In the previous post, I mentioned a slight vibration in the clutch assembly.... well, it quickly became apparent that we had to fix this vibration. There was a lot of testing, brainstorming, and trial solutions, but eventually we realized that there was no fix-it solution - we had to cut out the coupler and start again. The problem was in the weld between the shaft coupler and the flywheel. It turns out that the coupler wasn't completely flush and level with the flywheel when it was welded on. Now, considering that the welder knew that this was for a spinning application, I would have thought that he would at least have clamped the pieces together before welding. However, I'll take the blame as I never specifically asked for clamping and levelling. The result is a large "wobble" as the flywheel spins. The term is called runout. Often, the required runout tolerances are in the 0.1mm range or smaller, and we had a runout of 2mm! This, aided with poor tolerances of the shaft coupler on the shaft (very slight play prior to fitting the set screws) made for a very poorly spinning flywheel, and a ton of vibration. We're going to do it properly this time at a machine shop where they will make sure that everythink is centred, balanced and fitted precisely. During the brainstorming process we also came up with some ideas to improve other parts in the system, which we will be implementing as well. The parts are at the shop today and we may be able to do a trial run early next week. At least we're finally moving the right direction.

We also spent a lot of development time lately on the BMS for our LiFePO4 batteries. These batteries have a few unusual characteristics, and it has taken a while to learn their quirks. We have also made a numer of design changes to the BMS circuit boards to improve measurement capabilities and to handle real life charging scenarios better. We are still using a top-balancing system, which is certainly not ideal. However, we are not strictly enforcing the top voltage of each cell. The cells that reach the top first will slow down to let the others catch up, but only within about a 0.1V to 0.2V limit. This, along with a time limit in absorption mode should allow for some top-end imbalance in the cells. Couple this with regular (not sure how often yet) bottom balancing cycles, and we should have a robust system. We had to modify some component values in order to handle the extremely unbalanced state that we started with after we charged a few cells during board testing. The last change in resistor values will be completed early next week, and then we can started doing some actual charge/discharge cycles. This will allow us to test our new controller as well using the test facilities at Electrocraft.

In all, progress is slow, but we are fixing every problem as it arises, and learning valuable lessons as we go.