Kikusui PCR500M AC source teardown

[D Straney]

To complement the wouldn't-want-to-lift-these-all-day Chroma AC sources I looked at before (https://www.eevblog.com/forum/reviews/a-tale-of-two-ac-sources-part-1-(chroma-6408-teardown)/ and https://www.eevblog.com/forum/reviews/a-tale-of-two-ac-sources-part-2-(chroma-61501-teardown)/ ), here's a much more portable comfortably-benchtop-sized little guy from Kikusui.  Voltage and current ranges are comparable to the Chromas, and it can put out 500VA compared to their 800VA and 1000VA ratings, which is not bad at all considering how much smaller it is!  It's also a good deal cheaper as well (and not just the shipping).  Some of the disproportionately-reduced size could be from less overhead, closer-to-the-limits design (the internal magnetics do like to buzz, especially with weird loads like phase-cut lamp dimmers) but some of it could also just be that it seems more modern than the Chromas and saves some control circuitry space by going all-SMT, as you'll see further down.  Internally there seems to be a good deal less energy storage than the Chromas, and I'm pretty sure it wouldn't be able to deliver the same kind of turn-on-at-peak-voltage surge current, for example.  The UI is a little weird too, but overall, these seem like good value for what they are.

All-important outside view, with Standard Reference Hand for size:

After pulling off the top cover, it's PCBs as far as the eye can see - there's a power board on top...

...a view into the guts from one side...

...and a control board taking up the other side:


With a few cables disconnected, the top power board (which turns out to be for the inverter) can conveniently fold out:


The overall structure is something like this: the bottom ("DC board") contains input EMI filters + bridge rectifier, an auxiliary supply for all the logic/gate drive power, and most importantly dual half-bridge converters which provide two isolated DC voltages.  It's controlled by the small right-angle card coming off of it.  The two isolated DC outputs go to the inverter board, on top, which contains two independent H-bridges; the outputs of these can be connected either in series or parallel to the output terminals, depending on whether higher output voltage or current is required.  The control board sits on the side and runs the show, mostly the inverter stuff and the front panel controls.  It's actually almost exactly the same architecture as the Chroma 61501, except that it's only 2 stages instead of 3 - the Kikusui doesn't have the Chroma's additional boost converter between the AC line and the dual-DC-DC stage.  That probably helps with the size too; must've gotten a wide enough control range on the half-bridges that they can do some reasonable PFC as well.

DC board
Here's where it all begins.

The AC line input enters at the bottom-left, where there's a fuse then a few stages of EMI filter inductors and film caps running up the left side of the board, to a relay (not sure why, to be honest) and a bridge rectifier on the small-ish heatsink (which incidentally is turned so that the fins gets minimum benefit from the airflow of the giant fan right next to it).  The top and bottom halves, the two half-bridge DC-DC converters, are almost mirror images of each other.  Each has a couple huge film input caps, two transistors against the shared heatsink, a transformer, then a couple output diodes and a pair of bulk capacitors.

Here's some not-very-exciting proof that, yes, that really is a bridge rectifier on the small heatsink:


...and here's a close-up of one DC-DC primary...

...and secondary side:


Those power resistor and disc cap pairs by each diode on the secondary side are almost definitely RC snubbers, while the inductor hanging out near the diode/resistor/cap grouping on the primary side either could be added leakage inductance to the transformer for ZVSing the MOSFETs...or could be part of some kind of interesting energy-recovering snubber.  It's also worth noting that each transistor seems to have its own fuse, judging from the silkscreen labels.

There's a bit of circuitry near only one of the DC-DC secondaries, which does who-knows-what, maybe some OVP or a discharge crowbar or something:


The control card for the DC-DCs has a UC3854 generic SMPS controller, some op-amps for sensing/protection I'd guess, and what looks like two gate drive transformers for driving the two half-bridges:


Auxiliary Supply
This lives at the front-panel end of the DC board:



Has its own EMI filter and bridge rectifier; looks like an all-in-one IC driving a flyback with 3 outputs, one of which goes through an extra linear regulator.

Inverter Board
The two DC outputs from the DC board go up through a pair of short cables to the inverter board, where the sinusoidal magic happens:

Like the DC board, the dual power stages here are mostly mirror images of each other.

There's a bunch of control circuitry over at the left:

This mostly seems to be digital logic with maybe some basic analog here and there, so I think it most likely takes PWM signals generated by the control board, and just applies some dead time or phase shift logic to create the final gate drive logic signals.

Next, there's a pair of supplies which create isolated gate drive power for each of the 3 "zones" on each H-bridge:

These are super-simple self-oscillating push-pull/"resonant Royer" supplies - no serious regulation needed usually for gate drive power, and not having a drastically-variable load on the output along with a likely-regulated input helps too.

You can see the results of having actual isolated gate-drive power, as each H-bridge MOSFET has its own optocoupler and gate drive chip, as opposed to the Chroma way of using gate drive transformers for everything:

The optocouplers like they're Avago ACPL-M60L's, much faster than your typical slow-phototransistor-output optos with a photodiode and detector amp living on the secondary side.

There's also a temperature sensor on each heatsink:


After each H-bridge, there's a series of filters for the PWM output: first a common-mode choke and a capacitor, followed by another (yellow-tape) common-mode choke...

...and its capacitor, followed by a 3rd common-mode choke and some final output capacitors:

The final caps have some discharge resistors and what may be a small "artificial ESR" resistor for damping resonances:


Partway through that filter chain is what looks like over-current protection:

No serious measurement going on here, but just some current-sense resistors with the photodiode of an optocoupler stuck across them: when the current gets high enough that the current-sense voltage reaches the forward voltage of the opto's LED, the LED will turn on and it'll give the "oh shit, shut it all down" signal back to the control logic.

At the output of both filters are two relays, which I'd bet money on doing the series-vs.-parallel combining of the two H-bridge inverter outputs, and a pretty large current-sense resistor:

The current-sense resistor has a dedicated amplifier nearby to boost the current-sense signal before running it across half the box:


Output wires then run to the rear output connector, which has a few extra caps stuck on it, for that last little bit of EMI suppression:


Control Board
Finally, the brains of the operation lives along the right side:


Near the left is a TI TMS320 DSP, which probably handles all the sinewave generation, PWM, and RMS math.  There's also a Holtek HT16511, which is slightly strange as it claims to be a VFD driver, and the front panel definitely has 7-segment LED displays.  Looks like it has general-purpose open-drain outputs though, so it probably gets used just for the keyboard scanning / segment mux'ing and the VFD-specific stuff gets ignored.  At the right side is an H8 processor, which I'm guessing gets used for the high-level control and interfacing.  The two large SOICs in the middle are the AD73311, which are ADC-DAC pairs with programmable input and output gain - these must be useful for sensing voltages/currents and generating setpoints (fast over-current comparators, and the voltage-regulation setpoint for the half-bridge DC-DC outputs).

There's also an empty option slot, with no access from the back panel:


It also has a mystery daughterboard:

All I could see underneath the daughterboard were what looked like SMT resistors, and the daughterboard didn't have any components on the back side, so my guess is that this is a PCB capacitor for creating a compensated voltage divider that senses the output voltage.  I saw something similar in a Yokogawa power meter; the very end of my post on that has some discussion of "how" and "why".

[Dexter2]

Thanks for a nice tare down. Just added the front panel PCB picture. I had to replace the tactile switches on mine as a few had become non responsive with others getting poor on an quite old unit...
Fitting replacement switches with LEDs are: TE Connectivity FSMIJ63AA04 and can be purchased on Mouser or Digikey.

[NiHaoMike]

For the hobbyists on a budget, I wonder if a good enough substitute would be a subwoofer amplifier plus a transformer of some sort to step up the voltage. The automotive amps are particularly capable of handling low impedances and will readily run from surplus server PSUs.