ssn circuits

Thursday, March 29, 2012

Blah! And this is why I should publish all of my work!

!!!! I need to post my work more often!!!!

If you haven't read the Hack-a-Day link above, "Mike" posted a video about modding an Atten APS3005S power supply to quiet the always-on fan. He changes out an in-line zero-ohm jumper for a thermostat switch (probably the usual cost reduction nonsense that happens, but see below about the new version of this supply). He also "upgrades" the fan to an ultra silent one (not entirely sure if has similar CFM, though, which is fairly important). If you're going to do this, make sure it's got at least as much airflow, or better yet, test the power supply under a fairly beefy test load and make sure none of the primary components (transistors, transformers) exceed a reasonable temperature at full load. After all, this is a bench supply in a test environment, and will be taking quite a bit of abuse from your experiments.

Worked on this problem a few months back because I got really annoyed. Everything the other poster said is more or less completely true--this power supply is loud. Oh, and a clarification: the CircuitSpecialists CSI530S supply is equivalent to the Atten described in the post. I went along another approach, though. The one I had was actually someone else's that I borrowed for one of those projects where you end up needing 3-4 different power supplies for one reason or another. Problem is, that meant I really didn't want to mess around with the board too much, lest I accidentally destroy the power supply, so I spliced the control board inline with the fan power wires. I also wanted to be extra safe about not overheating anything in the supply because, well, it wasn't mine.


So I designed a somewhat simple little redundant thermistor / comparator based circuit. You can see in the two tape locations a few images down where the thermistors are mounted--directly on the two transistors. The design of this particular supply parallels two 2n3055 (?) TO-3 power transistors, which are pretty common in cheap linear bench supplies, to achieve an overall 5A current rating, so I attached one thermistor to each one. This adds some redundancy (in case one thermistor fails) and in case one transistor heats up significantly more than the other (bad matching / part-to-part variation). These went into a simple comparator-based (LM393 or equivalent, don't remember exactly which one I used) circuit with hysteresis, which were then ORed together with diodes and then drove a jellybean transistor to drive the fan. The hysteresis, of course, means that the fan only turns on when it exceeds a higher temperature, and will not turn off until the temperature falls to a lower threshold. I think I set these to roughly 50C and 35C. I recall there being some trickiness involved because the LM393 / equivalent ICs are open-collector, and this makes doing an diode-OR, along with hysteresis and the transistor driver circuit, fairly difficult. If you're interested in


Anyway, here are some pictures. The supply works wonderfully now--at low currents (<0.5A-ish, for most voltages I've used) the fan doesn't turn on at all. Otherwise, it roars, but I guess I'm cool with that for that much power. It turns out most of the cost in many linear power supplies comes from the large heatsink necessary; this linear bench supply pulls off having 5A output primarily by replacing that large heatsink with a wimpy metal plate and sticking a huge fan on it, which might not be optimal, but definitely how they achieved the $80 price point with 5A.

Thermistor mounted on one of the power transistors, with thermal glue and Kapton tape for wire strain relief.
The control board, spliced in between the fan power cabling and mounted on the rear with double sided tape.

Other important notes

As I said before, I did this project a few months ago. The power supply pictured is actually in New York now, so I can't physically get access to it. I've also lost the schematics (well, I never saved them in the first place--this was one of those <1-hour design-on-scrap-paper exercises, and I figured it wouldn't be interesting enough to save), but if anyone wants to reproduce the circuit, ask me and if there's enough interest I'll try to redesign it.

What would I do if I did this again? Well, there's one more significant heat source that might need to be protected--the transformer. While it's not directly blown at by the fan, it's definitely in the path of the airflow. Might make sense to have another thermistor there. Additionally, a cheap PIC10F/12F (or ATTiny for you Atmel guys) would've probably made the whole endeavor a lot more flexible. I recall not doing that because I was in the mood for some analog and figured it'd be simpler than pulling out a programmer.

For some time I was considering publishing this and making mini PCBs because a few other people I know did a group buy of these and they all have extra-loud power supplies. Unfortunately, I soon found out that this was unnecessary: the new model Atten power supplies added the fan temperature control feature. I ordered a couple of them off of SRA Soldering Supplies for a club, and found out that all of them already had built-in fan control.

Anyway, if you've got any questions, just ask. If I find time I'll try to answer them as best I can.