I picked up a Mitsu HF-SP102 servo motor for my OSW direct drive wheel project. It’s much lighter than a Mige 130, weighing only 6.5kg. I also like that it is made in Japan. It has lower inertia than the Mige, and it should put out enough torque (constant 5Nm at 6A RMS, and maximum 15Nm with 16A RMS).
I’m going to use it with a Granite Devices Ioni servo drive, and STM32F4 Discovery controller running the MMOS USB HID firmware. This means it needs to have an incremental encoder, rather than the proprietary absolute encoder it comes with.
I’m going to remove the high resolution Mitsubishi encoder, and mount a CUI AMT-102V encoder (very basic compared to the Mitsubishi encoder). When ordering the parts, I was unsure about whether the AMT102 would fit. The AMT 102V kit comes with a range of shaft adaptor sizes, but I couldn’t find out the encoder shaft diameter of the Mitsubishi. I opened up the back of the Mitsubishi motor and removed the encoder unit, hoping I would be able to mount the new encoder.
The optical encoder wheel was glued onto the shaft, so I used a heat gun and melted it off. It has a 5mm shaft, which was ideal for mounting the AMT102. The mounting holes for the encoder on the motor housing were a bit narrower than the AMT bracket, so I had to enlarge the AMT bracket mounting holes, and cut a couple of notches out of the AMT102 alloy housing to allow it to fit over the bolts. I was a bit disappointed with the build quality of the CUI encoder – the plastic is rubbing. However, it is only a $30 encoder, so I can’t ask too much. The resolution is 2000 PPR, which works out at 8000 counts per revolution. Initially it was producing a slightly noisy signal, this was easily fixed by connecting the encoder ground to the motor housing. The AMT encoder works well for direct drive.
The picture below shows the AMT102 mounted on the mitsu. The original encoder cover fits nicely over it, so it looks tidy when it’s all sealed up. I used a 5-pin DuPont connector to connect the encoder to the Ioni servo drive, using cat 7 network cable. I used the spare 3 wires to run an e-stop signal for the Ioni, a force LED from the Discovery to show FFB clipping, and a potentiometer to adjust torque on the fly with the anain pin on the Ioni.
The external sockets used to connect the encoder and power cables are proprietary on this motor as well, so I made some small wooden pieces to hold a plastic cable gland, which holds the wires nice and secure to the motor. I’ve made the cables detachable on the servo drive end, and I’ll leave them connected to the motor. This saved me from spending $100 on a new set of plugs and sockets to mount on the motor. The timber case alongside the motor contains the power supply, Granite Devices Ioni servo drive, and the STM32F4 Discovery USB controller.
Update: After using it for a while, I can say this motor is working great. It has enough torque to make my arms very tired. There is no cogging sensation. The bearings are buttery smooth. Its feels nicer to turn than my real car, so solid and smooth. I’m glad this motor doesn’t weigh 12kg like the Mige, it is much more compact. It is a very high quality motor, and an absolute bargain for what I paid on the bay. New, they can go for $2000. I paid a tenth of that.
Note: Mitsubishi also made the hc-sf102, which is an older model which seems very similar. I’m not sure if the encoder shaft will also be 5mm diameter.