ET System EAC 500 AC Source
A Teardown and some kind of Repair Log
A big and heavy box with no controls
This is what has turned up one day at my doorstep. Just a power on light at the front panel, and
a power switch at the back panel. No other controls besides a RS232 labeled D-Sub connector and a DIP switch for
an address setting. That's all she wrote.
Beeing okay so far as this was perfectly what I expected, I visited the manufacturers web pages: This thing appeared to be well unknown to them.
No manual, no spec sheet or anything. From the type number, I guessed this has to be an 500VA model, otherwise, the output ratings are written on the type label.
Failed to operate
Using your favourite internet search engine gave similar results. I found some of the type numbers explained, e.g. option "FA" meaning "no controls, remote only",
and some manuals and block diagrams of their newer models. These manuals include a description of the remote control commands, and there is a remote control
software available for download. I had downloaded and tested this software, but I couldn't establish any communication to the unit. Neither using the software nor
using a simple terminal. I wrote an email to the manufacturer, asking if that available software can be expected to communicate with this unit, and if there's any
kind of manual available, but got no answer at all.
Inside the unit, I found an internal IEEE-488 bus connector, located on the CPU board.
There's also an NEC UPD 7210 IEEE-488 controller and the neccessary drivers on board.
So I made an adaptor cable and tried to communicate via IEEE-488. Result was the same as above: Nada. That thing just didn't do anything.
Now for the fun part
Now, I'm left to my own devices. Either I'd find a way to control that AC source or I've got a real nice brick made of aluminium, iron, copper and dust:
A bit of reverse engineering
Within the card cage, one finds seven boards, four of them related to the "digital" side and three "analog" ones. Top left is the auxiliary power supply, employing its
own small transformer. The heavy and huge transformer at the right hand side provides power for the output stage. Two heat sinks are located in the middle, each one
loaded with 20 MOSFETs. Together they form the output stage, a linear H-brigde amplifier configuration. Above the heat sink, vastly covered in dust, a pair of
SCR/Diode modules form the controllable main rectifier. This configuration provides variable power rails to the H-brigde output, reducing power dissipation at lower output
voltages. There are three SSR (solid state relay) modules stuffed between the large electrolytic caps and the transformer. These are used to
connect the main transformes secondary windings either in parallel or in series. This way, the power supply can provide more output current at lower output voltages.
The board has nothing fancy, just a Z80 based controller system, consisting of the usual
stuff like the main CPU, some EPROM and RAM, the IEEE-488 interface and an address/data bus at the card connector. Quite the
usual way one built such stuff back in the days.
DAC board 1
That's where it starts to get interesting: This board has two identical DAC circuits. At the right hand side, one can see the few
components required to interface with the address/data bus, then there is an isolation barrier: A small DC/DC converter, some filtering
circuitry, optocouplers. These provide power and serial data to the DAC chips.
The DACs require an external voltage reference and some amplifying. I guess the circuitry here is similar
to the reference circuits in the datasheet, but I didn't check that any further.
This board outputs two voltages through the nicely coloured ribbon cable. So I've probably just found two analog setpoints for the waveform
generator circuit. The reference voltage is 10V, so I guess this board provides
two 0...10V outputs. There's nothing else connected to the ribbon cable here.Looking at pictures of similar units having manual controls,
one can see multiturn pots used to set the output parameters like frequency and voltages:
The available settings are output voltage, frequency, and output current on one of these.
DAC board 2
This one has one DAC channel, built the same way as one channel of the DAC board 1. The other channel
isn't populated on the board, instead there's some kind of relay board mounted at the otherwise free space.
The DAC output is routed through that relay board. As it turns out later, this board handles the switching for
the DC output mode and the DAC provides the setpoint for the DC output. Interestingly, the DC output mode provides
only a -82V ... +82V range, while in AC mode the unit can supply up to 260Vrms.
ADC and control board
Another small DC/DC module supplies an 12bit ADC, a parallel to serial converter (just a 16 bit shift register) and
optocouplers take care of the isolation. Additionally, we find a bunch of ULN2803 drivers, these are used to provide a 32 bit wide digital output
from the main CPU to control the unit.
The eight small reed relais form a differential input signal multiplexer used to route various measurements from the measurement board to the ADC. This
is the way the CPU reads these values back. Manually controlled units provide the user with an LED display and some keys to select the measured value
like frequency, output voltage or current. Only five of eight possible signals are used here.
One of the ULN2803 drives the eight relays on board. The next one appears to be completely unused, as its outputs measured open circuit while I investigated
the signal routing. The two left over ULN2803 drive relays and optocouplers located on one of the next boards to be visited.
So far, one can conclude this unit is able to be controlled by using switches (that replace the ULN2803) and control voltages. No fancy communications and CPU
required to operate the AC source. One has just to figure out what these control outputs do within the generator circuitry, that's a typical case of making some
educated guesses and then perform a classic trial-and-error procedure.
Anyway, these were the boards that are connected to the CPU board. The other boards are mounted using the very same backplane PCB, but there's no electrical
connection between the CPU related boards and the AC source boards on that backplane.
This is the first board of the actual generator / AC source section within the unit. There's some interesting old style analog ICs on this board:
two RMS converters
an analog multiplier
another analog multiplier
an isolation amplifier
a shitload of trimmer pots
some operational amplifiers
and even your classic NE555
The isolation amplifier isolates the current sense signal - there's a shunt resistor in one of the amplifier outputs. The floating power supply required
for the isolated circuitry isn't on that board, it is part of the common power supply board. The ribbon cable routes measurement signals to the ADC board.
The mentioned RMS converters are used to provide output voltage and current RMS measurement, quite obviously. Now switching to guess mode: One of the
multipliers provides the output power measurement while the other might calculate output VA (apparent power), or do something completely different. Anyway,
while checking the signals available at the measurement board, I didn't find either of them (neither output W nor output VA). The LM399 and the NE555/NE556 timers
apparently are part of the protection circuitry, as the unit shuts down for a short period of time if some output overload occurs. The relays aren't driven
from the ribbon cable but from the backplane PCB. I have no idea what's their purpose.
Now we find a ratsnest of bodged wires, one of the relays is obviously replaced by more relays located on that little DC mode board mentioned above.
The internal oscillators output signal is routed through that DC mode control board, switching from the sine wave generated here to DC output mode, controlled by
the second DAC board. Another of these relays is used to switch the amplifier input to external oscillator mode, with the input signal located at one of the unused
pin headers, which I found useful. One can use any kind of external (supposedly arbitrary) waveform generator to supply the input signal to the power amp. Quite
useful for testing power supplies for voltage dip hold-up capabilities.
Anyway, there are two interesting old style analog ICs found here:
The AD639 trigonometric function converter
a voltage-to-frequency converter
Looking up the AD639 datasheet, one can find some hints on how this oscillator board may work. I have no idea what the SN7473 Flip-Flops surrounded with tantalum caps and
2N2222 transistors are used for. And there are more relays than I could find working control lines for. The 10pin ribbon cable with the connector brings the control voltages
from the first DAC board here. There's a third control input provided on the 10pin header, I couldn't figure out what it does. At the right hand side, the 26pin ribbon
is connected to the ULN2803 drivers on the DAC board. The control signals are routed through the yellow/white optocouplers, the relay coils beeing driven by the optocouplers.
This provides additional isolation (more on the unwanted interference than the isolation voltage side here). What I found out while testing the control lines, is that this
oscillator provides three preset frequencies (50Hz, 60Hz and 400Hz) and the variable frequency mode through the DAC provided control voltage. The AD639 provides a spare
multiplier that probably is used for amplitude control.
This is the amplifier board, driving a BTL (brigde tied load) power stage consisting of each 10 paralleled MOSFETs. As one can see by the small labels attached to them,
the MOSFETs were selected and matched. There's a small signal transistor with each MOSFET, presumably used for better current sharing and / or current limiting.
Located on the PCB, the TO-39 cans are discrete bipolar transistors, 2N3439 and 2N5416. Mounted on that aluminium bar are mode more transistors, MJE5850. Two ribbon cables
route the drive signals to the MOSFETs on the large heat sink.
A TCA 785 phase control modulator provides the firing signals for a pair of big-ass SCRs to provide a variable supply voltage for
the power stage. At lower output levels, the supply voltage is lowered to reduce the MOSFETs power dissipation. A large mains transformer provides a substantial amount of power.
Another smaller transformer provides a bunch of secondaries to an auxiliary power supply board. Some ribbon cables route a variety of power rails to the backplane PCB.
And now for something completely different
From what I have discovered, it shouldn't be a big deal to make the AC source controlable by some simple means like pots and switches, just like the style
the orignal models with front panel controls were made.
A block diagram
First, draw some block diagrams. This is an overall view of the modules located in the box:
This is a diagram of what I believe how the modules work together:
So all the control from the CPU module goes through either the DAC modules or the latch and driver on the ADC module.
The signals are rather straightforward, first analog control:
An analog voltage to control the DC output voltage
An analog voltage to control the frequency
An analog voltage to control the AC output amplitude
Then a number of static digital signals to control some operating modes:
Preset frequency or variable frequency
DC mode on/off and output polarity
External Oscillator mode
Output stage standby / shutdown
Figure out how to control it
Next, I did replace the DAC chips on the DAC boards with potentiometers:
and the ULN2803 drivers with some DIP switches:
Stuff everything back into its place:
Now, it's a matter of making educated guesses and systematic testing to figure out
what each control does. One may take some notes while doing so, as this helps with remembering.
Having figured out most of the controls, and found the external oscillator input, let's do
Internal oscillator set to 230V / 50Hz
External square wave at 100Hz
Same square wave, showing an about 50us rise time
DC output mode with load
Making a control box
Now knowing which switches to control the AC source, I've put the potentiometers and switched into
an external box, connected through the now useless RS232 port:
Still some labels required for the box.
Do some measurements
One potential use would be to measure the efficiency of power supplies.
As an example, I've set up this measurement for a rather lousy 12V brick:
The power meters are showing 43W input power and 36.1W output power, this calculates to an efficiency of 84%.
Teardown pictures / Gallery
Goto here for some teardown pictures.
What the heck is an AC source?
What is it: It's an AC source - that's the technical term for it.
What does it in general: It outputs an AC voltage of selectable / variable frequency and amplitude. In particular, these units output typical voltages of 100V to 250V (or 400V for a three-phase model), at typical frequencies from 50Hz to 400Hz, with some respectable output power (this unit is specified for 500VA, one can get them in the range from 200VA to several kVA and three-phase models).
So if you see the output specs, what does it in particular: It simulates an AC power supply (like your household mains) for testing purposes (e.g. of power supplies at different line voltages and frequencies). More advanced testing includes short voltage interruptions, sags, and harmonic content, you'd require an arbitrary waveform generator to do so with this unit, modern ones have that built-in.
Just imagine a variac that isn't adjustable for output voltage only, but also for frequency (and more recent units also waveform) and doesn't depend on mains stability.
Some components datasheets