We’ve seen lots of power supply projects that start with an ATX PC power supply. Why not? They are cheap and readily available. Generally, they perform well and have a good deal of possible output. [Maco2229’s] design, though, looks a lot different. First, it is in a handsome 3D-printed enclosure. But besides that, it uses a TFX power supply — the kind of supply made for very small PCs as you’d find in a point of sale terminal or a set-top box.
Like normal PC supplies, these are inexpensive and plentiful. Unlike a regular supply, though, they are long and skinny. A typical supply will be about 85x65x175mm, although the depth (175mm) will often be a little shorter. Compare this to a standard ATX supply at 150x86x140mm, although many are shorter in depth. Volume-wise, that’s nearly 967 cubic centimeters versus over 1,800. That allows the project to be more compact than a similar one based on ATX.
The project is nicely documented and has features including colorful meters and USB ports. An import buck converter module give you a lot of options on output voltage.
The original build had a top and bottom plate made from plywood, but the current design files have STL files for those parts, in case you prefer to print them. Be prepared for a long print though. [Maco2229] says that even though he prints between 100 and 120 millimeters per second, printing at 0.28 layer height took about 14 hours.
We’ve seen plenty of ATX builds. Some modify the power supply while others don’t.
“As well, this is not laboratory power supply. I don’t know the level of noise or any other parameters which is required for lab PSUs, so don’t even think about this thing that way.”
At least this designer was smart enough and honest enough to provide this caveat. Should also be said that the enclosure can be a fire hazard given the available current and VA.
Story time. Nephew was a post-grad working as the resident code monkey and electrical designer for a university bio lab. The biologists and botanists and whatever put together several ATX power supplies for the analog front-ends between the sensors and data acquisition equipment. The scientists complained about the poor signal quality out of my boy’s signal conditioning circuits. My boy told them them about PSRR and other such stuff, but they insisted that their power supplies were quite wonderful. It was all messy fun and games until the ATX did not have the required min load and went crazy and killed some sensors and instruments. They fired my boy and put some other post-grad slave labor in place to do lab instrumentation, but nothing improved. Almost six years later, my boy is on the interviewing committee that screens these ‘brilliant’ PhD life scientists. He was ignored by his boss, and his employer hired the two “scientists” that had fired him. His employer had to terminate both hires (for cause) less than a year later.
Lessons to be learned.
1. Life scientists oft have poor understanding of physics and math.
2. Life scientists oft do not know that they have poor understanding of physics and math.
3. Life scientists should never be allowed to prank lab mates because they are oft too ignorant of how to determine long-term resultants.
4. Life scientists and physicians should always be supervised by engineers and physical scientists.
5. Bad power supplies cause bad things to happen.
I work in a physics lab and I don’t know anyone stupid enough to use a switched-mode supply for something sensitive – we once had to run an experiment off a car battery because we wanted low-noise DC. I did convert an ATX supply to run a Peltier-based cloud chamber for science outreach and it worked fine for that.
My undergraduate degree is in systems programming. Fresh out of college, I was the scientific programmer for a biostatistics lab. They were brilliant as biologists / statisticians, but as coders, not as much. The code I received was looking for statistically enriched sequences over the entire human genome using nested for-loops to generate the search sequences, and doing string comparisons… it would run out of memory and thrash even on the group’s relatively beefy HPC, and it took >48 hours to run. I rewrote it to encode and test the gene sequence as an integer (2 bits per base pair) and replaced a multi-gigabyte table of strings a much smaller table of integers. Then it would run on 45 minutes on my desktop.
They were pretty happy with that. Good times.
This is itself a research field, which made me feel much better in grad school: “end user software engineering” – people writing code to get their job done, not because their job is to write code.
In general, I don’t agree with generalizations :-P
No, really. It’s not about unserstanding physics and math (which many life scientist have no problems with), but but about understanding your *requirements*. So, an ATX power supply may be “perfect” for some problems and, at the same time, bad for others.
Side note: has you nephew considered seeking help with his communication skills? Having one’s opinion being heard is a complex art that requires as much logic as persuasion, something that is difficult for many among us, the more technically inclined.
I was a research assistant in a Behavioral Science Lab in 64-65. Everything you say is true. It takes a Behavioral PhD to “require” you to do something which involves the repeal of the known laws of physics.
More and more projects focus more on the 3D printed enclosure part than on what is inside this enclosure.
Here we have a nice box and nice displays, but as very well said above, low-cost switching-mode supplies are not the best choice inside any serious lab.
Found this discussion quite interesting, as when I get the time I’ve been working on a hybrid power supply architecture that combines both good efficiency and low noise for use as a lab power supply.