My power supply arrived today. I picked it up relatively cheaply ($243.56 AUD to my door) off eBay to run the 36V motors I have.

Problem is... it's a 48V supply, not 36V. They had 12V, 24V & 48V models, and I thought I'd better get a larger one and tune it down, rather than a smaller one tuned up. I've seen a number of electronic components are rated at around 30V, and if I took a 24V model with 30V rated components and ran them at 36V, I might be pushing the envelope a bit. I think it's safer to get a 48V unit, and try to run it at 36V. The unit was "user adjustable +/-10%", and being a switched mode power supply I assumed that changing the output voltage a bit more (25%) should be easy enough.

How difficult could it be?

On opening the case, I found a lot more than I expected. I guess the extra componentry is for the extra features (Auto on/off cooling fan, Built-in EMI Filter, Over Load and Short Circuit Protection, Over Voltage Protection, Auto-recovery after protection, etc). Luckily the user voltage adjustment was easily located, and tracing that back along the PCB resulted in a fairly straight forward couple of components before the signal hit a riser card full of epoxy and SMDs. I'm taking a guess that the riser is the control smarts.

 

Tweaking the pots changed the output voltage from 44-52V but the voltage to the riser was always 2.49V according to my DMM. Based on this, it uses a simple voltage divider and the control circuit attempts to maintain 2.5V on the sense line.

Assuming the pots are centered, 2.5V at 2K85 needs 48.11V at 54K85. For 36V to give 2.5V a total of 41K04 is required. Replacing R2 with 37K19 would give 36V. 39K is the closest E24 value, and that would give 37.6V. With a little adjustment, 36V should be possible. I fired up my trusty soldering iron & swapped it in. Hold breath... switch on...  tweaked the pots... voilà! 36V :-) I've put one of the motors on the supply, and it kicks like a mule during take-off while voltage holds at a rock steady 36.0V. I don't know what effect dropping the voltage has on its ability to deliver current, but if it could give 20A at 48V it should do at least as much at 36V. If it can maintain the same power output, 27A should be possible.

Without knowing too much about the workings of this power supply, I appear to have changed it from 48V to 36V. I don't underestimate the complexity and there's probably more to it... but for the moment, it appears to work.

I'd like to introduce you to my new baby. I've always wanted one of these... but they're just stupidly expensive. Tool shops sell the trolley for $600+ and the six drawer chest for $250+.

There's an important distinction to make when looking at chests & trolleys like these: slides (the runners that the drawers are on).

There are alot of cheap trolleys and chests available now but they all skimp on the drawer mechanism. Often called "friction" or "smooth action" slides, the drawers drag, skew sideways, stick and rattle. Yuck!

Friction slides - Cheap & Nasty

If you want to load up the drawers and not attack the box with an axe you'll need ball bearing slides... and that's going to cost you. 

Ball bearing slides - Expensive & Nice 

I'd been looking for a while, and had found the contender, a trolley and chest combo for $400 (+$50 shipping). I'd been watching some nice Supatool pieces on online auctions, but as is common with auctions... prices were higher than you could buy else where if you looked around.

Whilst strolling thru Bunnings one day, there was a pallet with brown cardboard boxes... not well marked... and $399 written in chalk. I recognised the SKUs from an auction I'd been looking at (they sold for way too much). It was a 3 pack including trolley, extra drawers, and chest. Bargain. So, I got one. I keep thinking I should have got two. (In fact I did get two, one for a friend, when I bumped into some remaining stock at another store).

Now I just need to load it up...

I've been playing with my little side project, and that involves Atmel AVR microcontrollers. I love the Atmel AVR family because they're simple and compiler friendly. Friendly enough that even GCC supports them... under Windows too (WinAVR). I'm writing some code to generate PWM signals to drive brushed DC electric motors. Now, these great little chips often support built-in PWM generation. In fact, the particular model I'm using supports six hardware PWM channels. Call me a crazy, but I decided to ignore this perfectly good hardware feature and flip the output pins on and off in code.

In doing so, I've got a set of raw output variables which I convert to a PWM structure. When I call the conversion function in my main loop, it works fine. When I call the conversion function from a interrupt handler (connected to a timer), it doesn't update. WTF?

It's the little things in life... hours spent scratching your head about a simple code change. "Hmmm... works here... not here. Okay... Interrupt handler is definitely called."  A trip to AVR Studio's emulator and setting break points on the lines where raw outputs were changed was the key. Quite simply... I couldn't.

When the call that uses these output variables was only within an interrupt handler (and not called from the main body), GCC would optimise away setting (and even calculating) the output variables. My structure was never updated because GCC couldn't see where it was ever used. A similar (non-microcontroller) example is given at over at LinuxDevices.

Adding the volatile keyword to my variable declaration fixed the issue (yay!). Is there a performance impact? Certainly. Enough to worry about? Not yet.