Let's look at both an older (6408 model) and newer (61501 model) AC source made by Chroma and compare them. Most of the features are the same - you set an amplitude and frequency, and it'll generate a sinewave for you (500 VA for the 61501, 800 VA for the 6408). They're good for either generating something not easily accessible otherwise like 400 Hz aircraft power or a different country's line voltage/frequency, or alternatively for just having a clean and protected (isolated, current-limited, etc.) version of your domestic AC line for testing. The second one is more important than it sounds, the normal AC line can source some pretty huge fault currents before a breaker trips, and I feel so much better about plugging an untested prototype into an AC source that can start from zero phase for a soft start, and only deliver a really limited amount of power if something goes wrong (vs. 50A+). Both these AC sources can be controlled remotely over GPIB, RS232, etc. and also measure important parameters like power factor and RMS current draw. The 61501, being newer, can also do things like add a DC offset (or even just be a DC source), have settable output impedance and slew rates, or create harmonics and programmed power line disturbance sequences for testing.
There's a rotary encoder knob at the far right for setting parameters, but it got left out because I'm bad at photograpy:
Both AC sources use the same very basic principles to create their arbitrary AC output: the incoming AC line is rectified and stepped up to a DC bus voltage (AC-DC stage), then an inverter uses PWM + filtering to generate an AC waveform from the DC bus (DC-AC stage). The difference is in the details of both these stages, as you'll see on the inside.
There's a lot going on with both of them, so let's look inside the
6408 first:
There's a display board against the front panel which mostly just drives the LED displays there and connects to the buttons and rotary encoder. Behind that, against the front wall of the power section is a control board with all kinds of things going on. The rest of the instrument is split into two halves: the bottom half has two stacked boards, with the DC-AC stage (visible here) stacked on the AC-DC stage. The other half has a pretty large hunk of steel in the form of a transformer, and some filtering which leads straight to the output.
Control BoardThe control board has a bunch of op-amps/comparators and digital logic chips, probably for sensing and protection:
A large chip: the AD1674 is a 12-bit 100ksps ADC. This is probably what reads in any and all sensed values (output voltage, output current, etc.) for the digital stuff.
An FPGA, along with an 8 Mhz crystal and an AD7542 12-bit 2us multiplying DAC: maybe this generates reference frequencies and/or an analog sinewave reference and/or the PWM signals for the inverter board's gate drive?
...and a DSP, the Analog Devices ADSP-2105, with an NEC D71054, which is apparently a programmable timer peripheral: there's also (not visible here) a DAC08E 8-bit 1-Mhz-multiplying DAC just to the left.
There's also an extra board which looks like an expansion or I/O card, but it only has some digital logic on it (most notably, a Hitachi HD64180 Z80-based processor) and doesn't actually connect to anything except the control board.
My best guess is that this is the general control/display/UI processor, and that there wasn't enough space to fit it on the control board at the front, so they had to put it onto a separate board and find a place to squeeze it in. Doesn't sound terribly plausible but it's the best I've got.
PFC BoardThe PFC board (AC-DC) holds a boost converter which steps up the AC line voltage to a DC bus (< 350V judging by capacitor ratings). This board also has the auxiliary supplies which power the digital and analog controls.
* EMI filters for incoming AC line are on the left side.
* Inrush-limiting resistor (the giant gold 100-ohm) and its bypass relay(s) (white boxes) are just to the right of the filter.
* The bridge rectifier is at the left end of the heatsink.
* Boost converter consists of two power transistors (paralleled), a diode on the heatsink, and the 3 inductors behind the heatsink. Control and gate drive comes from the 16-pin ML4812 PFC controller IC below them. DC output goes to the big aluminum caps and common-mode choke above the heatsink, and exits through a cable up to the inverter board.
* Aux power has its own bridge rectifier (bottom right) and a flyback converter using the right-most transistor on the heatsink, and the transformer to the top-right of the heatsink (components clustered around the transformer are most likely snubbers and similar). It looks like the AC line input is connected to the boost converter's input all the time, but the front-panel switch connects/disconnects the aux supply's input. Control comes from the 8-pin UC3844 towards the bottom right. Three outputs are visible, each with its own ouptut diode, set of caps, and linear post-regulator. Two of the outputs share a ground, while one is isolated from the others. Regulation is done by the secondary-side TO-92 feedback amp (looks similar to a TL431) through the left-most optocoupler (white IC).
This particular PFC board was made in '87.
Here's a closer view of the power devices:
The aux supply also has a crowbar circuit, with an SCR on one supply which gets triggered through the other optocoupler by an LM393 comparator on the AC line side of things. Maybe this crowbars the digital supply when input voltage gets too low (as in, switch has been turned off) to prevent brownouts? One product I worked on had a similar problem with a clock distribution chip which would refuse to power on in the correct state unless its 3.3V power pin got below 50 mV or something really low.
Here's a schematic which shows the general workings of this board:
What's up with the boost inductors anyways? Well, for one thing, they have multiple layers of Litz wire for low AC resistance:
But why are they split up into 3 separate pieces? Why do they have those weird jumpers in between, in the form of thick pieces of wire with spade terminals on both ends?
Each jumper footprint seems to be labeled either "110" or "220", which makes me think that this is an input-voltage-selection scheme. Maybe for 110VAC, they connect some of the inductors in parallel for a lower inductance but a higher current-handling ability (since the input current will be higher for a given power). Then with 220VAC, they could connect some of the inductors in series instead, for a higher inductance (which would give a lower current ripple). This way, they can also manufacture the exact same parts for both 110 and 220VAC models. Not sure why they would use these comparatively-hard-to-install jumpers instead of soldering some wire staples down on the board, but maybe they wanted to have the option of reconfiguring a unit between 110/220VAC in the field.
One of the MOVs in the EMI filter blew up, leaving a bunch of soot on the bottom of the inverter board, but I only found this out when taking it apart because it had been working fine.
Inverter BoardThe inverter board (DC-AC) now takes the DC bus from the PFC board, and generates a 60 Hz sinewave.
The DC power comes in at the back edge. You can see the heatsinks for the power transistors, some passives for filtering in the middle, and some control and gate drive circuitry (especially the DIP-4 transistor pairs and gate drive transformers) up in the front.
Here's the power transistors and the secondary-side gate drive circuitry:
There's 2 bipolar power transistors on this side, but those are unrelated to the inverter: the left one is a linear regulator for the fan (possibly adjusting speed?), while the right-hand one has its collector unconnected, using its Vbe as a temperature sensor for the heatsink.
There's 8 MOSFETs, but only 2 gate drive transformers, and adjacent pairs of MOSFETs share gate drive circuitry and common source connections; this heavily suggests that the inverter consists of an H-bridge, with pairs of MOSFETs connected in parallel. Checking the bottom-side traces confirmed this, and also led to this weird-looking schematic of the powertrain:
The two toroids and film cap provide some differential filtering to average the PWM signal before it reaches the isolation transformer. The common-mode choke helps prevent remaining high-frequency components from coupling capacitively to everything in sight, from the massive-area capacitor plate formed by the isolation transformer's windings.
I honestly have no idea what the current transformer is doing in there. It seems to be sensing the sum of the current to both sides of the half-bridge, but also has 1-ohm resistors bypassing it (which only makes a difference at fairly high frequency, I think, where the wire/CT inductance starts giving it a higher impedance and sending more of the current through the 1-ohms). The thing is, since current transformers can only sense the AC portion of the current, the sensed current will end up looking something like this, for a resistive load on the inverter's output (this only shows the 60 Hz portion, as I'm assuming the PWM frequencies are filtered/ignored):
Any light shed on this would be helpful.
The control stuff on the inverter board doesn't actually include any PWM controllers; there's just a couple
ICL7667 dual gate drivers, some generic digital logic to the right, and a bit of analog to the left. The FPGA or DSP on the control board probably generates the actual PWM signals, and the logic here creates some dead time or whatever else may be needed. The analog stuff seems to be for temperature sensing and fan control, judging from its connections to the fan-regulator and temperature-sensor BJTs.
The gate drive circuits on the power-MOSFET side of things are kind of interesting. I was wondering what the smaller MOSFETs were doing, so I traced out the circuit, which was luckily very simple:
The Gx and Sx are the gate and source for each of the two paralleled power MOSFETs.
The 47nF and 22 ohms at the left looks like a snubber, likely to damp the spikes/ringing from the huge leakage inductances that must be present as a combination of the gate driver transformer's imperfect coupling between its windings, and the super-long traces connecting the gate drive transformer (GDT) outputs to the power MOSFETs.
When the voltage from the GDT secondary is positive on the top and negative on the bottom, the bottom MOSFET turns on (connecting G1, G2 to negative) and current flows through the body diode of the top MOSFET (making S1, S2 positive). This makes the power MOSFETs' Vgs negative, so this corresponds to the off-state.
When the GDT secondary voltage is negative on the top and positive on the bottom, the bottom MOSFET is now off (because it has negative Vgs) but current flows through its body diode and into G1, G2 (connecting them to positive) as well as turning on the top MOSFET (connecting S1, S2 to negative). This makes the power MOSFETs' Vgs positive, and corresponds to the on-state.
So why even have the MOSFETs in this circuit at all? In steady-state on or off, this circuit behaves as if those 2 little N-channel MOSFETs weren't even there. I think they're there to help out the transient response of the gate drive - they seem to create fast rising and falling edges on the power MOSFETs' gates. When the voltage from the GDT is changing, the GDT voltage won't actually be connected to the power MOSFET gates until it reaches a positive or negative level high enough to activate the IRFD110s. Additionally, because both charging and discharging happen with a body diode in series, a decreasing positive voltage or increasing negative voltage won't actually change the power MOSFET Vgs! So in the end, having these MOSFETs in the circuit makes it wait until the GDT voltage is pretty reasonably high or low before turning themselves on and dumping charge from the 47nF cap into or out of the power MOSFET gates; this lets the power MOSFETs turn on and off faster, spending less time in their lossy linear "halfway on" region. A picture is definitely called for:
Hope that all makes sense.
Output SectionAnyways, after the inverter generates its nice low-frequency sinewave, that gets passed to a giant transformer in the back-left corner of the instrument:
This is what provides the isolation between the Chroma's AC output and the input AC line (the floating output is one of the important features of an AC source).
The transformer actually has two sets of outputs (4 wires), which go to an output/filter board, from which wires run directly to the AC output connectors. This board seems to do some final EMI filtering, and also some output voltage/current sensing (as it plugs into the control board at one edge).
From the transformer's 2 separate output windings (labeled "VS1-", "VS1+", "VS2-", "VS2+" on this board), and the collection of relays on this board, I'm guessing that the transformer's secondary windings are completely isolated from each other, and can be connected either in series or parallel to extend the output voltage range (parallel for low-voltage high-current, series for high-voltage low-current). I couldn't lift up the board to check the traces on the bottom, but if I had to do something similar with the components shown here, it would look something like this:
By the way, a quick discussion of dates:
The revision date printed on the inverter board shows it was designed in '85, while the PFC board's date sticker shows '87. However, the date sticker on the inverter board is from '98! For one thing, this must've been a good design to be in production of some kind for at least 13 years. It also makes me wonder why the inverter board is so much newer than the rest of the AC source - the original instrument must've been from '87 originally, but then the inverter board failed much later on, and got replaced.
That's all for the 6408; next post, I'll be showing the inside of the newer 61501 and its significant differences.
Home
Help
Search
Login
Register
