What a monster. Not the biggest piece of equipment I've run across, but weighs a ton. Nice heavy construction all around.
We open it and find...paper?
Yes, really, lots of paper. Heavy non-flammable paper - have seen similar stuff used before (under the name Nomex I think?) but didn't expect it in such quantities. Check out the nice folding and tab-in-slot here...
...and the random little bit added here...
...oh, and some underneath the fans as well:
Ok, let's get that large sheet out of the way and take a look underneath. The inside looks something like this:
Lots of fans for everybody in here. Two big side-by-side fans at the front, one to blow air over the fins of those big black blocks which are the heatsinks, and one to blow it over the components in the whole middle section. There's also one on the side which blows air over the magnetics.
Is it just me, or does the entire mechanical design around directing the airflow seem really hack-ish, between the many bits of paper and the weird piece of bare FR-4 over part of the heatsink? It feels like something which the engineers spent a day patching together with tape in the lab trying to contain the airflow just right, and then which got immediately put into production with zero DFM in-between. A one- or two-piece molded plastic shell to drop over the top and screw down feels far more professional. Then again, maybe these were made in such low volumes that it would've been more effort than it was worth? Maybe their one mechanical engineer was out on maternity leave? Maybe my opinions are way off-base for other reasons?
Power flow, from input to output:
* AC power comes in at the back-left; there's a bunch of what looks like EMI filtering in the form of common-mode chokes wound on triple-stacked(!) toroids, some big blocky film caps, a relay (likely for using that big power resistor as an inrush limiter), a bridge rectifier, even more big inductors, then a couple MOVs and a bank of 10x 1,800uF 200V bulk caps, for 18,000uF total.
Crazy amount of energy storage there. Since there's no power transistors over here I'm 99% sure this thing doesn't have a PFC stage, just straight peak-charges this bank of bulk caps from the AC line. No wonder it needs an inrush limiting resistor. The inductors following the bridge rectifier also probably help slow down the peak-charging, turning each current spike into a slightly wider, lower current spike (this will give it a not-quite-so-awful power factor as well as not absolutely hammering the bridge rectifier with huge peak surges).
Here's a close-up of the bridge rectifier:
...and here's a close-up of those triple-stacked toroids:
* The board ("MOSFET board") at the front end on the other side of the heatsink holds 4 power MOSFETs behind it and some film caps.
At least some of these caps (2 red ones?) probably provide the local buffering, but I'd guess the others (2 round yellow ones?) sit in series for DC-blocking. It looks like on the main control board just underneath the "MOSFET board", there's spots for 4 more MOSFETs, probably placed in parallel for a higher-current model in the same series.
With the possible-DC-blocking caps, this could either be a full-bridge converter or a half-bridge converter with each set of 2 MOSFETs paralleled. Maybe it uses the series caps for resonance.
* Next up are some large, heavy, and super-crusty magnetics.
I'm getting cancer looking at these things. I think the left one is a transformer (which also provides the output isolation), and the 2 right ones are output inductors. Two wires run from the "MOSFET board" to the transformer; two more wires exit the transformer and go to another board right next to it, which I'm going to call the "rectifier board", as 4 output diodes in large TO-247-ish packages are pressed up against the heatsink behind this board, which sit there full-wave-rectifying the output of the transformer.
This board also has some power resistors (white rectangles; it's metal deposited on a ceramic base) and some small inductors and orange caps.
The resistors are clearly labeled as 270 ohms, which is far too large to be doing output current sensing (no other connections to get out a current sense signal anyways). With some close inspection and a multimeter I also confirmed that inductors are one single continuous winding (not transformers), despite the groups of turns - they also sit between the incoming secondary of the transformer, and each diode. This, together with the 270-ohm power resistors and the caps nearby, leaves no other real option but that the diodes have an RLC snubber like this:
Never run into those before. The series inductance slows down changes in current, which may be useful here to suppress current spikes from the diode reverse recovery.
* Two wires with the rectified power exit the rectifier board, take a few turns around a shared free-hanging toroid (common-mode choke for the output) and run to each of the large inductors. After passing through these inductors, the output power now goes to the "output board" against the back panel, which has a whole bunch of capacitors; two big bulk ones and a surprising variety of ceramics including an interesting yellow DIP-package-like capacitor array.
* There's an additional board mounted against the opposite wall of the "powertrain area", which has a big power transistor mounted behind it plus a 10-ohm power resistor. There's also a 358 op-amp on board, a ribbon cable to the output board/main board, and heavy-gauge wire to the output board.
The little of the traces I could follow showed the power transistor and the resistor connected like this:
I'd guess that this is a linear current source, with the op-amp driving the power transistor and regulating its drain/collector current based on the voltage it senses across the power resistor. The current source seems to be put across the DC output: probably does a controlled discharge of the output caps when the output voltage setpoint is reduced and at power-down.
Overall, the powertrain schematic looks like this (with the caveat that it may be a MOSFET half-bridge instead of full-bridge):
Main control board looks pretty generic, has a bunch of standard op-amps, comparators, etc. plus a bit of digital logic and some standard SMPS controller ICs.
The front panel has a couple ICL7107s, which display the voltage and current readouts - these are convenient all-in-one ADCs which interface directly to a set of 7-segment LED displays and handle everything all the way from the raw voltage input to driving the displays directly. This is not a fancy modern digitally-controlled power supply - the pots on the front just change the setpoints for the onboard analog control, and the front panel displays read out whatever you happen to get; one of the buttons on the front does do a "press to display setpoint" function though, so it's way better than most older supplies in this way (as in, you can see what the voltage is supposed to be
before you turn on the output).
There's also a lot of temperature sensing in here. A temperature sensor on the metal wall just above the input bridge rectifier feeds a little bit of circuitry near the 18,000uF capacitor bank.
A temperature sensor near the MOSFETs has to a cable that connects to the main control board, and the traces run who-knows-where from there.
Another temperature sensor sits on the heatsink near the output diodes, and it connects through a cable to sensing circuitry (an OP07E DC-precision op-amp) living on the output board.
Glad to know the supply has your back to some degree and won't just happily chug along as the power devices reach their melting points.
Haven't actually used this supply much as it's wired with a circular plug and needs 208VAC I think, so there's only one place to plug it in. There's better ones like the Chroma DC source I showed before, so this Xantrex mostly gets broken out only when there's a test equipment traffic jam going on.