Survival course

Yesterday was a day full of sparks – electrostatic discharge sparks. Tests were done by shooting sparks to to different targets including cables, casing and open connectors while the drive was operating. Also various ESD types were tested: contact discharge, air discharge, varying voltages and different polarities. All ESD tests were passed without a single hiccup.

ESD test table and ESD gun

However, ESD is only one mode of high voltage harassment drive must cope with. Another test type is EFT (electrical fast transient) which is induced to cabling through capacitive clamp. Several kilovolts with nanosecond scale rise times were injected as 75 spike bursts into the clamp at frequency of 5 kHz and repeated 10 times. This simulates interference spikes emitted by nearby sparking or EMI. No problems recorded.

Encoder cable going through capacitive EFT test clamp

The last test was a surge test. A high energy and relatively long pulse were applied to supply voltage inputs. This is typically the most difficult harassment to filter out in electronics.

Surge & EFT test equipment – above is the 24V PSU and the device below is the EFT & Surge injector.

Difficulty of surge may be imagined by the funny fact that cheap lab power supplies tend to die during the test even when it’s not the target. So heavy duty PSU had to be used as supply to avoid it breaking.

In this test we found a problem. The voltage regulator of Argon died when shot with 400 V and ~40 A pulse in negative polarity to 24V input. However, drive survived a 500 V positive polarity surge. 500 V is the required level to pass the test. Failing one tests means that some circuit/layout changes has to be done to address the problem.

EMC test have begun!

A day in a EMC lab yielded over 40 radiated emission measurements and lots of useful data about how to install the device to minimize interference.

GTEM type EMC chamber

GTEM type EMC chamber

Inside GEMT type EMC chamber

Inside the chamber

Few radiated emission test results shown below. The green line represents the limit to meet IEC 61000-6-4 limits in industrial environments, so any spike above it would mean that device is not compliant.

Chamber & measurement equipment bacground noise

Chamber & measurement equipment background noise. The small spikes in GSM frequencies can be seen beyond 900 MHz. Notice that graph has different scale than the ones below.

These tests were carried out while drive was executing 10 Hz high current square wave torque command which was found to produce the worst case scenario.

Driving square wave torque command without any added filters on cabling

Driving square wave torque command without any added filters on cabling

By experimenting we found out that a ferrite bead core on 3-phase output leads (as common mode filter) before motor cable makes a quite significant difference. The graph below shows a success in emission test. This test was repeated several times in different ways to eliminate the possibility of error.

Driving square wave torque command with a ferrite bead filter on phase leads

Same as above but with a ferrite bead filter on phase leads. A spike visible at 360 MHz is probably a third harmonic from the 120 MHz CPU.

In addition to emission tests, also radiation immunity test was carried out. No sign of disturbance was noticed on drive operation or communication to PC, so immunity test was passed as well.

Next up: ESD & EFT tests. But first I’m going to visit Germany and SPS IPC DRIVES fair in Nuremberg. See you there!

Preparing for EMC testing

Tomorrow will be a big day for Argon; it will be tested for EMC (electromagnetic compatibility) standards in a test laboratory. The tests consists several measurements: radiated emissions, conducted emissions, radiated interference survival, electrostatic discharge and surge current tests.

Argon will be tested against the most stringent standards that the CE regulations specify ensuring (if passed) that it may be installed in virtually every environment without violating the regulations. For example different cable length classes exists. Some classes require testing up to 3 meter cables but I chose to use the class that requires passing all tests with 30 meter cables in addition to 1 meter cables. In some cases 1 m cables may give worse results so testing with a such short cables is also a requirement.

Milled enclosure sheet metal

Milled enclosure sheet metal

The device under test has a self rolled enclosure milled with a self made JV2 CNC router. The sole purpose of this prototype is to check fitting and provide EMI shielding during the tests. The final enclosure will have more professional finish.

Argon testing with work-in-progress enclosure

Argon testing with work-in-progress enclosure

Short circuits

Argon has pretty sophisticated circuit protection functionality. The device has sub-microsecond response short circuit sensing circuit that can detect phase-to-phase and phase-to-ground shorts. To counter harsh environment EMI and prevent false fault triggering, there is variable length digital delay filter for short circuit protection  – and this parameter is also user adjustable.

SC scoping

Short circuit scoping. Yellow trace is IGBT phase current 67 A/div and blue is power stage shutdown signal from DSP.

The oscillograph above displays a phase-to-ground short circuit when shorted by a piece of wire right at output terminals. Drive was driving an AC motor at supply voltage of > 200 VDC while wire is shorted to ground. The yellow output current graph shows current of 170 Amperes and pulse length of about 2 µs before shutdown was issued. I believe the blue digital trace is fluctuating because board’s ground plane is struggling to carry the enormous current.

The test has been repeated dozens of times with varying protection delay times and supply voltages, and no harm to hardware was ever induced. I’m satisfied.