Category: MIT & Boston

Anything that happens in MIT and beyond.

Loose Ends and Tag Closing for Bits of October: Site Updates, Chibikart and Mini-Jasontrollers, New Expensive Things!

chorl1 Comment

Now that the season of Dragon*Cons and Maker Faires and everything else has finally settled down, I’ve reached the curious state of having nothing to do with my life, being between large builds in much the same way you’d be between coffees or meth hits. My day to day activities revolve around managing the IDC (excuse the cheesiness) fabrication facilities, of which there will be some updates shortly, and monitoring & mentoring the classes running in the center, including the renowned How to Make a mess out of Almost Anything. I’m not a TA for the class per se, but part of the process of making sure the shop isn’t lit on fire is some times giving extra attention to those who would be most likely to do it.

That isn’t to say that my life is entirely empty and devoid of meaning. I’m tending towards taking the downtime to fix up my eternally problematic go-kart children, starting with Chibikart2. During some hard running at the Powerwheels race, I lost one of the Jasontrollers to Sudden Jasontroller Death Syndrome, a fairly common failure mode for them when they are over-run. The failure is always gate drive destruction since the circuitry is so fragile, and always not worth repairing to myself because it involves replacing so many small shitty transistors. Next up on my list after this is probably to add the electronic solenoid shifting to burnoutChibi and finally get rid of my super-rigged cable linkage. I’ve also been collecting many prospective parts for the “Chibi-Mikuvan” project, so stay tuned for a massive Beyond Unboxing the likes of which have never been seen!

But first, by popular request, I’ve added Pad Thai Doodle Ninja and Colsonbot CAD files to the References page. PTDN’s files are only made of 3D printable frame parts, but Colsonbot is the full bot – you’ll need Autodesk Inventor or a compatible viewer for anything but the STL files. All of the details on these bots are available in their respective build threads.

Onto Chibikart’s controller update. Like the dual controller mount I made for BurnoutChibi, I designed up a two-mini-Jasontroller snap-fit mount which also holds an 80mm fan. Essentially the same idea of BurnoutChibi’s. I was planning to current-hack these controllers to 40A, and for sure they will need supplementary air cooling.

 

The mount was printed on my Up machine, and this is about the largest object I’ve found it can handle reliably. It came out well, with minimal warping. I sincerely recommend the Up (now on 2 Plus!) to anyone thinking of getting a small hobby-class 3D printer.

Short of popping it in the Dimension, the Up is my go-to for structural parts. The ABS formulation they use is a higher hardness/toughness than the soupy generic stuff you feed to RepRaps and Makerbots. I was concerned about the snap fits being too aggressive and snapping off themselves, but they turned out to be just on the side of the acceptable line.

The mini-Jasons were cleaned up of unnecessary wires, leaving only the motor phases, power, the Hall sensors, throttle, and the ‘regen brake switch’ which may or may not be wired in in the future. The regenerative braking on these things is a fixed low current on-off kind of affair, so it’s not very helpful.

I plucked the 80mm case fan from stock – there’s nothing particularly special about it.

In the past, I’ve current-hacked these things with a blob of solder on the current sense shunt, but it’s such a bad hack and is unreliable – I’ve actually had the blob melt back off before. To remedy this, I began looking for large current shunt resistors packages that fit in between the leads of the existing shunt. This is the result – for a mini-Jason, a “2818” (.28″ long, or so) package current sense resistor is a nice fit. One that is 8mm (“31xx” – “35xx”) will fit even better and not require much solder bridging would fit better, but I could not find any that were not also square in shape – rectangular, the long way, is preferred.

I actually had this hack vicariously tested by Daniel (YAMEB) a while back. These shunts are 5 milliohms (not 10 – I measured erroneously the first time), so it took a nice sandwich of 10 milliohm resistors to get my 40 amps. The exact part number I used was WSHA-.01CT-ND, and it has a 5 milliohm brother in the form of WSHA-.005CT-ND.

I cleaned up the floorpan of Chibikart after removing the old Jasontroller – it was positively disgusting and filled with 2 years of floor grunge buildup, plus mud and dirt from running at two slightly wet Maker Faires. The new installation drops right into where the old controllers used to sit, after redrilling some mounting holes.

Systems wired back up. The first test drive was without the fan hookup, and without the sensors connected.

To rehash, these controllers “self-calibrate” sensors if you connect them and then run once to full speed. I couldn’t achieve this on the ground since the vehicle never really reaches “full speed” in the space available, so I had to freewheel it, being mindful of the 4700rpm-ish commutation limit. After one power cycle, the controllers had learned the sensor configuration and Chibikart could apply “static pressure” to something again. To get a good transition between sensored and sensorless, the sensors have to be aligned properly first (check out Equals Zero Designs’ page where I have an actually well documented example.) – and that’s all you need to do, not actually try and optimize their timing position.

This was, of course, the important part.

Now, the 12v PC fan could not handle 24 volts, so I just dropped a giant 40 ohm resistor in line so the fan only saw about 15v. This resistor surely dissipates more power than the fan actually removes…

With two motors on 40 amps, instead of on ~25 before, Chibikart2 is way more fun. Not, say, tinykart Black Edition level fun, but it is far more peppy. The small Colson wheels are starting to reach their traction limit.

I hit 1.1kW on indoors testing, and there is much room for improvement yet. Because the controllers are doubtlessly still running constant current before I run out of hallway, the power will only increase with vehicle speed.

Say, I haven’t garaged something properly in a long time (mostly because said garage was under repair construction this past summer). Maybe it’s time to take Chibikart back to its proving grounds.

Next, some of the ongoing facilities improvement projects that I have going on in the space! The place is kind of like I-95 around New England – always looking like someone’s working on it and the construction shifts every once in a while. I swear, though, it’ll be over soon – just like they say in Connecticut (In my six years in this area, I have never once driven through 95 in CT without hitting some kind of construction…)

First up, a full size Shopbot – the full 5 x 10, gifted by the Architecture department. I’ve been itching to have one of these for a while – with an 1/8″ carbide bit, they’re practically mechanical waterjets! Expect some Shopbottables to emerge on my end soon due to the “It’s the closest tool next to me and I don’t have to ask anyone to use it” effect. It will be very handy for producing Chibi-Mikuvan’s body panels since they’re all larger than what can be stuffed into the laser cutter.

Above, Media Lab students operate the machine as part of the MAS.863 “Build something big” week.

Next up, the legendary Form 1. Full disclosure: There’s four Form 1en in the space at the moment – this one is “The Lab’s”, and the rest belong to researchers and classes residing in it. Four. That’s more forms than Formlabs (okay, probably not), but the Form 1 density must be up there.

The Form is a SLA-like machine which can hit much higher resolutions, but the  material isn’t too strong – it’s an acrylic resin, so it has some mechanical strength, but does shatter and snap. Dat rez tho…

These are some of Brian Chan‘s insects. Check him out on Shapeways! I also printed the crab, lobster, and some other doodads from his collection.

Of course, with every 3D printer that makes it in here…

The  model is “Pillared Miku” though I used the version without built-in pillars – the Form software generates its own support lattice.

Now, moving up in the Expensivity scale is our latest acquisition:

 

An Objet24 (By Stratasys™)! This is just contributing to the slow rounding out of 3D printer technologies in the space. Objets are incredibly high resolution, very nice, and very expensive. This unit was purchased used from a local company for only $7,000, but you’d easily eat up that much per year in materials alone. The Objet Goo comes in 700 gram jugs that each cost $300-350 and up.

And this is the entry level machine.

The Objet technology combines SLA (light cured resin) with inkjet style nozzles so it can control the deposition very finely. No giant bubbling cauldron of goo here. It also has its own Windows XP computer built into it.

Now, I know XP is pretty much the OS that saw the Internet grow up with it, but this machine was built in 2011….

…and even worse, it requires a very specific network setup to talk to. Objet-Stratasys (ObSys? Stratajet?), I’m going to publicly shit on how bad the Objet communication infrastructure and software are. I should not have to configure a point-to-point LAN, disable Windows Updates, and disable firewalls just for it. The whole setup procedure gives me the vibe that they had to ship the machine and had 1 day left to write the drivers, so took whatever the developer’s computer was at the time and just made that the exact requirement. That, or given Objet is an Israeli company, probably just opens your computer up to direct monitoring by the Mossad.

I’m amazed I didn’t have to start Space Pinball and log into Pandora before the printer would communicate.

The slicing software is also slow, prone to crashing, and has an inconsistent UI. For such a beautiful piece of hardware, the software end of it seems so incredibly rigged.

Of course, the first thing to do with every 3D printer that makes it in here…..

Yeah. This was like a $30 Miku given how much of the material I used.

This corner of the room has been reconfigured to become what we now affectionately call “printrgartn”. The Form 1 is immediately off to the right, as is the Replicator 1. I’m trying to commission an Up for the lab (in supplement to my personal machine).

What’s absent, sadly, is a powder printer. I need to do some Powder Print Affirmative Action here.

Loose Odds and Ends from the Past Week or so

chorl3 Comments

Things have been getting exciting in the last week or so as the SUTD Summerkarts Global Leadership Program students have arrived, and we’re now well under way into the Silly Vehicle design phase all over again. Global Leadership Program. That’s such an epic name for 2.00gokart, guys. To be fair, there’s plenty of other things going on for the students too, most of which are ‘leadership’ flavored. In this running of the class, since I’m not being watched over by The Department and need to make sure everyone has paper lab notebooks, the student groups will be blogging their builds! I’ll post a list of links once everyone gets set up.

mikuvan

I’m currently at 380 miles.

Having essentially reached the limit of things that have been going wrong, I’ve been faced with no choice but to start attacking rust. This thing hasn’t even so much as hiccuped a single time since Operation: Bad Timing. I would say that at this moment, having checked everything I think is important, I’d trust a trip out to at least New York City (about 240  miles).

I’ve mostly been spending the past 2 weeks psyching myself out and picking up some materials.

 

Based on estimating the clooooouuuuuud Internet and asking friends, I got a pile of things from Eastwood – panel sheets, a bucket of sealer, and some rust converter. I was skeptical about the “rust converter” – it allegedly converts iron oxide (“rust”) to iron phosphate or iron tannate, which is some shit I’ve never heard of and only appears in product descriptions plagiarized from Wikipedia. This sounded shady, and success seems to be hit or miss. However, it’s relatively cheap, so we’ll see what this magical potion does. That, plus another haul of random abrasive and sheet metal banging tools from Harbor Freight, ought to round out the basic non-sketchy rust patch. I’m going to try the pound-and-weld-metal route – falling short of soldering since it sounds a tad too hardcore at the moment – instead of throwing fiberglass at it.

All that, and waiting for several days of hot and dry weather to do the majority of the work, just to smoke out any residual moisture from the body holes. Recently, it’s actually been hard to come by, with the Northeast in its Periodic Random-Ass Storm Season (PRASS). There’s no point in tying a puddle up inside my work. Worse come to worst, I’ll point a space heater at the trouble spots for a day before doing anything.

What I’ll do first is probably do all the sanding and grinding; the full-depth investigation, basically, and then post it publicly to get some opinions and appraisals. The idea is to cut or grind off what I can get to, covert and cover over what I can’t abrade off, and then slather external underbody repairs in sealing compound. I also managed to find a matching Chrysler color at Advance Auto Parts to repair the exterior paint afterwards, tested by blasting random areas and staring at it a few minutes later. This may backfire horribly.

Part of the reason I’m hesitant to start is because I have the feeling that things will get more and more Death Race 3000 if I mess something up or discover more structurally unsound areas than I previous anticipated. What you can’t see….

Though, in the limit of Death Race 3000 style modifications, this thing will probably look more and more like a classic wedgebot.

testing the mini-jasontroller

I replaced the full-size caseless Jasontroller in RazEr REV2 with the mini version detailed last week.. I’m definitely a big fan of these now – they’re basically the same as the full size, in a much more useful package.

This is the smaller controller uncased and fully cleaned. I essentially took out every wire I didn’t need, and also locked the speed setting to high internally by jumping the right side orange wire; by default, the “3 speed range” switch comes in the medium range, which means it divides down the throttle input) This has no bearing on its upper speed limit which is still around 540-550 eHz or so, but for low speed motors the throttle response will be substantially retarded otherwise.

These controllers have a discrete logic power switch, unlike the full size Jasontroller, so I also hardwired that internally (left).

I discovered that the entire controller case fit inside the space where my old full size Jasontroller went, minus one corner. So, instead of redrilling the mounting holes for the smaller heat spreader bar, I cut the entire case into an L shape with the board resting in its stock location…

…shrink wrapped the whole thing, and Velcro-mounted it in. Clean and waterproof, and the additional aluminum should still offer some thermal capacity.

The overall height of the controller when stripped of its case is under 0.8″ (in my configuration, it’s not much shorter due to the existing case outline remainder), which opens up the potential to be stuffed into even more things.

I’ve noticed no difference in riding behavior between the mini-Jasontroller and full size, once again confirming they’re basically the same thing. I’ve noticed some slight difference in starting behavior – the mini doesn’t twitch backwards, at least not often. More observation will be needed to discern the differences. In the mean time, I’m officially qualifying the mini-Jasontroller as Certified Legit. You can buy it on this page, and maybe soon from Equals Zero. If you make something using one, post it!

I’ve added this controller to my Scooter Instructable in the EV controllers section.

more silly rideable things

One of the downsides of having 160 cubic feet of self-motive cargo volume is Oh man, this free stuff on Craigslist looks awesome. I’ve previously been limited by what I felt like carrying back on Melonscooter, or worst case, ride back independently. That’s no more.

What you see here is a most-relevant-to-my-interests free Craigslist haul of two nonfunctional electric bike (-like-objects) from a closing e-bike shop. The one on top, as it turned out, is quite the machine. It’s a TidalForce IO cruiser bike, from another one of those small EV companies with an illustrious but ultimately short lived existence, in complete condition. The bottom red pile is a generic Chinese “电动车” or “Chinese moped”. These are sold here and there under various names (here’s one example, and most likely the company that retailed it since it says GREENPOWER on it!). Its condition was a little more beat up, but seemingly just devoid of batteries.

Here’s a better shot of both of them. Being me, I’m actually more a fan of the little red moped – it’s a little weirder and has that Chinese charm to it, but the Tidalforce was much more complete, so I began messing with it first.

The backstory of this machine was that the customer dropped it off for battery service and abandoned it. Apparently, these bikes were notorious for having their NiMH cells degrade very quickly.

Lacking a legitimate charger, I jacked it in on a power supply to 45 volts CV and fed it at about 0.8 amps for basically the better part of a day. The cells inside are nominally 8Ah, so the charge rate is a nice C/10 trickle charge. In case any of the cells were permanently toast, it wouldn’t cause thermal runaway. The battery came off nice and warm, and I rode around until the bike shut down from undervoltage. This charge lasted basically 5 miles with very little pedaling. The original advertised range was 15 or 20, but according to the storytellers realistically 6 or 8 miles, so it didn’t seem that far off the mark.

The termination condition is dictated by the battery management system onboard, and this is where things got difficult. No matter what, I couldn’t convince the charge-o-meter to go above 20%, even when I’ve clearly left the battery on slow trickle for many hours! I suspected that this artificial BMS meddling is what shut the bike down in the first place, since it didn’t feel like it was about to slow down.

I did some research online and came upon this useful page for decyphering the onboard controller for the bike, as well as this flamewar thread on Endless Sphere where someone mentioned that the battery needs to be discharged to under 32v to resynchronize the charge indicator.

This battery is too damned smart. I couldn’t get any output voltage from it unless the bike was on, since it has internal FET switches to shut off the cells from the pack output, so I couldn’t artificially drain it. And even at 44 volts off the charger, I couldn’t get the bike to move more than a couple dozen feet before the BMS shut me down. I hate it when batteries are too smart – I’m forced to crack them open.

Off the front wheel comes. It’s on a quick release, so a latch and some cable pulling later and it comes cleanly off.

Removing the case screws and side, check out this holeaphobia-inducing lotus flower of cells! The terminals all had bits of corrosion on them, but there were no signs of leakage that I could observe.

Hammering on the opposite side of the wheel makes the entire battery structure fall out. This is the important side of things – the BMS board. My mission was to artificially brick the pack via the CELL tabs, draining them to under 32 volts, hoping the BMS would reset or something.

I used this shady arrangement of power resistors, totalling 15 ohms, to drain down the pack over the course of about 3 hours, getting the whole array down to about 30v. Afterwards, I immediately closed everything up and threw it back on the charger. It did exactly jack shit.

The battery meter blinked 20% the whole time! I’m going to guess I did this wrong somehow, or more likely, forgot that Ni batteries bounce back in voltage very well after an initial discharge. By the time I was done connecting things back up, the battery voltage could have been well in excess of 32v, making the BMS think everything was still skullfucked. But it should at least recognize the 7Ah I dumped back into the battery, right?! No such deal.

As of now, I’m currently riding this thing around day to day to burn down the charge in a useful fashion. Apparently, the charge meter blinks in its entirety when the BMS reset point is reached, so I’ll hopefully be ready then.

Why am I trying so hard to use this proprietary-ass stock battery when I could very well just hack the “B” battery with any number of potential long running packs? I’m hesitant to do that because I don’t actually like this thing. It weighs nearly 60 pounds and is enormous, clearly built for a much Manlier Man than I. Plus, I can barely stuff it inside my front door. I guess I’m used to smaller and more portable scooters which can be rolled inside – this sucker is going to need the bike rack. Not really my style.

For now, though, it’s alive and working as yet another Craigslist impulse that turned out to be a little neurotic but otherwise livable day to day. What’s with me and that kind of thing lately?

Let’s move onto the Little Red Moped.

After diddling around with the TidalForce for a few days, I decided one night to get this contraption running along with Adam. I cleaned up the mechanicals and repaired the existing wiring while he created an impromptu brick of 12V7 modules I have on standby for the summer EV design class.

Look at that beautiful… 20 gauge? wire going to the hub motor! This machine is capable of Real Power.  The hub motor appears to be a 48v, 500W (or 750W) brushless type, like this.

Unhitching the electronics box, I discover this wad of wires. If you ever wonder what Jasontrollers and their ilk are actually used for, this is the answer. As you are reading, millions of Asian moped bros are cruising about on machines exactly like this one.

During my wiring cleanup, I found a spider!

Someone clearly was derping around with this after-market and the controller is likely not the original. Someone was also terrible at this. There were plenty of examples of wires just twisted together and electrical taped up, and solder joints like that.

Whatever. It worked, and all I really did was replace some of the decomposing electrical tape and resplice some of the signal wires appropriately.

With the impromptu 48v battery and a random found bike seat, it was ready to roll!  And roll it did. The acceleration was brisk and utilitarian, and the suspension was a bit underdamped but compliant even when riding up curbs. It’s very quiet, and there’s a pedal assist sensor which almost sent me into the wall a few times when I instinctively kicked the pedal out of the way.

Riding it in this form makes me envision myself wearing a straw hat and dark brown Mao suit, riding along a dusty Chinese road to my factory job. With a cage of chickens on the back to be sold at the market later that day.

We agreed it would be more amusing once completely rewired and running on 72 volts, but sadly, Mao’s Little Red Moped did not see that day. For at Swapfest, I was riding around aimlessly for no more than 10 minutes before someone bought it off me on the spot. Sans batteries, but still.

So my net wheel gain for the past few weeks has been 2, both won by the TidalForce bike. Unfortunately, that may increase again, because I’m considering…

a playmate for mikuvan

Your job, Internet, as the guardians to my sanity, is to tell me I do not need another one of these.

Let’s face it. I was originally looking for a science project with Mikuvan, but elected to put in an honest repair effort to have me some of that thar “auto tech” larnin’ y’all kids are into these days. But now it’s running too well, and a few of us are basically invested emotionally in it, and I have a harder time with the thought of tearing everything down again than when it wasn’t running.

Mere weeks after I stated my life goal Passive Non-Career-Derailing Desire was to collect the Legendary Van Trifecta, I discover that I might have a chance to nab the rarest of them all: the USDM Nissan Vanette. Yes, the one which was well known for lighting on fire.

The back story for this find is quite circuitous indeed. It wasn’t by weeks of stalking Craigslist, or a “Hey, I hear you like derpy vans” referral from my “Hey, I heard you like trashy electric scooters” network. Instead, while doing research on the other members of the trifecta, I found this Jalopnik post for a Nissan Van (-shaped-object) in North Carolina. Some link hunting led me to the original sellers album…from 2011.  Out of sheer shits and giggles morbid curiosity, I emailed the seller what amounted to “lol do you still have this”. MFW the answer was yes:

I’m at a loss about what to do.

On the one hand… Whoa, a chance to capture the rarest Legendary Pokévan and train it make it the base for this electric drive project. After all, the way the world apparently works, as I’ve handily found out in the past few months, is you have a functioning car, then you get an explicitly nonfunctional one to mess around with. This van is so explicitly nonrunning the FCC and ASE are about to join forces to erase it from reality. The chassis mechanicals appear to be comparatively rust free (then again, so I thought with Mikuvan).

The downside? Space. Parking. Not even counting the (once yearly) cost of registration and (fairly low) monthly insurance premiums once it’s operational, there’s no such thing as a little empty grassy patch to stick a nonrunning vehicle here. I was lucky with Mikuvan that my one allotted parking spot was open. What I cannot justify is paying hundreds of dollars a month for a parking spot or garage space for a van-shaped lump without the knowledge that I will immediately be able to attend to it, rare or not.  I’m currently in the process of exercising my social network™ to see if anyone is willing to put up with my bullshit. Ideally, there’s a back alley of a nearby industrial space somewhere that I can slip into, or someone’s back yard who thinks this is all too hilarious. I’m not going to try very hard.

So the dilemma goes. I will probably not see one of these in such a complete condition for many years, but maybe in said years I’d be in a better position to start Big Chuck’s Van Adoption Service.

 (All pictures above of the vehicle were provided by the seller)

I’m filing this post also under Beyond Unboxing since so many things were taken apart in one way or another.