The effect of photoconductive shunting on HIT modules

[Mark Campanelli]

To whom it may concern,

  I have been updating PVfit, and I have updated the getting started demo for IEC 61853-1 matrix fits to compare the "simple" SDM with the "less simple" SDM that adds a photoconductive shunt as the auxiliary equation for the shunt conductance (the inverse of the shunt resistance).

  For context, some of the nine modules tested, PVfit's simple fitting actually suggests a zero shunt conductance at STC (equiv., an infinite shunt resistance at STC). Interestingly, for these (few) modules, adding a photoconductive shunt doesn't actually change the fitting to zero of this parameter. (I had expected otherwise.) However, it does improve some of the fits that already have a non-zero shunt conductance for the simple model. Importantly, I observe that it significantly  improves the low-light fitting of the one HIT module in the group:

  What is also interesting here, is that the "improved" PVfit model does not add any new fitting parameters to describe the photoconductive shunt, which is a simpler approach than PVsyst (which IIRC adds at least two fit parameters here). I also note that fitting to only the spec sheet parameters appears to be much more difficult when there is a significant photoconductive shunt, which makes sense because there is no light-level dependence information being used in PVfit's algorithm for this.

  So, if anyone has any IEC 61853-1 datasets for modules with high photoconductive shunt, then I'd be happy to compare notes, especially with respect to any PVsyst results.

Thanks,,

Mark Campanelli

[Anton Driesse]

Hi Mark,

Your photoconductive shunt without additional parameters sounds just like the CEC/De Soto model (conductance proportional to irradiance).  

Is it the same?

Anton

On 2026-06-04 15:57, Mark Campanelli wrote:

To whom it may concern,

  I have been updating PVfit, and I have updated the getting started demo for IEC 61853-1 matrix fits to compare the "simple" SDM with the "less simple" SDM that adds a photoconductive shunt as the auxiliary equation for the shunt conductance (the inverse of the shunt resistance).

  For context, some of the nine modules tested, PVfit's simple fitting actually suggests a zero shunt conductance at STC (equiv., an infinite shunt resistance at STC). Interestingly, for these (few) modules, adding a photoconductive shunt doesn't actually change the fitting to zero of this parameter. (I had expected otherwise.) However, it does improve some of the fits that already have a non-zero shunt conductance for the simple model. Importantly, I observe that it significantly  improves the low-light fitting of the one HIT module in the group:

  What is also interesting here, is that the "improved" PVfit model does not add any new fitting parameters to describe the photoconductive shunt, which is a simpler approach than PVsyst (which IIRC adds at least two fit parameters here). I also note that fitting to only the spec sheet parameters appears to be much more difficult when there is a significant photoconductive shunt, which makes sense because there is no light-level dependence information being used in PVfit's algorithm for this.

  So, if anyone has any IEC 61853-1 datasets for modules with high photoconductive shunt, then I'd be happy to compare notes, especially with respect to any PVsyst results.

Thanks,,

Mark Campanelli

--
You received this message because you are subscribed to the Google Groups "pvlib-python" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pvlib-python...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/pvlib-python/bc8ce488-1176-4c08-acc5-6b4d25858115n%40googlegroups.com.

-- 
PV Performance Labs
Emmy-Noether-Str. 2
79110 Freiburg
Germany

+49-761-8973-5603 (Office)
+49-174-532-7677 (Mobile)

www.pvperformancelabs.com

[Mark Campanelli]

Anton:

  Indeed, nothing fancy or really novel here. G is proportional to PVfit's implementation of effective irradiance, F=Isc/Isc0, i.e., G = F * G0.

-Mark