Until then I used and ATX power supply with an ATX breakout board. I provides all voltages needed for electronics: 3.3 V, 5 V, and 12 V. But you can't set the current. It only comes with a 1.25 A polyfuse.

So I decided it was time to buy my first bench power supply. But is had to fulfil the following criteria:

• set voltage and current
• wide voltage range (0-50 V)
• silent
• have an earth connection
• not expensive
• not noname
• provide a communication interface (most important)

I found and bought a used power supply on eBay for ~ 80 €. And now I own an Elektro-Automatik EA-PS 2084-03 B power supply. And this fulfilled are the requirements:

• set voltage and current: in 0.1 V and 0.01 A steps
• wide voltage range (0-50 V): 0-84 V, 0-3 A (not a lot, but enough for electronics), limited to 100 W
• silent: not fan, full passive cooling
• have an earth connection: on the front panel
• not expensive: ~ 80 €
• not noname: Elektro-Automatik (never heard of it, but it's German and they have a decent website)
• provide a communication interface: USB with proprietary documented protocol

Here some more documents about this power supply:

• datasheet (archive)
• manual (archive)
• programming manual (archive)
• driver (USB ACM for Windows)
• easyPS2000 software (Windows, demo version)

When powering the device on, I noticed that the voltage on the display did not change while I was turning on the knob. But the Digital Multi-Meter (DMM) actually shows that the voltage actually does change. Just the reading on the screen is inaccurate.

It was time to measure how inaccurate this power supply is. But doing that manually just takes to long. So I decided to implement the protocol to control the the power supply, and I will measure the set output using a DMM connected to the computer.

To control the power supply I implemented the protocol described in the programming manual. The source code is available on git.

The control programming will increment the voltage from 0 to 84 V in 0.1 V steps, at 1.0 A. It will set the voltage and current which are set, actual (measured by the power supply), and measured (measured by the DMM)

To measure the output of the power supply I used two UNI-T UT61E. These are good multimeters for electronics which you can get quite cheap, with 22000 counts, and a connection to the PC. More functions and its accuracy are available in the manual (archive).

The DMM comes with an RS232 UT-D02 cable. To connect to the PC you need a RS232 to USB converter (are PCs with COM ports still manufactured?).

The cheapest RS232 to USB converter one is based on the CH341 chip. Sadly the 7O1 mode used by the multimeter isn't supported by the linux driver. I also tried the patch, and after toggling DTR I get wrong data out.

I also has an old ARL3116 based RS232 to USB converter, but there too the mode didn't seem to be supported.

Finally I found a FT232-based RS232 to USB converter. This is an expensive cable (but good quality), from an evil company, and it worked.

But instead of using it, I decided to connect the cable to a CP2102-based UART to USB converter. Then you have to keep two things in mind:

• RS232 signals are between -12 to +12 V, while UART uses 0-5 V. This was solved by simply using the 5 V pin from the converter to power the cable.
• RS232 signals are inverted compared to UART. This was solved by inverting the signal using an NPN transistor and two resistors.

More details about the cable and modification is described in this picture:

For the second multimeter I used a UT-D04 USB cable. This time the data doesn't come over a serial port, but rather a HID device.

To read the data from both multimeter I used sirgork. It supports the UNI-T UT61E and both cables.

Once sigrok-cli installed you can record the data using the following command:

sigrok-cli --driver uni-t-ut61e-ser:conn=/dev/ttyUSB0 --samples 1 -O analog

or

sigrok-cli --driver uni-t-ut61e:conn=1a86.e008 --samples 1 -O analog

depending on the cable.

I've run 5 experiments:

• go from 0 V to 84 V in 0.1 V increments, without any load
• go from 0 V to 84 V at 1.0 A in 0.1 V increments, with a 678 Ω load
• go from 0 V to 11 V at 1.0 A in 0.1 V increments, with a 10.2 Ω load
• go from 0 A to 3 A in 0.1 A increments, with a short
• go from 0 A to 1 A at 10 V in 0.01 A increments, with a 10.2 Ω load

After changing a value I've waited 3 seconds for the measurements to stabilized.

The measurements and accuracy calculations are available in this spreadsheet.

Here are the resulting graphs:

• go from 0 V to 84 V in 0.1 V increments, without any load

• go from 0 V to 84 V at 1.0 A in 0.1 V increments, with a 678 Ω load

• go from 0 V to 11 V at 1.0 A in 0.1 V increments, with a 10.2 Ω load

• go from 0 A to 3 A in 0.1 A increments, with a short

• go from 0 A to 1 A at 10 V in 0.01 A increments, with a 10.2 Ω load

As you can see the measured values are most of the time higher than the set values, but within the 0.2 % accuracy (to 84V or 3A). But the actual values displayed by the power supply is way below what is set, and outside of the accuracy, particularly on the low voltages.

Conclusion: don't trust the displayed voltage (it's too low), but you can be confident the output is right (except for the very low voltages and currents).