So, my trusty cordless drill last battery just died. It was not bad, but also, not a brand name model, so I just could not get new batteries for it. BUT, I do have spare Li-Po batteries, so, it’s brain swap time!
First, remove the old 18650 batteries and faulty electronics and connect the Li-Po instead.
Add black hot glue for that professional look:
Some rotary sanding will take care of the tight fit in the handle:
But the battery will still not fit all the way, so an extension is modeled to be attached to the old clip-latch endcap. To model it, the angle is extracted with a curved calliper:
And the shape is photograped and imported into Fusion360:
Finally, add some bling:
Print and fit:
Some more black hot glue (oozing to be cut after it’s cold. Also both surfaces were rough, so adhesion should be excellent):
Photo Finish!
Surely it’s more unwieldy than the original, but it’s better than having a useless battery dead tool, right?
Have you ever done a ride so long, your right hand is left useless? For those gaudy fucks who have cruise control (real cruise control) I salute you with my erect middle finger. For the rest of us mortals who have a bike with no lavish fly-by-wire throttles, there are other solutions.
You have fancy ones:
Less flashy ones:
And things you should not even use to clamp a sub’s dick to a wall:
But, and it’s a big BUT, I have realized they are all wrong, because, you see, they put the control in the same hand that you want to block. What you really want is to lock/unlock with the left hand, leaving the right alone to do it’s speedy-speedy/stoppy-stoppy thing. And no, using your left hand to hold the throttle is definitely stupid beyond even the shittiest throttle holds. /transfers over the connection and SLAPS YOU for thinking that/
BUT, how, you might be asking, can one accomplish that? (you are probably asking why, instead, but I don’t give a fuck.)
Well, my CB500x has a continuous tube handlebar, wich means I can run a bowden tube through all it’s lenght and connect something on the left side, to something else on the right. Easy.
A bit of drawing later:
So, at first I tried to just cable actuate a similar screw-clutch type than the more streamlined locks, but that proven complicated by requiring both a 90º vector turn in the cable with no micro-pulleys avaliable, and too much lenght of travel cable-wise.
A printed test model:
I must admit It did work to an extent, but that was not the bestest way by any means. I knew it was a failure well before even finishing the current print.
Meanwhile I was waiting, I happened to discover that the throttle handle body, at least in my particular model, happens to go beyond the end of the metal tube that holds it. I could definitely use a drum style brake in there!
Once I had tested, and failed, with the first dumb idea, I dived on the second one.
Since I needed measurements from the various internal diameters, I decided to bite the bullet and dissasemble the counterweights in my own bike. You can’t imagine my face when I saw that this was inside the handlebar:
I had always assumed the thing was just the outher portion you could see, attached with some kind of inclined plane nut (like the first model at the top of this post), but nooo, it had to be massive and hidden away and act as a surprise when I decided to overengineer a stupid thing in it’s place.
Oh well… ¯\_(ツ)_/¯ …in any case, that is just an elongated nut to add mass, back to the studio:
Damn, this thing looks like a German WWII grenade.
So, this is the drum brake thing implemented on the right, throttle brake side:
The central plunger separates both halves that contact the white nylon drum in the throttle grip, braking it.
But, you might ask, does it actually work?
Well shit, yes it does, shut the fuck up already.
Here’s a video of the brake shoes moving (white piece is teflon for maximum slipperness):
The rubber band is just to hold them. They can’t fall once installed. I did implement a groove to leave the rubber permanently in there, but that would also reduce the braking surface, and as said, once installed, they won’t go anywhere. (Think of it like some bearing retainers in those flexible clutch/brake levers that can fall down when you work on them, but won’t once finally installed).
On the controller side, I really want to look both unconspicuous AND simmetric to the other side, so I made the effort of operating the cable pull using the existing sleeve coupled to a rotary inclined plane:
That inclined plane piece links to the outside grip cover with a passing headless screw.
Should you where been paying attention, you might be asking by now: “What is that weird groove in the bottom of the cross-sectioned piece?”
This one, you unattentive bitch.
Well, as I told before, this is a bowden driven mechanism, that means that somewhere in the piece there is a cable and a sleeve that run along the thing. But this also has an angled nut to retain itself in the handlebar tube. Unfortunately, those types of nuts have the bad habit of twisting around until they are fully locked, and that would put the bowden in immediate risk of shear.
To prevent that, an antirotation pin was added, and it’s seat in the nut elongated, so it can move up and down when tightened/loosened. Said pin also doubles as axle for the brake pads to pivot.
Here’s a frontal view of the bowden/pin interaction:
When the screw is loose, the amount of play built in the front piece means that the nut top groove for the bowden tube can’t go above the shear line. As you tighten it, the nut can only go down (in the graphic) and also can’t rotate, keeping the bowden safe at all times. The pin would act as downward limit for the nut, but you should not be using this in a tube whose diameter is so much different than the nominal diameter of the piece.
Pretty nifty if you ask me.
Since I was at it, I decided to go all fancy out and make the parts to be the same for both the left and right sides for TWO reasons:
Why not?
I can.
Surprise, motherfucker!
Actually at manufacturing is easier if I have to make more of only one model, albeit slightly complex overall, instead of different models. That can streamline the machining process a lot.
As for holding the outher grip pieces in place, I made simple ring clips and recesses on both sides of the piece:
It’s a shame I can’t use steel wire that looks like enamelled copper, the color contrast would be wonderful. With a copper ring the sleeve stays in piece really well, but still can’t risk it in road conditions.
Unfortunately, the cable puller piece has to be machined out of metal even for this test.
The pressure required to hold the cable with that little vertical M2 screw and withstand the pin sliding in the inclined plane would be too great for the PLA to hold. That will have to wait for the next chapter of the adventure.
For now I’ll leave you with a size comparison between the metal counterweight and the 3D printed model.
No, I didn’t print the longer nut because that would be a waste of PLA and it doesn’t add anything to the mechanical requirements for a test.
Meanwhile distracting myself from the lack of components due to the world pandemic, I started working on a novel approach to cable driven joints for robots.
One thing in particular is that it requires a continuous loop of steel cable (no fancy dineema string avaliable for tests either), but achieving that is no easy task. The device can’t handle knots in the cable because it loops over itself to allow more grip in the pulley. Given those requirements, and current material avaliability, I decided to try butt welding the ends of the 0,45mm cable I have around.
To do that, I definitely needed more hands than the ones that mother nature decided I should have. So, I designed a support that would hold the cable, bottom copper anvil for spot welding and would support it under a digital microscope to be able to see what the hell I’m doing, this isn’t something you do without magnification.
None of the features that make it mass efficient/lighter where added other than to make it pretty, because, why not?.
Printed it at 0,1mm layer, 0,4mm nozzle, 100% infill.
Cables in the photo are NOT soldered yet, just as a measure of scale and visual guide.
Detail from the holder:
Now. How am I supposeed to weld this, actually?
This is not a huge amount of mass to be welded, so the amount of energy is small compared to usual spot welder requirements. For a trial, I used my bench power supply, set at 20V and 3.5A (max current), and the discharge it can produce before the CC kicks in. Didn’t even bother into shaping the bottom anvil, so it would embrace the cables together. As top anvil I just bent a sheet of copper to use the small radius as contact point. Also added a smidge of soldering flux paste, because it couldn’t hurt, I guess.
I got totally surprised by this:
It’s not the best weld in the world, and after inspection, a single wire that had bent, was totally not soldered, but as first tests go, it was surprisingly good.
So, after all this, how was the weld? I’m not afraid to say fine*. Quickly tied the ends of the test in a loop and pulled with my fingers until my skin started to hurt a little(no force gauge avaliable, sorry), and this thing didn’t break.
Also rolled it on 10mm pulley bearings (9mm diameter bend) AND looped it over itself on a bigger pulley while mantaining tension. It struggled slightly on the underpass, but didn’t break either. Definitely good enough for tests on my new robot joint Drive.
Apart from that, the plastic covering the weld should be taken care off, either by adding a drop of resin in there, so it sits flush against everything else, or, the best solution, knowing the plastic, making a small injection mold to reform the coating around the weld.
BUT…
One step at a time, for now, I can practice more, and when the welds are nice, worry about other details.
*Of course not an ideal weld, but come on, I’m improvising here. Now all this requires is practice and testing for a while, but I’m sure I’ll get proper cable welds in no time.
Until the next time, techno-cowboys!
————— UPDATE —————
After a bit of practice, I am starting to get “full strenght” welds. They’re not pretty, but I can’t break them, my skin hurts too much with the force applied:
So, I have a long therm project that will require a display with some particular needs.
I want it to work with 3V3, retro-looking and curved.
Yup, nothing off the shelf is going to work for me (flexy o-led displays with a big aspect ratio are not going to be a thing in time, and they don’t exactly look the way I want the thing to look anyways.
We could say I want to inspire the display in these:
The gorgeous HDSP-2000’s wich are nigh impossible to obtain nowadays at good pricings (and besides, they are small for my application). But the amber dot matrix look is just too gorgeous to let it pass, so I started drawing my concept:
8×8 groups of leds, using a SN74LVC244 line driver as row selector, and 74HC595’s shift registers to select the columns using a clock signal (and a start pulse). The thing is, when I started this, I was not sure I could program a microcontroller well enough to actually use such a display (for some people it would be awalk in the park, but to me, programming is always a chore I don’t particularly like to do). Of course, since this post doesn’t have a doomed kind of tittle, and I’m actually writing this post, is fairly evident I succeeded.
Oh, sorry, spoiler alert. XD
Anyways, so I just put myself to Altium the shit out of this thing, one schematic at a time:
Then PCB it:
Damn altium. XD
For a moment, I was also afraid I would have to chase every individual led, however Altium was kind enough to group them in couples of 8 leds:
At first I just put them horizontally, altough I knew I was going to rearrange them to be as compact as possible (1206 led footprint):
90º rotation and compaction afterwards:
The other componentry:
It’s not like it’s a particularly difficult board, but I spent an insane amount of time compacting it for a 2 layer board. (remember, this is a prototype to test my programming skills (and secondarily. that the circuit worked).
8 row input lines, a clock input and traveling pulse input.
And, since I was at it…made it BLACK:
Nope, there was no way of avoiding those three jumpers without heavy board remodelling, and I was tired of it already. Two weeks later:
First, a bit of testing:
Once I was satisfied I assembled the rest, and lo and behold!
Consumption is moderate at 40-ishmA, as per design.
However, not everything was perfect in the land of the led. Previous tests where only lighting one led at a time sequentially, and that is innefficient both because I designed a parallel load, AND because reduced overall light intensity per led. Given that I was already working at very low levels (5/8mA per led) reducing it even more was not going to help with visibility. When I tried to start doing more, this happened:
That line was supposed to be one led thick, but somehow, the line was doubling horizontally.
Hummmm…
Quick check with the scope:
LMAOOOOOOOOO, no wonder the leds where doubling, I had made the software in such a way that it put the (blue) led activation, in between column pulses. XDDDD
Some softwareing later:
Nice and cozy!
Beautiful pulse-within-pulse (blue led activation, yellow, column selection).
tl:dr
Got a good job, learned to survive in Arduino IDE on STM32, Opened a Discord chatroom about Semiconductors and bought a workshop bigger than my old home.
—— Work ——
So I got into a research and developement position on a trucking company of sorts. At first I was supposed to just repair messed up things from the monkeys they have as employees. However it quickly became a position where I have to basically supervise manual mechanical fabrication AND develop control circuitry and whatnot to drive it.
I can’t really show you what I’m doing because altough I haven’t actually signed an NDA, it would just tell the world what my boss wants to do, so better not to blow the whistle, right?
 —— Coding ——
Funnily enough, just before I got called from the job I do now, I had decided I wanted to give a try at coding again. However, I felt that the PIC16F873 I was using, was long past it’s prime time and better coding IDE’s and microcontrollers should be avaliable.
After inquiring on twitter, it was clear that STM32 was the new way to go. However, as my coding has always sucked, I asked if Arduino IDE (wich I had in mind it was some sort of block coding like LEGO mindstorms) and everyone told me it was a good approach with my limitations. I could always jump to something else more complex if I outgrew that.
So I am happy to say I do defend myself quite well on Arduino IDE. It makes for certain things to be soooooooooo damn easy (I’m looking at you, I2C! ¬¬ ) I feel very empowered. With the knowledge I carried from PIC and Microchip IDE, I found very easy to do the things that where just so damn complex before. Now I can reach serial comunication, wireless comunication, and quickly program things I only dreamt before (I’m still not a genius coder, do not expect that).
—— Discord ——
So, after Twitter showed it’s limitations as chatroom, I was shown TheDiscworld Discord. I quickly grew fond of it, but my semiconductors interest was limited by the control others had of their chats, and with that, I decided to have my own playground to share that: https://discord.gg/a7kkUUr
You are welcome to join, but remember, it’s extremely semiconductor specific. There are other Discords for more general chat. (not that that is forbidden, just that the topic is very particular)
—— Workshop ——
TOP: From the middle to the entrance.
BOTTOM: From the center to the back:
Long workshop is long.
Measuring about 25m in lenght (yes, that’s fucking long) by about 3m wide. (with width variations along the rigth lenght, but one flat-ish wall to the left. That gives 78sqm of my own personal space, but the best part is that it has installed a mezzanine of about 40sqm more.
AND I always have wanted a workshop with a mezzanine.
I do have literally TONS of work to do. The whole floor has to be removed in it’s entirety, because it’s crooked and just bad quality. One non-main wall has to be toroughly repaired (brought back to vertical and clamped) and a few repairs here and there, wich will make a lenghty process before I can populate the space with tools.
BUT…
It’s my own real workshop. Bigger than many makerspaces, and it only has to hold a single person, wich is a plus. XD
And that’s pretty much it. The workshop renovation will be thoroughly documented, so expect updates on that from time to time.
So, one day I thought I felt that my helmet had become noisier. Weird, because I have a deflector on my windsheld AND I’m quite short. When I stopped to refuel I realized that the butt bluetooth plug had fallen off:
I thought that replacements might be avaliable, but having a 3D printer and skills, it was just easier to take some measurements and make a new one.
That turned into a downward spiral of creativity I hadn’t planned at all…
The thing is, I have an internal bluetooth (Just Speak) because I hate the bulkyness of the external ones. The only drawback is that for charging I have to fully open the helmet and remove a foam flap to access the microUSB port, wich is slightly annoying.
As you can guess, while working on the plug I wondered…”Can I add a microUSB port to the plug and connect it to the inside of the helmet?”
Well…it seems I can.
So I proceeded to design a flexible PCB to connect it to the inside.
A few weeks later:
This was my first flexy pcb, so I messed slightly. Especially by not optimizing the connector position (vertical instead of horizontal) so the pcb was more expensive that needed. It also looked a bit weird, with a forced bend and such.
In any case, the thing worked, but then I realized that I would not be able to check the status light/charging, so a wild idea came to my mind…What if I put some fiber optics in there?
So I modified the design to include a 1mm PMMA fiber in there. Interestingly enough, I already had to add a spacer in the inside connector anyways, because the male part was 2mm longer than the socket.
And the spacer:
And also, redesigned the pcb:
HOWEVER…
Meanwhile I waited for the pieces, I realized that the endcap didn’t had any form of retention, and altough that wasn’t a problem per se, I knew it would be lost at some point. So…
Yeah this had to be done. And with that, when all the pieces arrived:
Fucken awesome!
Open:
Closed:
Damn it looks cool. Also, the insides are prety clean now:
Yay!
Btw…I opened a Tindie shop, and this is for sale (this is only for CABERG DUKE SMART helmet + CABERG JustSpeak bluetooth:
(a step by step guide to use said machines, or similar, to make the semiconductors. A work in progress, updated often.
A video resume for the project has also been made:
I will be attending the Hackaday Superconference in November 2-5, and will be at the Poster sessions (think of a grown-up science fair) on Friday 2nd, in some obscure corner I pressume.
I will sit beside this poster and bore people to death about all the tooling.
And that’s it for now. Once I come back from the Superconference, posts should resume as “normal”.
See ya!
So, while waiting for the turbopump to arrive, my mind was left meandering around, and thinking about the PMMA masking, and the difficulty to obtain very low molecular weight masking liquid…
And I thought…If Silicon is sort of transparent to 10600nm laser light…but PMMA is not…can’t I just use laser etching to patern low resolution features on my test wafers?
With that thought, I prepared a test vector file with lines separated 100µm – 200µm – 300µm – etc… for the laser cutter:
Then proceeded to engrave some acrylic I had around, and McGyvered the shitty microscope we had at home:
That is nice!
That looks very consistent to me! Also, linewidth seems pretty constant (no micro power fluctuations in the laser supply)
For scale, a 110µm copper wire was placed in one trench:
The process idea is as follows:
First etch the PMMA with the laser, but not trying to go all the way, just to the top of the silicon wafer. Then, using PMMA solvant, you eat away some thickness from the leftover PMMA, revealing the silicon on the bottom of the trenches.
After that, an anneal step and you can etch the wafer. This should enable 100µm features, with 100µm spacing.
However, not everything was perfect in the land of the led. Previous tests where only lighting one led at a time sequentially, and that is innefficient both because I designed a parallel load, AND because reduced overall light intensity per led. Given that I was already working at very low levels (5/8mA per led) reducing it even more was not going to help with visibility. When I tried to start doing more, this happened:
More about it later, as it deserves a post of it’s own.
It was very cheap, and that way I don’t have to deal with my own homemade.
The Electronic Mercenary 