHL today, so another few days to wait at least. :(

[timefox]

clicky keys or quiet keys? or linear? (i'm not sure what linear means in mx switch speak)

i also think this cherryMX approach to be much nicer than those rubbery button light things.

 

 

 

[latigid on]

The best thing is you can choose the keys you want:

http://matias.ca/switches/

The options like you say are click, quiet click or linear. I though I'd like the linear, but actually I prefer the quiet click. Some people enjoy clicky keys, so no worries there!

PCBs are here, unfortunately they didn't do the cutouts for the LEDs, so there will be a delay in testing. 

I also need to get some parts orders done -- getting there slowly!

[jarvis]

quite excited by the new layout, excellent work!

[latigid on]

Perhaps a few teaser pics:

Closeup of the wCore, note the USB/SD "modules" above:

 

One tricky thing was to mount an LED but not go beyond 160mm x dimension.

This was the answer:

 

There was room on the back, so this is dedicated to the guy who loves the Beat LED:

 

A bit hard to capture the mask text, but it reads:

"the big beat LED somehow belongs to a MIDIbox SEQ, it's like a defining feature :D"

 

[mongrol]

Awesome looking work Lat. You've been very busy.

[Hawkeye]

Very nice!

[CJ55]

I think, I will build a second Midibox SEQ V4(+) soon! 

I see you are using a Core407Z. Do you need a JTAG or SWD programmer (st-link v2) for it? Have you tested it already with the midibox seq firmware? I like that is much smaller than the Discovery board!

I like your USB module! I build something similar with switching between USB B and USB A (Host). How do you connect your USB module to the core?

[latigid on]

Hi,

We might have to move the discussion soon, but what the hey. 

Yes, the Waveshare boards look good and are much smaller than the Discos. There are some others which have an even better arrangement (mounting holes and top-mounted header pins), but on the face of it Waveshare look a bit more trustworthy. Still have to find time to test!

The downside is exactly that: no STLINK onboard. It could be possible to flash on UART0 (I think), but you still need a USB-UART converter.

I bought something like the following:

http://www.ebay.com/itm/ST-Link-V2-Compatible-STM8-STM32-Programmer-with-4-Spring-Loaded-Pogo-Pins-/201303925381

but if you have a Disco board it should also work to connect using the normally jumpered header pins on the ST-link side.

The USB module optimally aligns close to the Core on a 5 pin header, but it could be placed further away if needed. A main reason for moving to another carrier is that the USB MCU pins are on the header, so no more flaky microUSB! The host/slave switch is probably not proper USB spec, but should be okay for DIY purposes. The SD card sits on another PCB underneath (J16E), with the LEDs and reset button. Hopefully that's okay given the lack of power switch. I didn't put one in because it would've complicated the USB spec (should apply power prior to data pins).

[jarvis]

My main concern with the design is that, at least according to the renderings, the data wheel is not flush with the surface of the case. I would expect this to lead to accidental knob nudges when pressing the frequently used function buttons surrounding it, or at least make some of the ergonomics a little odd since you have to angle your finger around the edge of the data wheel. Is there other hardware you've modeled this layout after?

Edited June 2, 2017 by jarvis
clarification

[latigid on]

2 hours ago, jarvis said:

My main concern with the design is that, at least according to the renderings, the data wheel is not flush with the surface of the case.

The idea uses a DK-38 knob from ALBS, and I've asked Adrian to accommodate for it by cutting a hole in the front panel. So it will already be sunken down a bit. But no, it's not flush with the panel. They also have a fairly long stem, which is why the wheel on the Wilba version is "panel mounted" to the PCB.

Do you know of a datawheel knob that sits lower? Please tell me a part number and I'll have a look.

 

2 hours ago, jarvis said:

I would expect this to lead to accidental knob nudges when pressing the frequently used function buttons surrounding it,

The suggested encoders are 24ppr/24 detent, so you'd need to turn the knob 15 degrees to register a click (or is it half that with the fast encoder scanning?). Doing a bit of maths says you'd have to turn the outside (least torque required) almost 5mm.

 

2 hours ago, jarvis said:

or at least make some of the ergonomics a little odd since you have to angle your finger around the edge of the data wheel.

Unless you have tiny Trump hands, you should have more than enough room to reach all of the buttons without bumping into the encoder. The usage is not 100% clear, but it's possible that the wheel and buttons will be used together quite frequently. 

 

2 hours ago, jarvis said:

Is there other hardware you've modeled this layout after?

It's mostly original and aims to use the remaining space left over from the LCDs/OLEDs. But perhaps something like this:

 

 

[mongrol]

Here's one with no stem at all.

 

http://au.rs-online.com/web/p/potentiometer-knobs/7777283/

 

[latigid on]

Looks like a nice wheel! 

Compare with the DK-38:

http://www.albs.de/ecom/images/866004.pdf

The RS one is just 1mm shorter in total height, so I'm not sure if that would make too big a difference.

[mongrol]

Interesting article on custom keyboard building.

http://hackaday.com/2017/06/05/mitosis-anatomy-of-a-custom-keyboard/

 

[jarvis]

On 6/2/2017 at 4:09 PM, latigid on said:

The idea uses a DK-38 knob from ALBS, and I've asked Adrian to accommodate for it by cutting a hole in the front panel. So it will already be sunken down a bit. But no, it's not flush with the panel. They also have a fairly long stem, which is why the wheel on the Wilba version is "panel mounted" to the PCB.

Do you know of a datawheel knob that sits lower? Please tell me a part number and I'll have a look.

I'm not worried so much about the encoder values being bumped, its more about ergonomics. The USB jog wheel you shared was nice, but it has rounded edges on the jog wheel as well as a contoured surface. Combined, these features give more room for finger placement. Now imagine that same layout with a square-edged knob and flat control surface - its not a negligible loss of space.

I don't know of any better solutions except a flat data wheel (I got one at Adafruit but it has a long stem - I'll have to scrounge up the part number) and/or more space between the jog wheel and buttons. I'm by no means speaking from a position of experience, it's just something I noticed that might be worth checking out.

[latigid on]

For those playing at home, there's a delay on Matias switches, apparently as they are producing a newer design with "reduced key wobble." I didn't think the wobble was too bad, but there you go. The restock is due at the end of August. In the meantime I will try to test things without switches and also look at an order of keycaps.

[latigid on]

 

LEDs are nice and bright; this is from the rear side! I'll need some help/fiddling with the switch portion as this is not a 1:1 BLM.

 

I noticed a similar concept from 40%club:
http://www.40percent.club/2017/06/alps-48-rgb.html

I considered mounting 5050 LEDs in this way last year, but IMO the available ones are not really bright enough. As well, I'm not sure how reliable a "bridge-soldered" LED would be in the long run after a lot of keypresses.