A Tale of Two AC Sources, part 2 (Chroma 61501 teardown)
We're back now for a look at the newer of the two Chroma AC sources, the 61501. (Here's the first one:
https://www.eevblog.com/forum/reviews/a-tale-of-two-ac-sources-part-1-(chroma-6408-teardown)/)
So as I mentioned before, the 61501 has a lot more features than the 6408, including DC output capability. Let's take a quick look at the front panel again:
...but quickly move on to the part that matters - the guts!
Against the front panel (at the left) is a display-and-controls board. Behind two pretty large fans are two sections: each actually has two stacked boards. The top section here has the control board visible, while the bottom section here has the inverter board visible.
Control BoardThis is the control board:
The revision date on the silkscreen text says '06, and various date stickers say 2014, but it looks kind of old by '06 standards, being about 2/3 through-hole components. Could be an incrementally-updated older design, or the Chroma engineers may just not like change (which may be justified, if they have no trouble sourcing parts and already have an in-house wave soldering line or something similar).
A lot of the board seems to be dedicated to analog sensing circuitry. However, there's a TMS320 DSP near one edge:
Next to it are 3 AD73311 ADC/DAC pairs. These seem to have been originally designed as speech CODECs, where a 64ksps 16-bit delta-sigma DAC and ADC are paired with a PGA each for gain control; however, now it's marked on the Analog Devices website as a general-purpose analog front-end, so they must've realized that these were useful for anything relatively slow but high-resolution. The Chroma probably uses the ADCs to sense voltages and currents for the DSP, and the DACs to generate setpoints and threshold voltages from the DSP for some analog controls.
There's also an XP Exar ST16C550 dedicated UART chip with 16-byte receive and transmit buffers: the datasheet I found was dated '94, which says something right there. UARTs are all integrated in processors now as on-board peripherals (or if one has special requirements like a large buffer memory, then probably implemented in a small FPGA or something).
There's also a mystery option card which plugs into the control board with an edge connector (but which doesn't actually expose any ports on the back panel):
It holds an analog switch array, a few op-amps, and a multi-channel multiplying DAC. I hadn't heard of multiplying DACs before, where instead of feeding a fixed reference voltage to the digital resistor or current source array, an input signal can be brought in instead. This effectively multiplies the input signal by the digital DAC value, which makes a sort of programmable attenuator/amplifier. These seem to be used for digital gain adjustment and calibration, like an older substitute for where a digital potentiometer might be popular instead now. I have no idea what this option card does though.
PFC BoardLet's dig down now to see where the AC line input enters, which is on the same side as the control board. Getting the control board off was a serious pain, as it had both a horizontal connection to an interconnect-backplane, and also a vertical connection to an I/O board underneath it, as you can see here:
A whole metal panel comes out, with the I/O board (to back-panel comms connectors) in a box:
Anyways, with that out of the way, here's what turns out to be the PFC board:
You can see the connector where the AC line enters at the bottom-left edge. There's an EMI filter with lots of nice green components (
color-coordination matters, folks), a cluster of relays (at least one is almost definitely to bypass those square white power resistors for inrush limiting), and an auxiliary supply at far right for the analog/digital controls, with linear post-regulators on the multiple outputs just like in the 6408.
Power devices are sandwiched underneath heatsinks and hidden from our view, but we can make some guesses. The bottom-left heatsink, for example, has the bridge rectifier for the incoming AC line:
(See that "DB" label?)
The mysteries that lurk beneath the other heatsink aren't hard to guess either. Control is by a
UC3854 high-power-factor preregulator:
Now enjoy some...interesting...rework examples while you ponder:
Even without seeing the power devices we can figure this one out. Power-factor correction? Single inductor (the size of my hand) with no visible isolation or multiple windings? Inrush limiting necessary? It's got to be a boost converter.
So that's what's on the PFC board, a high-power PFC boost converter and the auxiliary supplies. On the output, there's a nice large bulk capacitor and a common-mode choke for EMI filtering, then a cable carrying that DC over to the other half of the instrument:
DC-DC BoardThis is the point where we pretend that the inverter board temporarily doesn't exist, and instead dig down underneath it to see the other board stacked below. This one has what looks like an isolated full-bridge converter, with 2 independent outputs:
There's the output connector over at the left, and a cable which comes down from the inverter, but we're going to ignore that output section for now until after the inverter's covered.
Now see how the cable comes in from the PFC board at bottom-center, and goes to a couple bulk capacitors, a transformer, and a set of power devices mounted on the heatsinks on either side. There's also 2 current transformers, for what reason I'm not sure. Controls are just left of this converter, with a generic UC3xxx controller in the DIP-16 package, and a couple daughterboards mounted vertically. There's also a hexfilar (?) transformer, likely for gate drive:
Anyways, on the output of this transformer is a set of 8 diodes (2 full-bridge rectifiers' worth), and a symmetrical set of 2 EMI filters and 2 bulk capacitors, which really hammers in the fact that this DC-DC supply produces two identical outputs:
The two DC outputs now connect to screw terminals, and through those to a pair each of really long standoffs (!), which run upwards to the...
Inverter BoardThings are going to start making a lot more sense soon.
See those weird little daughterboards with the big white boxy power resistors? Those attach to the standoffs where the 2 DC inputs come in:
Those resistors (30K each, 3 in parallel for 10K total) sit across the DC inverter inputs, most likely to discharge the capacitors on power-down?
Anyways, look back at that overview. The two DC inputs comes in through the standoffs (which has to be the most bizarre but also kind of reasonable way I've ever seen power transmitted between two places) near the left, get buffered by the two giant blue aluminum caps, and then power the two mostly-symmetrical H-bridge inverters; one in the top half, one in the bottom half.
The control section of the inverter looks like the 6408: some digital logic (likely dead time generators or something similar), gate drive transistors which feed the gate drive transformers for the inverter's power MOSFETs, and a comparator and op-amp at the right (probably for over-temp sensing and fan control). The actual PWM signals seem to be generated on the control board, either by one of the many mystery chips (some with labels could be programmable logic) or maybe by a peripheral on the TMS320 DSP itself.
Each inverter has 4 2SK1518 MOSFETs, the same 0.25-ohm 500V model used in the older 6408 too:
That MOSFET-side gate drive circuit looks really similar to the one I rambled about in
painful excruciating detail for the 6408, except that there's now zener diodes too, and ferrite beads for noise suppression on one lead of each smaller transistor.
The only part of the inverter lacking in symmetry are the two bipolar transistors tucked against the heatsink on one side only:
I would bet many beers that these are used for the exact same purpose as in the 6408 - one for fan control, and one for heatsink temperature sensing.
After the H-bridge, each inverter has a differential filter consisting of a multi-stacked-toroid common-mode choke, a capacitor (with series resistor for damping resonances), and another common-mode choke:
(That picture was from before I realized that the plastic airflow-guiding cover over the inverter had removable screws)
After the filter, the inverter's outputs now run down to the connectors at the back side, and through cables (not more standoffs?!) back down to the output section on the DC-DC board underneath.
DC-DC Board (revisited)
Those two inverter-output cables come down, go through a couple relays and some more EMI filtering, and then finally to another pair of relays (barely visible at the bottom edge here) and the output connector at bottom-left. There's a 20-milliohm resistor which is probably for current sensing.
The second (final) pair of relays are almost definitely for disconnecting the output, but what're those first two relays for? Well, this whole section is reminding me of the dual secondary windings from the 6408's isolation transformer, and its output filter board. Just like in the 6408, the first two relays probably choose series vs. parallel connections for the two inverter outputs, to allow a wider voltage range. In fact, here's the schematic-guess I drew for the 6408: replace "transformer secondary windings" with "inverters", and it's a drop-in replacement here.
Overall ArchitectureSo now, the DC-DC converter with two identical but separate outputs, and the dual inverters, should make sense. This is the replacment for the 6408's isolation transformer with its two output windings.
Let's compare:
Instead of the isolation for the output (as having an isolated output is one of the big reasons to use an AC source) happening
after the inverter in the isolation transformer, the isolation is now happening
before the inverter(s) in the DC-DC stage.
Now for one thing, leaving out the big isolation transformer lets the 61501 output DC, which the 6408 couldn't do. Whether it's actually cheaper or helps in other ways (when replaced by an entire large board full of semi-fancy electronics) is less clear, but still possible it could make a difference.
It bothers me a bit on some level to use 3 whole stages here, with the associated complexity. This may be done for very good reasons that I'd discover only when looking into it in serious detail, but if it were up to me, I'd at least have started by investigating these options first:
* Isolated PFC stage with 2 outputs, followed by the 2 inverters: leave out the PFC boost, use something similar to the 2-output DC-DC stage here
* PFC boost, followed by 2 inverters with isolated outputs: in this case, each inverter would essentially be an isolated DC-DC converter (with a changing setpoint) with active rectifiers on the secondary side so that the output polarity could be flipped at will (or a separate slow-switching H-bridge could be used).
On the other hand, one reason why you might want to use 3 stages is the output voltage range, and for ease of development. This Chroma can output relatively low voltages, and making an isolated PFC stage (as in my first proposal) which can take an input from maybe 50V-180V (or to 340V if it has a universal input) and put out a DC voltage anywhere from 5V to >200V, while behaving well over that entire two-dimensional operating range, could be a tall order. In some ways it makes sense to add complexity by having an additional stage, but having each stage focus on only one job: the PFC stage boosts to a stable, fixed-voltage DC bus. The DC-DC stage takes in only fixed-voltage DC and puts out only DC, but can step down voltages over a wide range. The inverters only have to generate sinusoidal PWM at the chosen output frequency. My second proposal (the isolated inverters) avoids the wide-range PFC issue, but makes the inverters operate over a much wider range of duty cycles to do the step-down on their own, which can lead to pretty tight timing requirements for gate drive, etc. when the minimum duty cycles get very small.
So that's pretty much it for the Chroma AC sources I have access to. There are higher-power models in the same family, and you can see evidence of that in some places on these boards, where there's spots for extra inductors (inverter board), inrush resistors (PFC board), or an extra current sense resistor (output section on DC-DC board)...specifically for the 1kVA model as opposed to this 500VA one. Considering the way the dual inverters share their load nicely when paralleled (because of the high output impedance at the PWM frequencies), I wouldn't be surprised if the higher-power models had different PFC and DC-DC stages, but for the inverters just used multiple copies of this same dual inverter board.
Home
Help
Search
Login
Register
