cooling concern for AX 3500

The AX2550's enclosure is a heatsink, but the AX 3500 does not come with one. When the 3500 controller is used, does it have to have a heatsink fabricated for it? I'm assuming there might be a few components on it that will heat up quickly like a computer processor and a fan alone won't be able to dissapate the heat fast enough. Is anyone using this thing with a basic plastic or cheap handmade enlosure?

on the closeup page; dev.roboteq.com/dev1/ax3500closeup.shtml
there is a Perspective image of the controller. if you look under the board where the power transistors are, you will see the heat sink. in the manual, they recommend that one mounts the 3500 in a vertical position lying on its side so the heat sink can use convection currents to assist in dissipation.

Heat management on the 3500 is much more important than on the 2500. I tried one enclosed in a basic sheet-aluminum housing where my current draw was between 10 and 40 amps. The unit overheated within a few minutes. The thermal/current reduction system works pretty well to keep the magic smoke inside the ICs. I then tried a small fan to keep the air moving. This increased the running time by three or four times but it would still overheat.

Now, my original goal was to reduce the cost while still maintaining continuous operation in desert heat. Unfortunately, by the time I added enough cooling and miscellaneous hardware, the cost savings wasn't much so I'm back to the 2500. It's really a shame because the 3500 design is so nice and it seems to spit out much less EMI. If the Roboteq boys would make a 3500 that could handle higher temps without external cooling, I'd standardize on it.

Hello all,

Since somehow this post is still alive, I figure I should follow up with our results.

I used a plastic tupperware box and an aluminum piece of sheet metal I machined on a mill with pc motherboard standoffs to mount the Roboteq on. It worked just fine.

My OU senior design team blew away all of the rest of our classmates and took the prize during our performance evaluation for graduation. I would recommend this controller to anyone, especially to students who aren't experienced with controllers. Thank you Roboteq for the excellent technical support and this discussion board.

Basic sheet-aluminum sounds like the material is/was much too thin to adequately dissipate the heat, thus forcing the rest of the circuitry to work harder.
The extra EMI is the most likely the result of the 3500 not being sealed up inside a metal box.
The 2500 will be cooler because it does not have to work as heard to deliver the same current.

I have an Electric assist Hub Motor on my bike, and whenever I accelerate hard, I can hear, and feel the motor soaking up all 20A from its controller.