Motor Controller Progress Update

12 11 2012

Don’t worry everyone, I haven’t disappeared off the face of the Earth. Just have been taking my time with my projects and also ran into a lot of snags on the way.

First up is the printed circuit board for the motor control schematic I posted earlier. The design is finished but I have had some troubles getting in made so I can test it. I also have had trouble uploading the KiCAD files for the board onto github in a way that is then downloadable. So, as a result, I am currently moving all of the modules that I used on the board into a single library that can then be downloaded with the board. Hopefully this will fix the problem.

Now, back to the problems I’ve been having getting the board made. Since I am a low income earner, I can’t easily afford to get my designs made by professional board manufacturers and have to rely on my home made boards. I usually use the photo-lithography (also known as photo transfer) method but ran out of the positive resist boards that I have for this purpose. The people who I use to buy these boards from no longer sell them, which means that only the negative resist boards are available locally and they are quite expensive. This led me to find a seller on the opposite side of the country that does sell the positive resist boards I am use to, but they charge $15 for freight up to 3kg, which was almost the same cost of the board itself.

I shelved that idea at that point, mainly due to me being lent a Roland MDX-20 CNC router. This then opened up a bigger kettle of fish. I have had a bit of experience with this machine in the past and gave up on it. This was a result of the bad quality software that comes packaged with it that will only run on windows. This time around, however, I was determined to get something printing from it in an open source environment. I jumped the gun and downloaded the LinuxCNC ISO then installed it on an unwanted old computer I was given by a friend. This didn’t work for me because the Roland router uses a specific language that was derived from HPGL called RML and LinuxCNC does not have the capability to send these commands to it.

After this, I tracked down two pieces of software. The first is a piece of software developed a MIT (link to come) and another called Tux Plot. The MIT software worked but it was very time inefficient and took over two hours before it was even half done on my test piece. I didn’t spend heaps of time on this and it may have just been a configuration issue, but it put me off anyway. Tux Plot didn’t work very well at all. It is meant to convert HPGL to RML then send it to the router. But, it was hard to calibrate and also did some weird conversions, like trying to drill through the entire board at the corners of my design. It also does not convert some of the commands. The HPGL format supports circles and arcs directly but RML does not and you must manually tell it to draw circles and arcs point by point. Tux Plot doesn’t take this into account and just leaves the circle commands in the code without doing anything at all. This results in the router going to the location of the centre of the circle then going down then straight back up without making the circle. The final blow for Tux Plot came when I found out that KiCAD outputs HPGL in a configuration meant for plotters (as it is originally intended) and not for routers. So, where the HPGL file tells the router to go is where a plotter would normally draw with its pen. Instead I have a milling bit that wants to cut stuff away. This results in the tracks being cut away instead of the gaps between them.

It was with this realisation that I knew getting this router to work with KiCAD in a reasonable manner was a project in itself. So, I have given up on it for the current time and will revisit it when I am finished the bicycle. In the mean time, I wrote a post on the hackaday forums trying to organise a group buy for the positive resist boards I found, but have not had anyone reply as of yet. I bit the bullet and spent the money on the negative resist board, just so I can get the project moving again. I think I will have to save up a bit and do a bulk buy myself and sell the excess on ebay, or sell kits of the finished design to use them up.

That’s pretty well where I’m at currently. My next step is to sort out my github issues and manufacture the first prototype of the alpha design for the power section of the motor controller.

BLDC Motor Controller Schematic

11 10 2012

Hi all,

Power section of BLDC motor controller for recumbent electric bicycle.

This is the schematic for the alpha version of the power section within the motor controller I am designing for my electric recumbent. It is fully untested so use at your own risk. It will be run by an Arduino on my project but I have designed it so that nearly any micro-controller with TTL logic levels can be used. I am in the middle of designing the PCB and will upload all of the files to github once I’m finished. Also, an explanation of the circuit will be put up soon too and yes I am aware that the hall effect sensors are missing. I am not sure if I will be adding them to this board or making a separate one for them. If I do add them then I will post an update.

Hope you all enjoy and let me know if I’ve made any silly mistakes or if you think I should add something.



Test stand for Smart Drive Motor

2 10 2012

I made a little bit of progress on the recumbent bicycle front today. I’ve made a makeshift test stand for the Smart Drive motor so I can start testing some control board designs before I put in the effort of installing the motor on a bicycle frame. This way the motor has no load connected to it, which will keep the current required to run the motor to a minimum during testing.

The stand basically consists of the shell from an uninterruptible power supply (UPS), bolted to a piece of steel plate, with the nylon hub of the motor cable tied to the UPS frame. The first thing I did was strip the UPS frame and reinforce it with a couple of cable ties.

The bare UPS frame with cable ties to prevent it from wobbling.

Four matching holes were drilled into the cover of the UPS and a piece of scrap steel I bought for $5 and then they were bolted together using M4 bolts and nuts. I had to buy this piece of steel because, it wasn’t until after I gave the steel case of the washing machine to the scrap metal merchant, I realised I could of cut it out of that instead of having to buy more. But, that is a lesson learnt I suppose.

Lid of UPS bolted to steel plate using M4 bolts and matching nuts

Basic stand after UPS frame is bolted back into its lid and attached steel plate.

I then cable tied the motor hub to the basic frame through the holes that were already present in the bottom of the UPS. The motor does have a bit of movement after it has been secured. However, if this becomes an issue during testing I will probably just put some double sided foam tape down between the motor hub and the UPS frame to prevent it sliding about. Time will tell on that front.

The motor secured to the frame via cable ties around the motors bearing hub.

After all is said and done, I think I have managed to slap together a passable test stand that should get me through to the time that I have to actually mount the motor onto the bicycle frame.

Finished test stand. The steel plate was connected so that more of it was on the side that the motor would be on. Due to the centre of gravity of it not being directly over the hub, but underneath the actual motor. This way it stops the stand from falling over.

Fisher and Paykel Smart Drive Schematic

25 09 2012

I’ve spent my free time over the last few days reverse engineering the control board out of the Fisher and Paykel washing machine I bought. I managed to find details on all of the components except one. I know it’s a MOSFET but, beyond that, I don’t have any more specifications. It is labelled H75309 G814BE and is a 4 pin SOT-223 package. If anyone has any more information on it, I eould very much appreciate it. I used this codebook to help find all the parts with SMD markings on them, which was exceptionally helpful and I would highly recommend it. I rewrote the schematic in KiCAD (Build: (2010-00-09 BZR 23xx)-stable), which I will share with you all here. Please note that this is not a complete schematic of the control board but only a partial of the power section that drives the motor. I wasn’t really interested in the rest of it and it took a long time just to do this part. So, I’m not going to bother doing the rest, except for the hall effect sensor section at a later date. I’ve also included the libraries I created for the parts that KiCAD didn’t have as standard. They must be added to your library list before the software can use them.

Next, I’m going to attempt to remove as many parts as I can off the board to reuse, then use them to develop a prototype for my controller. Hope this all helps somebody.


Electric Recumbent Motor

22 09 2012

The journey has begun for my electrically assisted recumbent bicycle. I’ve collected a lot of parts, including the donor bicycles, some steel for construction, some of the batteries and the all important motor. I’ve decided to use a BLDC motor from a Fisher & Paykel washing machine (must be a model with the Smart Drive motor). This is for a few reasons; they are cheap to get, powerful and have permanent magnets in the rotor. Cheapness is important to me as I am a low income earner, I got the entire washing machine for AUD$5.50 off Ebay. So, for an approximately 1kW motor, that is a pretty awesome deal I think. The permanent magnet side of things is also important because I am wanting to investigate regenerative braking on this build and I am led to believe that these are more efficient in that regard. Most people that use these motors as generators usually rewire them for better voltages, but I am going to go with the stock motor first and see how it goes.

Here are some pictures of the tear down of the washing machine and preparation of the motor for mounting. I got the washing machine for cheap because it had stopped working and after the previous owners spent some money replacing the water pump, they gave up on it when it still wouldn’t go. Most of the time the power drive transistors on the control board burn out on these models and a handful of components can get them working again. But, I’m not interested in riding a washing machine all day, I’m interested in bicycles and how to make them go fast.

Fisher and Paykel washing machine with Smart Drive motor. Bought off Ebay for cheap.

To remove the tub and motor assembly, the four support rods on either corner have to be removed first. The top right has already been taken off in this photo. I just grabbed hold of the motor shaft inside the tub and lifted slightly to be able to lift off the grey holders on each corner. It worked well for me.

This is a picture of the tub and motor assembly once removed. The motor is the grey part on the bottom. The problem with this design is that it is hard to remove the motor from the tub. So, I decided to just cut away the tub right back to the part that holds the bearings and shaft.

First I drilled a hole in the side of the tub big enough to fit my jigsaw blade through, then cut around the perimeter to remove the top.

At this point I removed the rotor and stator then started cutting either side of the support sections that run out from the central hub up to the next cross support. After that it was just a matter of snapping each part out along the cross support. With the wider sections near the outer rim of the tub, I had to cut them into smaller sections before trying to snap them because, if I didn’t, they would only bend and not snap.

Cut down to the last lot of support sections.

This is as far as I could cut down the tub with my jigsaw.

To remove the last bits of the supports I tried both a hacksaw and rotary tool with a cut off disk inserted. The rotary tool won out at the end of the day but I had to be careful when cutting because it melts the, what I assume to be, nylon and it can wrap itself around the shaft. As it builds up it ends up snapping the cutting disc. I got around this by cutting through in layers and removing the build up around the cut as I went.

I used the barrel sander attachment on my rotary tool to sand back the remaining stubs to leave a nice finish on the hub.


After the hub was finished I reassembled the motor. You can see on this photo that the stator has some cracks in it. Apparently this is a fairly common occurrence with these motors. I think I will end up impregnating the cracks with some epoxy using some kind of vacuum set up to prevent it splitting further.