OwnVerter grid following

General
1

Hello,

Here is a short update about my experiments with OwnVerter used as a three phase grid connected current controlled converter (i.e. grid following).

Just to let you know that, after much hardware and software debugging (*), the converter is finally injecting power back to the actual campus grid! (through a 15V transformer). Power injection tested at about 30 – 45 W (1A current injection amplitude).

The hardware is OwnVerter board tied with a crude inductive filter (three 630µH WE-SI inductances which are supposedly suited for up to 100 kHz switching frequency… OwnVerter is only at twice this value). DC source is a 60 V DC power supply.

The converter follows the Id/q current setpoint with the expected transient convergence (closed loop time constant set to 5ms or 20ms).

The most striking control defect is the presence of a quite large harmonic current perturbation (probably harmonic 5, perhaps 7 as well). The current control design is not equipped to reject harmonic disturbances and I suspect that the grid voltage has a few percents of odd harmonics, enough a quantity to explain the harmonic current of about 0.2 A.

Photo of OwnVerter grid connected
Photo of OwnVerter grid connected1920×1735 398 KB

Scope capture (1A setpoint. Yellow: Ph1 grid voltage, Pink: Ph1 Current, Cyan: Ph2 grid voltage)

Scope capture of OwnVerter grid connected with 1A current control setpoint
Scope capture of OwnVerter grid connected with 1A current control setpoint1024×600 88.2 KB

Same situation seen from the board. The current is quite noisy. Also grid voltage (phase to DC-) has a pretty awkward shape, but the PLL is robust/slow enough to handle this as a harmonic perturbation.

Embedded board scope capture
Embedded board scope capture2000×2500 305 KB

Thanks to OwnTech team for all the useful discussions in the last weeks!

(*) bug examples: flipped lower and upper bounds for PI control ⇒ 0 V output :exploding_head:, flipped phase order, or wrong PI control tuning formula…

2

Great news ! :partying_face:
With a bit of fine tuning I guess we’ll manage to reject these harmonics a bit better !

Have you tried lowering the switching frequency ? In your current case, there is no special need to run at 200kHz. (has there is no output filter like for a DCDC usecase)

3

For the reference: the code I’m using is now available in the ownverter-grid-following branch of my fork of the Core repo :technologist::

In particular, the entire code is in main.cpp

1 Like
4

Yes for the harmonics I have ideas, since there is a very good visual agreement between the grid voltages I’ve analyzed in details and the harmonics current which is flowing. See the scope capture when the Id current set point is set to 0 → only the harmonic current is visible. More precisely, the harmonic current is the opposite (× some conductance factor) of the harmonics voltage (red curve in 1st and 2nd panes), which is coherent since grid voltage acts as a “back emf”.

Grid_harmonics_analysis
Grid_harmonics_analysis1800×1800 416 KB

2025-10-03_GridFollow-steady-0A
2025-10-03_GridFollow-steady-0A1024×600 95 KB

5

This I haven’t tested yet.

  • I suppose that it should have no effects on the grid harmonic perturbation, since it’s independent.
  • On the positive side, I do hope it will reduce the amount of noise, especially in the current measurement. Still, if the noise happened to be more of a systematic error due to current ripple, then reducing the switching frequency (and thus increasing current ripple) may be detrimental!
6

@pierre-haessig

Can you share the python code you used to draw the harmonics?

This way we can also study the problem.

7

Done!