PSU Project: Solar Panel Simulator

[HackedFridgeMagnet]

Just had this idea yesterday so little detail.
Still working on a feature set.

something like this.

https://www.bkprecision.com/products/power-supplies/PVS10005-1000v-5a-5kw-programmable-dc-power-supply.html

but lower power and voltage and especially cost.
Want to be able to put in I/V curves, angles, ambient temps, clouds, partial shading(eventually). I guess it will interpolate curves to do these.
Maybe one day work out an appropriate step response to a change in load too.


Was thinking of a switching pre-tracking regulator and then a linear reg to match the curve to the current load to get the operating point.
I guess I will just measure the V and I currently output and use that to match against the current curve.
Load the curves via usb.

My biggest issue that I can see is the multiple control loops fighting each other.
1. Pre tracking
2. Linear output.
3. Mppt charge controller as a load.

but if I include some physical resistance as part of the output curve, which any PV system will have, it may be easy to stablise.

Any ideas, comments are welcome.

[fourtytwo42]

Just had this idea yesterday so little detail.
Still working on a feature set. something like this.
https://www.bkprecision.com/products/power-supplies/PVS10005-1000v-5a-5kw-programmable-dc-power-supply.html
but lower power and voltage and especially cost.
Want to be able to put in I/V curves, angles, ambient temps, clouds, partial shading(eventually). I guess it will interpolate curves to do these.
Maybe one day work out an appropriate step response to a change in load too.
Was thinking of a switching pre-tracking regulator and then a linear reg to match the curve to the current load to get the operating point.
I guess I will just measure the V and I currently output and use that to match against the current curve.
Load the curves via usb.
My biggest issue that I can see is the multiple control loops fighting each other.
1. Pre tracking
2. Linear output.
3. Mppt charge controller as a load.
but if I include some physical resistance as part of the output curve, which any PV system will have, it may be easy to stablise.
Any ideas, comments are welcome.

I often wish I had something like that for bench testing BUT I try to get stuff off the bench and outdoors doing something useful as fast as possible.  When I want to bench run something I cobble together a collection of transformers and electric fires to represent the PV source and more electric fires to represent the load.

Anything that requires precise control of parameters I do in simulation (LTspice it's free) as that's a hell of a lot cheaper than building test gear, after years of use I can get a pretty close to reality model quite easily and see anywhere inside it I want. The only downside sometimes is run time, if you want to run several seconds or more of realtime on something switching at 125Khz it can take several minutes but there are various ways of tuning a simulation for speed also.

[ken246810]

I just read this and thought I would comment.
I got hold of a PVS10005 a few months ago with no output and it was cheap. it took me a few days to fix it but it was mostly waiting for parts, it was an easy fix.

It was harder to find and order the BK custom mains lead but I did get one after a good bit of hunting for it.

Lovely PSU

[NiHaoMike]

Connect a constant current load in series with the output of a common switching power supply module. Not the most accurate but close enough for some basic testing of MPPT algorithms.

[XFDDesign]

I home-rolled something similar to what you're after, but it's only a 2.2kW setup. Cheap off-the-shelf supplies are stacked up (6 x 12), each with a 40A output. Then a dual-loop control supply (current first, then voltage for my setup).

For the loops, do the faster loop first, then roll it into the slower loop.

[f4eru]

I wonder if putting a panel without light in paralell with a lab PSU in current mode will give you a good sim of a panel with sun

[trobbins]

Do you have a range of PV inverters you need to service, or some specific need, or is this a generalised thought bubble?

It's strange that I couldn't identify any statement in that PVS power supply spec or manual that clarified if the DC output was floating, as PV inverters can have insulation resistance interlocks that would need to be defeated if there the supply wasn't floating with sufficient resistance.

You can make a simple simulator with a CV/CC supply, and adding a fixed series resistor and parallel resistor (after the CC sensor) to achieve some basic slope response of a PV panel/array, and then use something like a Picaxe to overlay some control over the raw supply CV/CC setting to achieve a close match of an actual PV curve.  I did that years ago with a phase reg supply providing 600V 20A.

[Berni]

Yep as trobbins said. A lab PSU can be used to approximate a solar panel.

The voltage set on the lab PSU produces the open circuit voltage and the set current limit sets the short circuit current of a panel. This makes a sharp rectangle around the solar panels specs. To make it a more smooth curve you can then add resistor in series to make the output slowly droop as current goes up. If you want the transition into current limit to be smoother also a parallel resistor can be added across the PSU.

Tho the resistors you use might need to be fairly big to cope with the power dissipation.

[f4eru]

and what about using a lab PSU in current mode, and a non-illumiated panel in paralell, to have a very simple to built but extremely close model ?

https://wiki.analog.com/university/courses/eps/photovoltaic

[HackedFridgeMagnet]

Sorry I haven't answered for such a long time. Other stuff came up.
the Project hasn't really gotten past a voltage controlled Current Supply. Based on LT3763. That works fine for what would be the inner loop. I put a STM32 on the same board but haven't mounted it or developed any firmware for  it.

So yeah the current mode PSU can do some things but I wanted it to be as close as possible. My LT3763 board is any much more than a voltage controlled Current Supply anyway. I was thinking of adding the series resistance too as it is a natural fit to when the panel is at the current end of its operation (curve).

Want to be able to put in I/V curves, angles, ambient temps, clouds, partial shading(eventually). I guess it will interpolate curves to do these.
Maybe one day work out an appropriate step response to a change in load too.

this project is still stalled as I wont have time for quite a while.

[trobbins]

You really need a CV/CC supply that goes to 0V output (Isc conditions) for starters, so UVLO and smps constraints that don't allow full swing to 0 current are a concern (although a parallel load to simulate Voc curve may suffice).  But I'd suggest you need a separate outer loop control that can modify the raw supplies CC/CV regulating levels - such as a PicAxe controller.

[HackedFridgeMagnet]

You really need a CV/CC supply that goes to 0V output (Isc conditions)

Yeah fair point, thanks for that.

[Berni]

Yeah a LED driver is also a good way to do it since its a supply with both CV and CC modes.

I wouldn't be too worried about it going down to 0V since the load being put across the solar panel will likely never get it there. But low voltages are not that big of a problem for buck switchers in the first place, if the controller is not stubborn enough to insist on a fixed frequency it can just stretch the off period as much as it needs for the voltage to drop lower. For a lot of buck switcher chips the low voltage limit is due to the feedback resistors hitting 1:1 ratio (no divider) once you want a output voltage that is equal to the internal feedback reference voltage. If you add extra circuitry on the feedback pin to boost up the feedback voltage then it will go lower no problem.

You can also implement constant current mode on most normal switching regulators that don't feature any CC by again adding some circuitry to the feedback pin to hijack its feedback loop.

[station240]

I'm building a 0-100V 0-5A switch mode supply, based on SG3526 and some opamps.
It's just the controller PCB, and looks easy to feed in external DAC voltages.

@HackedFridgeMagnet Your profile confuses me, as the flag says NZ, but your location says South Island (which implies NZ), so which is it ?
I've got 10 of these PCBs, so some of them will need a new home.

[HackedFridgeMagnet]

@HackedFridgeMagnet Your profile confuses me, as the flag says NZ, but your location says South Island (which implies NZ), so which is it ?
I've got 10 of these PCBs, so some of them will need a new home.

lol flag says Australia does it not?

Australia has a North and South island too!