Smelting – 1st Run

It looks like one of the parts that will be needed for the mower will be a couple of spacers for the rear wheels. The idea is to fit new wheels to the 17mm axle the only issue being that the ID of the new wheel is much larger at approx 25mm.

A search on Trademe didn’t produce any results for an appropriately sized section of aluminum billet so after some though I figured why not cast some. A bit of research on the subject confirmed that on a small-scale it is a relatively simple affair, make a really hot fire, chuck some aluminum bits in a tin can, wait until it’s liquid and then pour the molten result into some form of mold.

A week of ferocious can collecting ensued (and is still ongoing) along with a quick whip around the house looking for any no longer used aluminum things.

On Saturday Matt, Bill and I fired the lot up. An old bouncy castle blower connected to a length of air conditioning ducting provided the air supply whilst a bag of charcoal in a brazier was used as fuel and container. A couple of crucibles (old fruit salad cans) were heated up initially to burn off the label and any rubbish inside before the scrap aluminum was added to the larger of the two cans.

IMG_6277 (Resized)

IMG_6278 (Resized)

IMG_6279 (Resized)

The results were quite promising, the scrap aluminum heated and melted quickly and so a couple of drinks cans were added to the mix. These didn’t go quite as well however and instead of melting they just oxidized creating a lot of slag that then needed to be scraped off.

Aside from that the pouring went well, A muffin tray formed a mold with 4 and a half aluminum muffins being the result.

IMG_6282 (Resized)

We decided that there must be a better way of melting everything down that addressed the major issues encountered;

  1. The fuel burned through too quickly and released a lot of ash that then landed in the molten metal
  2. The size of the tin made it hard to effectively scrape off the slag
  3. The drinks cans oxidized as opposed to melting

On Sunday we reconvened for a second attempt after refining the technique and equipment. Firstly the charcoal was swapped initially for cedar and then once there was enough heat in the fire, for coal. The can was swapped for a larger metal pot which we figured would allow us to more easily remove the slag and also dunk crushed cans under the surface so that they would melt and not oxidize.

IMG_6283 (Resized)

IMG_6284 (Resized)

IMG_6285 (Resized)

IMG_6287 (Resized)

The larger size of the pot meant that more fuel was needed to get the aluminum to melting temperature but once it did it was a lot more stable and less affected by the addition of new material. Overall the amount of fuel used to produce each muffin was about the same but the lower cost of coal made it slightly more cost effective. End result for the day, 11 muffins and one uncast chunk of solid aluminum for next time (we ran out of fuel before we could get it to a decently flowing state)

IMG_6290 (Resized)

Next step looks like it will involve the manufacture of some form of furnace. It looks like you can mix cement, crushed kitty litter and something called Perlite (apparently available from gardening centers) together and then form it inside a metal cylinder to make a rudimentary furnace (if anyone has an old gas bottle that has expired / not passed re-certification then I can get rid of it for you?). You then have the option of either using charcoal / coal or propane to heat the furnace which should work a lot better due to the greater amount of insulation. Hopefully this should be the only step up that is needed to turn what is really a once off procedure (due to the inefficiency) into something that is feasible long-term. With that will also come the final stage of recasting the aluminum into the final shape.


Mower Restoration – Initial inspection

So recently I acquired an old 2 stroke lawnmower off Trademe for a real good price. From what I can find so far these particular engines were produced between 1954 and 1977 so the mower is very likely older than I am and yet still runs on the first pull.

IMG_6268 (Resized)

Whilst the mower runs perfectly I decided to take it apart to have a look at the workings and see what maintenance / cleaning could be performed to prevent any future issues from developing. After removing the top cover the first problem that needed to be dealt with was the massive amount of gunk that accumulated on and around the engine since it was last cleaned. Apart from that most parts looked to be in pretty good order, there were some mis-matched bolts and nuts and some peeling paint but the rest looked to be in pretty good order. The only other could-be-potential-I’m-not-sure-it-could-be-normal issue were a couple of wet patches underneath the carburetor and cylinder head. I guess only once a proper clean has been done will it be possible to know what is really going on here.

IMG_6269 (Resized)

IMG_6270 (Resized)

IMG_6271 (Resized)

IMG_6272 (Resized)

After scraping most of the gunk off the shroud and fuel tank Bill and I began stripping off the oily residue and flaky paint that was left behind. The end goal is to get these both to the point where they’re nice and shiny and ready for spray paint. Next the mower was tipped on its side so we could open up the muffler (banana shaped part behind the blades attachment). The inside of this was pretty clean with only minimal carbon build-up so I assume that the engine is running pretty well.

IMG_6273 (Resized)

Next step will be removing the engine from the base. We’ll strip the paint from that and inspect the repairs that have already been completed on either side of the base. It looks like at some point the deck has cracked and someone has gone and riveted some extruded aluminum across the crack to hold it together. These patches will then get removed, the cracks cleaned and then properly fixed with some Durafix or similar. Something similar will be done to the chunks that have been taken out of the front of the deck between where the blades spin and the front axle goes. Then it’ll be strip, clean and spray for the deck.

In regards to the engine a good clean will be first, and then an inspection to work out how it goes together before too much more is removed. Following that a greater break down into individual parts for further restoration and investigation into the potential leak. From a first inspection it looks like the cooling fins on the head may also need to be stripped and re-sprayed with some high temperature paint as there is some flaking happening. Hopefully the project should come together pretty quickly without too many issues, the fact that the mower runs well, has good compression and only a small amount of build-up in the muffler means that aside from the oil leak hopefully not much other than a clean and paint will need to be done.


New Reprap

Generally my relationship with NZ Couriers goes something like this. I order an item, they pick up the item, item gets lost several times, travels through Gisborne, I call up, they investigate, I receive item 1 week after shipping despite it only having to travel across town… As such I avoid using their service like the plague and bang my head on me desk repeatedly when I find out that someone has dispatched a part on their service.

Imagine my surprise then when I ordered some Reprap parts from Christchurch and they managed to pick them up on a Monday afternoon and have them at my desk first thing Tuesday morning. I still have no idea how they did it, but anyway I received the following. (and no that is not my real address)

20131001_095910 (Resized)

20131001_100024 (Resized)

20131001_100050 (Resized)

Next step – Take received parts – Add printed parts

IMG_6229 (Resized)

Mix well

IMG_6228 (Resized)

Let sit overnight

IMG_6226 (Resized)

IMG_6241 (Resized)

End result after a weekends work on and off. The Silver and Black theme is purely accidental, it just happened that I had black filament in the printer and that the motors were black but I like it and think ill carry on with it. Printing the parts has been a bit of a pain in the butt this time around, a combination of upgrading both Slic3r and Repetier host means that he V2 needs a bit of fine tuning to get the result back to quality that it was before the upgrade but I think on the whole the upgrade was worth it as the bridges seem to be coming out a lot better. The other change that I needed to make was moving everything across from the Vista boot that I am running for general day-to-day stuff on the computer to the more stable (read not full of crap) Win 7 boot that I use for programming. The reason for this is that the printer was more and more randomly carking out whilst printing which really tears your gruds when there’s 5 minutes left in a 2 hr print.

The intention with this printer is to go with a .35mm nozzle and have a default lower layer height to obtain higher resolution prints on small or technical parts.  In addition to the nozzle the I3 frame seems to be a lot more rigid in both the X/Z axis and the Y/Z axis so this should help get the quality of the output up as well.Ill still keep using the V2 for larger parts to keep the print times down though.

At the moment the 4mm to 5mm couplers are on order as well as the lm8uu bearings so ill fit those when the arrive and complete the next step of construction. Until then it’s going to be a case of tweaking the V2 to try to get the output quality back to where it as and printing the remainder of the parts.

As a closing pic have a look at the inside of a mobility scooter gearbox.

IMG_6225 (Resized)

Finished Trolley

Wow, two finished project posts in a row. Something must be going right!

Friday evening a couple of weeks ago saw the purchase of two new 71Ah 12v deep cycle batteries for the trolley. Unfortunately this came at the cost of one of the spinners which cracked when it hit a curb (curse you perspex) so these have all been removed pending some serious thought into durability. But back onto batteries… Originally the scooter came with two 32Ah ones but they were pretty old and I only managed to get one decent trip (3 odd km) plus some testing before they didn’t hold charge properly and were pretty much useless. Looking through the service manual the specs list a range of 30kms with the stock 32-35Ah options and whilst this may be a generous estimation either way the new (71Ah) ones should be a distinct advantage.

After charging overnight and a few finishing touches on Saturday morning it was deemed to be time for a proper test. The course chosen was a short round trip from Torbay to Browns Bay which according to Google maps works out to be 8.3 km total. Going each way took about 45 mins so the top speed of the scooter is definitely lower than the quoted 10 kph but the better news is that after the trip the battery indicator still showed 4 bars.

The only annoying part of the whole thing is now the charge light on the scooter is permanently lit… Originally this only came on when the battery level had dropped until only the bottom bar on the battery level indicator was lit and also when the scooter was plugged in to charge. I’m not sure what is going on with this one at the moment, the stock controller has been programmed for the SLA batteries so it’s not like it’s thinking that it has a different type of battery, and as all four battery bars light up then I can’t see how it can be thinking that the battery is both full and also in need of charging at the same time.

Troubleshooting wise I’ve confirmed that the battery level indicated by the display is actually indicative of the voltage remaining in the cells (ruling out the battery management stuff being stuffed completely) and that there isn’t programming error that is reading the output from the original controller wrong. If there was still a LED connected to the original controller board then it definitely would be lit. Either something on the stock controller has been fried at some point (but very selectively as everything else still works), the scooter is confused with the extra capacity provided by the 71Ah cells (although this would be very weird) or the scooter is still getting used to the new cells. A trawl through the service manual has yielded no results so at the moment the only other step that I can think of taking would be running through the original controller board again and confirming that there isn’t something weird happening with the matrix that handles the input and output from the I2C expander on the original control panel.

Apart from that the rest of it is running well, the computer software has been finished to remote control everything apart from the steering and the final few hardware connections have been made connecting all the Arduinos to each other to allow all this remote control stuff to happen.

I’ll post a video (weather permitting) soon going through all the finished features as well as covering how amazing these Plega mobility scooters are for hacking, both in their simplicity and the extreme amount of thought that has gone into the design. It really comes down to the fact that for $150NZD you can pick up the complete chassis, controller, 2 motors and gearboxes, seat, steering stuff, wheels user controls and charger.

In the interim here’s some finished pics

IMG_6252 (Resized)

IMG_6253 (Resized)

Drivers View

Drivers View

IMG_6251 (Resized)

Lights and Siren Controls

IMG_6255 (Resized)

GPS and GSM antennas

IMG_6254 (Resized)

IMG_6256 (Resized)

Main power switch and Regulators

IMG_6258 (Resized)

IMG_6259 (Resized)

IMG_6260 (Resized)

IMG_6261 (Resized)

IMG_6262 (Resized)

Before and After

Before and After

How to Hack a PG Wheelchair controller

Recently I was very generously given a couple of large geared DC motors off an old electric power wheelchair and a PG motor controller to go with them for a future project.

IMG_6234 (Resized)

IMG_6233 (Resized)

Gearbox cover with lever for Freewheel release

IMG_6232 (Resized)

Brake unit

Brake unit

Whilst it’s possible to get a Sabertooth driver or similar off eBay it would be great if it were possible to interface with the gifted controller in a way that made it usefully compatible with an Arduino.

After trawling through the internet for a couple of hours two things became apparent. Firstly PG are very tight with what technical documents they release which meant that there was very little information out there about the controller and secondly what success people have had in interfacing with these controllers (which hasn’t been much) has been through the joystick user interface (which I didn’t have).

After cracking the controller open and having a poke around it became apparent that the motor driving stuff is all located on the top circuit board and all the control / user interface stuff is done on a separate circuit board that clips onto the back via a 22 pin connector. The  chosen plan of attack was to then write off any possibility of interfacing with the control board and just hack directly into the motor drivers.

Controller board that connects to back of motor driver board

Controller board that connects to back of motor driver board

This board basically consists of two H-Bridge drivers each driving 8 mosfets (4 pairs), one H-Bridge for each motor. Additionally there is a 12v regulator, a large central relay that switches the positive supply to the H-Bridge mosfets, and then the following

IMG_6247 ANNOTATED (Resized)

Mosfet 1 and Mosfet 2 – Power to electromagnetic park brakes

Mosfet 3 – Power to 12v Reg and Mosfet 4

Mosfet 4 – Power to Relay coil and ??

BJT 1 – Connected to Gate of Mosfet 1 and 2

BJT 2 – Connected to other side of relay coil and GND

After tracing as much of the pinout as possible the connector was (very poorly) removed and fly leads were soldered to the appropriate points as well as heavy gauge cables for the battery and motor connections. A computer power supply was connected to provide a 5v rail.

IMG_6242 (Resized)

IMG_6243 (Resized)

Testing then commenced with a couple of 12V batteries connected in series and one of the motors. Initially there was little success and so it became a game of working what needed to be powered on in what order. Eventually we got the following order worked out.

Mosfet 3 gate to GND (12v supply now active) Mosfet 4 gate to GND, BJT2 base to +5v (relay now switches on) Base of BJT1 to +5v (Mosfet 1 and 2 now powered, park brake clicks off) Pin 8 of H-Bridge to GND (enables outputs on H-Bridge) Pin 3 or Pin 4 of H-Bridge to GND (drives motors!!)

Basically now that it seems to be possible to run the controller without the proper joystick it should be a simple matter to connect everything to an Arduino and write some code to PWM control the H-Bridge to drive the motors properly. The rest of the unused flyleads appear to all be designed to be inputs back into the controller to make sure that the park brake is off, that the battery level is ok etc so it will also be a case of working out what gives what information and whether its worth providing that feedback as to what is going on to the Arduino. Ultimately though I now have a motor controller that I can control and even though this one was free they seem to be relatively cheap on eBay for the potential they have.

If anyone else wants to have a crack at this, first step will be to (properly) unsolder the white connector on the back of the driver board. Then starting at the left and working to the right (top row first) before doing the same for the bottom row the connections as I have worked them out are as follows

  1. GND?
  2. Mosfet 1 Drain – Monitoring of voltage?
  3. BJT1 Base
  4. H-Bridge Mosfets – Monitoring of voltage?
  5. H-Bridge Mosfets – Monitoring of voltage?
  6. Pin 8 on H-Bridges (DIS)
  7. H-Bridge Mosfets – Monitoring of voltage?
  8. H-Bridge Mosfets – Monitoring of voltage?
  9. Mosfet 2 Drain – Monitoring of voltage?
  10. Linked to 11 via cap
  11. Linked to 10 via cap
  12. Pin 3 H-Bridge 1
  13. Pin 4 H-Bridge 1
  14. Gate of Mosfet 3 via Resistor
  15. Gate of Mosfet 3
  16. 12v Reg output
  17. ??
  18. Base of BJT2
  19. Collector BJT2 – Monitoring of voltage?
  20. Gate of Mosfet 4
  21. Pin 4 H-Bridge 2
  22. Pin 3 H-Bridge 2