With a successful 2nd Kickstarter – The Spectrum Next will have between 8,000 and 9,000 users.
Lets Dream a little and imagine a Bright world where all the users have a Super LED Blinkenator 2000 installed….
9000 users = nearly 40,000 inserts to be made!.
lets say just 10% want the blinkenator, I still have to make nearly 1000 of the things.
I’ve been researching a little and identifying bottlenecks to SUCCESSFULLY produce and deliver my board in those quantities
There’s some scary numbers!
So, I’m now pressing forward with TWO designs. one design, the one you’re all familiar with, suitable for small time production in small batches here and there on my weekends, only ever endeavouring to sell maybe a 150 units ever
and the second, a ‘mass produced’ item that requires minimal ‘hands on’ time from me to deliver, but will require some significant outlay up front.
The pictures above are a first run result of my Design For Manufacture for the inserts….A different injection mould, possibly 2 parts, maybe 1 and using a flexible PCB!
some key notes……..
Advantage – no connector soldering needed on my part – currently I’m soldering 16 cheap ‘bridges’ to each main board. with this insert, someone will be soldering 8 SMT FPC style connectors
Advantage – it’s likely that this design will be easier to make ‘injection moulding’ manufacturable. the existing design is tricky, but not impossible
Advantage – FPC connectors are a bit more reliable and easier to use than my bridges for the end user
Advantage – FPC / flexible PCB ‘legs’ on the inserts will mean a little bit easier installation by the end user
Advantage – Uniformity of Light – This type of construction allows for a much thicker ‘top layer’ – which will diffuse the light far more. Also, more of the insert will be better lit up ‘from below’ rather than from the side that i’m currently doing.
Disadvantage – FPC connectors are more expensive
Disadvantage – Flexible PCB’s are more fragile
Disadvantage – Flexible PCB’s are more expensive than FR4 for small quantities, so prototyping ability is very limited. at The quantities I need though, there’s not that much difference
There’s more i’m sure, once the final numbers are ready, I can see if a kickstarter makes sense, it may not be financially viable if the whole thing needs to be sold at £80 each……
if I can get closer to that £50 mark, then who knows!
I’ve been updating the Dev group on facebook more regularly than here
Progress has been slow but constant!, the new Jumper method of getting the LED inserts to connect to the controller works well, if a little fiddly. I think there’s some changes I can make to allow for an easier installation experience.
A big milestone also – The BETA hardware is at such a point now that i’m happy to send it to the core Dev team for actual installation inside a Next….err, except they can’t have the bottoms on as the USB cable doesn’t fit, D’oh!, another re-design needed!
AND – software – My Arduino code’s finally quite stable – Also, from the Next side of things – the i2c code is great – it runs well at 14MHZ, allowing for some interesting sequences on 8 segments…..I’ll start uploading BASIC programs in the next month or two.
Also, a kind of fork in the road….
Throughout this project, I’ve had an end goal of maybe 5-10% of Next owners owning a Blinkenator. at 3000 Nexts, that’d be maybe 150-300 devices sold over a year or two, making my beer money fund quite happy
Things recently changed……and have made me realise that I’ll probably need to step up my game a little…..
Means that now, there’s over 8000 Nexts in the wild!.
Assuming the same targets, I’d now need to manufacture between 400 and 800 devices…
May not sound much – but at a top level, for just 800 units…….that means some big numbers…..
sourcing 3,200 Plastic inserts….
Sourcing 26,000 LED’s
and with big numbers comes Big Money….and long lead times.
IF someone landed me with an order for 800 Blinkenators tomorrow, at (say) 45 minutes per board, I’d need 600 hours to complete the order.
I have a day job that demands my attention for 160 hours a month. Wife and kids that demand me for a further 80 hours a month…then there’s the whole sleeping and eating thing..
It’d take me a year to be able to fulfil that order 😛
So, the fork in the road……….I may need to do my own Kickstarter!
I’m investigating larger scale manufacture – Full PCBA including through hole, better DFM and Plastic Injection moulding.
All that costs big up front ££…..hence the Kickstarter………is my 5-10% adoption figure massively optimistic. Is it woefully inadequate?
To have any chance at a successful Kickstarter, I need to turn this hobbyist , good quality (7/10, could do better) project into a slicker experience, a better presented finish and professionally produced, not at my dining room table package that would obtain a Crash Smash award, a solid 9.5/10 experience. I KNOW I’m capable of creating the hardware (i’ll learn the software). I’m genuinely uncertain at this time if I would be able to DELIVER that package.
Saying that, I know my limitations, I have a grasp of the fundamentals and i’m costed to the penny for small batches.
Extrapolating that upwards and figuring out where costs stand for different adoption rates is my focus now the BETA 1 boards are ready.
If 30% of Next owners buy this thing, that’s 1800 hours of ‘work’ to do. That’s a FULL TIME JOB!!
scary isn’t it. I have to create budgets that allow for an employee!!
A funny story about multi sourcing components and the importance of testing before shipping!
I used a supplier on Aliexpress to purchase a few thousand switches in a few orders over a few months but their prices went up quite drastically after the last order (doubled!!) they weren’t the cheapest to start with but were reliable and friendly, worth the extra ££
I found another supplier who did a good deal for a full bag of 4000! Ordered them and waited, very quick delivery and friendly also (will buy again!)
I built my first test new keyboard with the new PCB and switches
It didn’t work. Well, actually, it did! Work perfectly…but in reverse :-p …..
If you mashed every key simultaneously then only released the key you want to press….it worked!! Yeah, the supplier sent me 4000 ‘inverted’ switches! My fault for not checking prior to ordering, they ‘look the same’ so ‘must be the same’ was a wrong assumption on my part! (At least they all weren’t the shift lock type!!)
It’s a VERY easy fix though (found after several panicked hours of testing and building Keyboards)…rotate the switch 180 degrees and it’s perfect!
In each kit I’ve included a small errata note and list of basic instructions to help. It’s an annoyance but for you guys it really just means the silk screen doesn’t quite match the switch orientation so just ask first. Look at the pictures and of any doubt, email/messenger/twitter/Reddit me 🙂
Some quick steps right now – photos to follow.. Suggest have two tabs open, this one and the other PICTURES tab for reference
Some videos are up on youtube also
SUMMARY- SOLDER PARTS ONLY IN THIS ORDER
Cut one leg shorter on the diodes – Use scissors . About 1-1.5cm is good
bend the short leg side to a right angle
Note the orientation of the diode – The F Key diodes have a diode picture on them. The white bar matches the location of the black bar on the diode.
put diode in holes and bend slightly to lock in
repeat for all diodes
Solder all diodes
clip the excess legs back
you have a few spare diodes so don’t be afraid to experiment on one or two to get the right bend / fit
Probably best to solder these in now before you forget
I’ve found it useful to PLACE the arduino on the headers (DO NOT SOLDER YET) so it keeps the headers parallel
Make sure the black part of the headers is on the underside of the PCB
Solder one pin of each header
SWITCHES – STEP 1, JUST TACKING IN PLACE
Pay attention to orientation
don’t worry about straightening the switches at this stage, the goal is to just ‘tack’ them in with a single solder blob to hold them in place. They can be wonky, it doesn’t matter.
DO NOT SOLDER MORE THAN 1 PIN OF EACH SWITCH IN ONE GO
The switches are easily heat damaged – they become ‘sticky’ and no longer move smoothly if the plastic is melted due to excessive heat. During the entire soldering procedure for the switches, do ONE leg, move to the next switch. when all are done, move back to the first switch and repeat.
I’ve damaged only 2 switches this way soldering the prototypes but it can happen if you’re not careful
Note that the white part of each switch is asymetrical. One side has a ‘dip’ / inset which guides the switch up and down. the other side is smooth
there’s a marking on the PCB to represent this dip / inset.
ALL switches go the same way
Get a sheet of paper
Insert the top row of switches into the PCB
Place PCB on sheet of paper and fold paper over the top, tightly
flip the PCB over
hopefully all the switches stay in place
Solder just ONE leg of each switch – any one – say the top right
Repeat for Row 2
DO NOT FORGET TO SOLDER THE ARDUINO HEADERS IN PLACE
Repeat for Row 3
DO NOT FORGET TO SOLDER THE ARDUINO HEADERS IN PLACE
Repeat for for row 4
(Hopefully you didn’t forget to solder the Arduino headers in place?)
and finally the space bar
SWITCHES – STEP 2, Straightening
This is probably the most important step to getting a good looking keyboard with all the switches aligned. Spend some time getting this right, you have a handful of ‘spare’ switches so now’s the time to make mistakes and fix them whilst there’s only a single solder blob on them
I’ll post a few videos shortly but there’s a technique.
Hold the board in the air
Use your index finger to push in, and slightly down on each switch whilst soldering the previous blob. The goal is to move the whole switch slightly so that it’s slightly at the top, or the bottom of its footprint.
when you melt the solder whilst pushing in and down, the switch will move slightly, sometimes you’ll hear a little click or snap as the solder melts
repeat this for each switch, pushing in and down slightly – when you look at the final position, there’ll be some of the pad visible at the top of each switch
NOW IS THE TIME TO TEST EACH SWITCH FOR SMOOTH MOVEMENT
of the 5 keyboards i’ve soldered, I’ve had two defective switches, this is partly the reason why there’s a few extras in the kit
of the 5 keyboards i’ve soldered, I’ve broken 3 switches by either over-heating, or trying to remove after putting them in backwards. unless you’ve got a hot air gun, they’re tricky to remove intact, hence check NOW whilst there’s only one solder blob!
When you get close to one side of the keyboard, you’ll have to fiddle a bit to keep pushing the switches in the same direction. I’ve found that changing technique a little and ‘flip’ the board lengthwise works. hold the board against yourself and use your thumb to pull the switch down instead of push
repeat the alignment technique for ALL switches!
SWITCHES – STEP 3, Final soldering
This is the easy / relaxing bit!
DO NOT SOLDER MORE THAN ONE LEG OF EACH SWITCH AT A TIME
do it by rows, clusters, however works for you, but here’s what worked for me
Solder ONE pad of each switch, then move to the next
once all switches are done, start from the beginning
Solder another pad, etc etc
A SMALL CHEAT – You only actually need to solder 3 points. Two on the ‘bottom’ of the switch – these are the electrical contacts. ONE on the ‘top’ – this is for mechanical stability. As you look at the keyboard, the bottom two pins are the important electrical ones. Pick any on the top
on my prototype, I found soldering all 6 pins tiring, so on my second version I just soldered 3 and it worked perfect. Up to you, but DONT SOLDER MORE THAN 1 PIN AT A TIME
Note the orientation of the Arduino by the Small USB socket and a mark on the PCB. Also the silk screen on the PCB will match the letters on the Arduino.
these need a little more heat to solder to the pins
Back a couple of years ago, I lucked out on a facebook marketplace post. Someone was selling an unassembled 3D printer kit, that was tatty and had missing parts………Woo!
I did a 30 mile detour on my voyage home from Wales and grabbed it. excitedly pulling all the bits out of the oversized, filthy box and….Putting them into a rather smaller, cleaner box for a couple of years whilst not doing much with it.
Well, today, that £40 investment paid off.
by hacking a chunk off the 8mm threaded rod, using a few bolts and washers, and a quick mod using a spare Hama-beads square that we had lying around, I was able to fix the broken hamster wheel!
or, in other words, I’ve massively over-engineered a fix by butchering a 3D printer (kit).
On the plus side, I’ve a happy hamster AND saved £2.87 by not needing to buy a new one!
Also, It’s gotten me thinking….HAMSTER STATISTICS.
Scroll down to 19884 – See just what processors are sitting in the same area – what their retail price was! – this processor sits at exactly the same position as the intel i7-10700KF
A 125W DESKTOP processor that costs £400 on its own on ebay right now. Add all the other stuff needed to make a fully functional PC and you’re getting close to a GRAND on a DESKTOP!
In summary – I’m besotted….This laptop’s happily rendering in REAL TIME what my old laptop took tens of minutes in Fusion360
if you’ve been holding off buying a laptop, you really can’t go wrong with the Renoir series of processors.
and, for games – seems to run Fortnight,Starcraft 2, and a few others just fine at maximum settings! performance apparently is about the same as the Geforce MX250 which isn’t sloppy considering this is integrated graphics and has just a 25W TDP when under extreme performance. The MX250 graphics card has a 25W TDP on its own for this benchmark!
Fawning over, normal service will resume shortly 🙂 – in the meanwhile, ask away any questions, happy to answer them
I Purchased 10 controller boards with the SMT components ready assembled There’s a few small bugs……But, that’s what prototyping is for.
First major annoyance – I’d goofed and left VCC on the arduino as 3.3v in the schematic. so, the board wouldn’t power up inside a next. Quick fix is to short RAW – to the VCC pin on the Arduino
(RAW is a 5V OUTPUT when plugged into USB, or 5 and a bit Volts INPUT to power the Arduino)
This has the potential downside of back-powering the Next via USB via the 5V Line when it’s sitting inside a Next and someone uploads a sketch
Another change needed – The Arduino’s USB port fouls the case when fitted inside. With a USB cable in, the lower part of the keyboard inlay blocks the port. – it’s ‘just’ about bodgeable however if you really wedge it in there. But, not ideal So, i’ve tried with soldering the arduino ‘upside down’ on the wrong side of the board – that seems to work. The board’s mounted just far enough ‘up’ into the case that a USB cable can sit under it.
For the production version i’ll re-arrange the board completely
Another further change –
Removing the RAW pin entirely from the Arduino and shorting the Next 5V directly to the VCC pin on the Arduino
That’s about it for these pictures,
From top to bottom –
Original board – has a resistor bodged in and the RED led installed on the wrong side. Also has my ‘impossible to solder straight’ PCB fingers.
Third board – upside down mounted arduino with missing RAW pin
Second board – my ‘go to’ working one right now – also has the first run of my ‘quick fit’ connectors……i’ve done about 20 cycles of inserting now and they’re still working!
Due to my previous goof-up of having made the old LED insterts ‘backwards’, I designed new ones!
They’ve arrived and are all round much better
Slimmer – just 2.2mm wide
Here’s what they look like – the features are a bit too small for JLCPCB to handle on their mass production – their mill uses a 1mm bit – guidelines are minimum of 3mm between milled out slots
I’ve ordered 50 of these – should be plenty to fill up the 10 prototype control boards i’ve gotten made and leave a few over for experimentation / errors
There’s a small problem however with mousebites this small – each board takes a good five minutes or so of dremmeling to get ready!. I can do about 7 boards to a charge of my battery powered cutting tool….I’ve ordered 50 prototype PCB’s so that’s lots of minutes of work ahead! good thing we’ve a handful of N95 facemasks picked up from back in March just as the world went loopy!
Here they are in all their glory, powered up the RIGHT way round on a controller PCB.
I hand soldered all 4 of them using some new (expensive) solder paste,d 3 worked first time! the last one had two LED’s mis-aligned…Quite a good improvement!
Each insert so far on the prototype is taking easily 20 minutes from receipt of parts to working device. Each new controller PCB is taking over an hour of soldering and wire cutting for the quick contacts. That should come down quite a bit once i’ve figured out a process.
Yeah, spot the mistake 😛 Still, quite chuffed – these are powered up and working INSIDE the next! –
There’s no control YET – this is just a single routine that runs upon powerup in the Arduino….
And, just as a final superb moment – look at these results of a quick i2c Scan (there’s a lot of numbers due to using an HDMI monitor)….That 3rd Device at address 0x45 – That’s the Spectrum Next Super LED Blinkenator 2000 just waiting for a .DOT command to control it 🙂
Now waiting on enough Arduino Pro micros and switches to start making kits up!
Each DIY kit will probably contain the following – i’ll firm up with pictures once i’m done test populating a rev2 board
70 Standard switches
70 Diodes – Through hole (possible SMT option also depending on price)
1 USB Hub
1 PCB – Rev 2 or later
1 USB cable
two small pieces of heatshrink tubing
a couple of pieces of wire
1 Arduino Pro micro – Pre-programmed with QMK firmware and custom Keymaps
a set of FDM – Filament printed keyswitches – These probably won’t be ‘perfect’ so i’ll be chucking them in as effective freebies as I won’t be releasing the keycaps as a digital file.
About that last part – I’ve spent countless hours on creating these keycaps, and still have more to go. I’ll eventually release them as a Digital file, but for now, you’ll be able to at least use the freebies to see if new keycaps are for you.
If you wanted a professionally printed set, i’ll be arranging something with a printing bureau somehow… It’s also likely i’ll be able to source reasonably costed SLA resin prints of these…watch this space
And for the money shots…….I’ve finished the top row of key text!
Now I’ve gotten the first row done, the next three should be significantly quicker.
The text is recessed into the key by about 0.4-0.6mm – between 2 and 6 layers of 3D print, not really enough to be clearly felt – but enough to be ‘seen’
After that, there’s the optimisation for printing – Filleting the edges – trialling depths and generally finding out what actually works, looks and feels good
to use some PCB grounding spring contacts to provide a quick fit connection didn’t really pan out – the contacts simply didn’t solder on easily, too difficult to align correctly and quite weak – I tore a few pads off trying to get them aligned and correctly ‘grippy’ on the insert.
multiply that by 16 each board – the first one took me about 2 hours to get to be in an ‘ok’ state – Not really acceptable for a mass production product – not that this’ll be mass production but I’d rather not spend half a day on each of these getting them ready for sale……..
The second slight issue – See the photo below
The PCB is laid out on top the next board in the position it’ll be installed in.
There’s a prize for someone that spots the goof-up
Have a further look at the PCB powered up………..
Yeah, I got the inserts ‘back to front’ – That’s the result of working on a bottom mount PCB from the bottom…….
There’s two ways I can fix this
Simply rotate each LED by 180 degrees on the PCB and install a bodge wire to swap the input and outputs around………
I can simply re-design the LED insert and improve upon it!
a few reasons to re-design,
The first 6 LED version still has a bit of point brightness – I fear that even with the SLA printed inserts it won’t be diffuse enough..
8 LED’s – This should spread out the light more, reducing the hotspots a bit
0.4mm Slimmer – this lowers the LED’s further into the case, allowing much more plastic to sit above. I’m hoping this will de-focus the light more
Reversed connections to match the reversed controller board!. reversed is the new non-reversed now 🙂
New insert PCB’s are on order and should be here in a couple of weeks, I’ve bitten the bullet and ordered FIFTY….Also a few hundred more LED’s and a large tube of solderpaste.
Doing these first 10 dev boards is going to be fun – 320 1.5mm LED’s to be hand soldered!
The Controller board fits inside the case almost perfectly
I’ve slightly offset the J15 connector on my PCB to the one on the next. This offset gives a lot of friction, but needs some long term testing – the standard header I installed on one of my boards seems to work well as a friction fit. BUT, i’m not convinced that 32 LED’s, each pulling 5-20mA, (depending on which datasheet I refer to) – or between 160-800mA total depending on how I end up setting the brightness…..800mA is a LOT to pull
I’ve purchased a new gadget – A Riden RD6006 Benchtop ‘power supply’ so once i’m set up, i’ll charaterise the LED’s and current draw to set the software limits appropriatley
I doubt i’ll take nearly an amp on this board 😛
One further small mistake on the dev board –
When originally designing this controller board, I was to use a 3.3v Arduino to make it compatible with the Next’s 3v3 i2c.
For a few reasons, I’ve changed to using a 5V arduino and putting on-board a level translator device – this gives a 5V buffered i2c output that anyone can easily plug into
I forgot to change the net names….The board still ‘thinks’ the Arduino is either powered from RAW (it’s ‘unregulated input’) or 3V3 VCC…
The RAW input drops a few volts through a voltage regulator on the arduino to give the arduino a nice regulated 5V.
The next output is 5V….it’s not enough to power the Arduino through the RAW pin…
Took me quite a while – and a bit of soldering hackery to figure that one out as the speccy picked the board up perfectly when patch wired in place..
USB powered, it works perfectly
in the Next it doesnt….
The fix – I think I can just short the RAW pin to the unconnected VCC pin on this first batch –
See that i2c device, found at address 0x45………..That’s the Blinkenlight 2000 PCB :-), alive and inside the Next!!!
Great, I could shave down the underside of the insert and then I could shove a bunch underneath the insert –
The Pixel density isn’t so good 😦
the Inside radius of the arc of the insert is about 17.1mm…see picture below…
That gives us a total circumference of about 108mm for the circle. hat gives us about 27mm for a 90 degree sweep of the arc. There’s actually a bit less than 90 degrees, so lets say that we have 25mm of length along the arc where the white dots are.
25mm of length and LED’s spaced 6.3mm apart only allows us 3 LED’s along the arc……and based on previous experiments, that won’t be enough to give a nice looking line of light……
I’ve messaged the manufacturer to see if they can do anything with more LED’s – at least double, maybe 8 for the 25mm length about 1600 per meter ;-)…who knows!
So, onto another idea……maybe a small PCB?
Looks like I can probably fit a bunch of 0603 LED’s inside!
To the left is a bit of a trial to see if I can maybe get a Flexi-PCB in alongside the Insert….and another trial spacing to see if I can maybe get some degree of control of the LED’s – instead of a long block of 6 or 8 LED’s in parallel, why not 3 or 4 pairs of LED’s?
That way I can have a winking left/right status and a solid status…you could get some very basic animations going on all 4 LED’s split into 2 segments each – 8 individual lights…
The pipe with the new 53mm straight lengths comes a little short of the planned box enclosure end……
I’d calculated the ‘missing’ 57.5mm length based upon the centre path of the diameter having a diameter of 95mm……
The pipe itself has a diameter of 85mm
So, the actual pipe itself will need an additional 42.5mm either side of the path it follows………..
Which means, a new sketch………Something like this.
R47.5 is the 95mm diameter of the centre of the pipe
R90 is the outer diameter of the 85mm diameter pipe
10 is the thickness of the speaker enclosure
Finally, we have at least one outer enclosure dimension sorted! – 200mm – Err, that’s a bit bigger than planned! – The speaker itself is only 98.2mm diameter……Looks like some more iterating needed, BUT, now I understand the dimensions, I can experiment a little with the Tube’s cross sectional area and therfore diameter…….
In hindsight, the speaker enclosure dimension was fairly obvious…….
enclosure thickness ( 10mm ) +
Transmission Line Diameter ( 85mm ) +
enclosure thickness ( 10mm ) +
Transmission Line Diameter ( 85mm ) +
enclosure thickness ( 10mm ) +
The smallest enclosure possible has a Transmission Line area the same as the speaker – i.e.