Solar Freaks

AIM:
To see how much power is gained by cooling the panels from room temperature to Zero Degrees Celsius.

Notes: Because all panels behave differently to the effects of varying temperature, ive chosen to test John Jefferys' new solar modules as there were many teams using these panels this years, and im guessing more teams will be using them next year to.
So far testing has shown these panels to be really great, they have good power for their size and high fill factors.

METHOD:
1) Test Panel Power at Room Temperature (~25 degrees centigrade) .
2) Cool panel in freezer, until temp drops below 0 degrees centigrade.
3) Re-test panels at or below 0 degrees.

RESULTS:

Picture Above shows temperature reading before power test was taken at room temperature.


NanoMETER Room temperature Test results. Please note; the Fill Factor shown is about 1-2% below normal as i used a very long and thin panel leads (3 meters). These panels are very good and have fill factors towards 0.76.


Above we can see the panel is below Zero degrees.


As we can see, there is a significant advantage by cooling your panels. Nearly 6% more power per 25 degrees Celsius.

At 0°C: 8.55 watts
At 25°C: 8.12 watts.

Note Also the drop in Isc and Impp after cooling...
Going to test the temperature compensation formula to see if its on par with my results..

Excellent post Tony.
These are exactly the kind of results that I've been looking forward to. Particularly to help explain and prove to students that it is definitely worth cooling down your panel before a race.
While it has been mentioned several times in other posts that an auto tracking electronics unit will be very convenient here, it can still very readily be done with a set of electronics that require manual setting.
Once both John's and your new units come out, everyone will then be on the same playing field anyway.
It would also be good to get this translated into an actual time improvement of a car on the actual track, but what is shown above it is certainly the next best thing. Things can always be simulated from there. Having definitive power data is better in some cases anyway.

It will obviously vary slightly from panel to panel, but as a general rule you'd expect the voltage to change by about 0.5% per degree so the voltage changes are a little, but not too far, off for the 25 degree variation. The change in power above is a little smaller than what one would expect but who knows.

miseli

The formula for standardising solar panel temperature to 25° is:

PStandard = Pmeasured + Pmeasured x 0.004(T - 25)

After standardising our -1.6° power reading, we would have:

Code: Select all

P = 8.55 + 8.55 x 0.004(-1.6° - 25°)
  = 8.55 -  0.91
  = 7.64W

Which actually means the formula is overcompensating.

Therefore, by cooling your panels before scruiteneering, you could actually have a standardised power reading less than what the panel output actually is. We are actually gaining half a watt.

But, if we heat the panel and then test it, after standardising our power would exceed the panels actual power.

Yeah, I noticed that too. The potential for an advantage is certainly there. I will be making a mention of this to John, Ian and Stan.
Maybe the formula needs to be reviewed? Some more extensive testing on a variety of panels and cells might need to be done.

To make things fair, a rule might simply be added to the 2010 regulations that limit the range of temperatures at which a panel will be measured (say plus or minus 5 degrees of the ambient temperature).

miseli

I'll certainly be doing further testing with various panels.
Maybe that formula is based on the solarex panels everyone used to use.

Tony,
Marc told me about your experiments and I promised to comment.
The formula came from 'Solar Cells' by Martin Green, page 91, 5.3 Effect of temperature.
This states: 'The short circuit current of solar cells is not strongly temperature-dependant. It tends to increase slightly with increasing temperature.--------The other cell parameters, the open circuit voltage and the fill factor, both decrease.'
Then on page 92 it states: 'The dominant variation is that of Voc. This causes the power output and efficiency to decrease with increasing temperature. For silicon, the power output decreases by 0.4 to 0.5% per degree C.'
I don't recall how the lower figure of 0.4% was selected but maybe it needs reviewing.
Despite that, the temperature correction was originally introduced because it had become common practise for some teams to intentionally overheat their panels before scrutineering to obtain a lower power rating and panel weight, then cool their panels for racing, giving them a considerable power to weight advantage. The fact that an advantage can be obtained by doing what would normally be counter productive, that is, cooling your panel, definitely falls outside the spirit and the intent of the event and would be severely frowned upon. Now that the effect is to become common knowledge the rules will most likely be modified to remove any possible advantage being taken.
That said, you are to be applauded for demonstrating so effectively the effect of temperature on silicon solar panels.

Thanks for your input John, and welcome to the forum.

The purpose of this test was to see how much power could be gained by cooling panels before races.
It seems, that a cold panel will certainly be of huge benefit, and for a 12 watt panel, you could be gaining nearly 1 watt.

Abusing any possible loophole is definitely outside the spirit of the event. The reason why this was mentioned publicly is so the findings can be further investigated, and something can be done.

The formula might need reviewing. It was derived in the early 80's and cell technology has certainly changed.

I suspect that you are correct about the age of the formula and agree that cell technology has changed dramatically since then.

But it has worked, no-one but no-one comes to scrutineering with a hot panel any more.

As I am in the process of updating the rules for next year it would be much appreciated if you can either find the latest thinking on this matter or, failing that, continue you experiments and come up with a more suitable correction factor.

Those cells are laser etched and anti reflective coated poly crystaline, but I don't think that has any effect on the matter. I don't know if the quality of the silicon (fill factor?) has any affect.

I am forever learning about this fascinating subject. I studied engineering back in the 60s when most of what we take for granted now was not even the stuff of dreams. When we started solar car racing in the early 90s we had little idea what we were doing, and no-one would tell us, so we had to learn.

john jeffery wrote:As I am in the process of updating the rules for next year it would be much appreciated if you can either find the latest thinking on this matter or, failing that, continue you experiments and come up with a more suitable correction factor.

Im doing further research as we speak. So far, different panels seem to behave slightly different to cooling/heating effects, however power gains seem to have a linear trend. Im testing from 0°-> 25°, and 25°->50°. I might calculate the average power gained per 25° for each panel, average it, divide it by 25 and revise the formula on this basis.

Ive got solarex, scorpio, DSE, some 'no-name' commercial monocrystalline 10W panel, and your modules at hand for testing. What do you think?

I know it sounds like a lot of work but you seem to enjoy so it would be really informative to have data from lots of different panels to see if there is a trend. You need to test at a different temperatures, not just 25 and zero, to see if it is a linear effect or not, and you need to repeat the original test to make sure the result wasn't caused by some other influence. Could the panel have frosted up, for instance? Does the epoxy coating have an influence? Is your test procedure repeatable? You can simply allow the panels to warm up on the light box as you take multiple power readings, provided the warming is uniform across the panel.

No matter what you do, Ian Gardner will no doubt repeat your tests, and Stan has the final say taking lots of other things into consideration.

If different panels behave differently then maybe we can simply limit how far from ambient a panel can be for testing, and perhaps not correct back to 25 degrees but whatever ambient was at the start of scrutineering.

The effect of temperature on power and Vmp during a race is minimal since the panel warms very little during a sub 20 second race.

John there was some condensation on the panel, but no frost. Im not to sure how the epoxy encapsulation effects the panel output. Further testing has shown it may not be linear, especially in the cold temp range.

Note however, the standardising formula has a linear characteristic.

Here are the results from the 50° test.


Power test:


Standardizing these readings, we have:

Code: Select all

P = 7.37 + 7.37 x 0.004(49.6° - 25°)
  = 8.09W

8.09W is bloody close to our power reading at 25°, Which shows the formula works for this temperature range.
Hmm.. Further testing required in the cool temperature ranges to see whats going on there...

John, as you mentioned earlier, i think the best option is to just limit the temperature range for testing panels.

PS: We can certainly see the benefits here of running with a cool panel. Marc, i think if you cooled your panels before your test lap, you would of broken the 15.17 record.

Tony, interesting that you think it is linear at higher temperatures but not at lower temperatures. Even some condensation could have an effect so it will be interesting to see if cooling below the dew point introduces non-linearity. I know that last year (I think) in WA Winthrop had big problems because they overcooled their panel and it frosted up, slowing the car dramatically.

Yes Tony. I have no doubt that the test car would have gone close to, if not go under, 15 seconds in the right conditions.

The times that we were getting after the event was over certainly indicate that. The panel was also on the warm side so who knows what could have been done in 120% with a cold panel.
After all, even the 15.75 seconds in 108% recorded by last year's car which has almost twice as much air drag would have gone around 15.5s in 120% (less weight I know, but also poorer panel FF).

Reducing the temperature increments at which you are taking measurements would be handy. Taking measurements at every say 5 degrees as the panels warm up on the light box could be a possibility.

Whether there is a general trend for a whole variety of panels or not, it would be fairest in the end to just limit the temperature range for panel testing anyway. That way, even if the correction formula is still a bit off, there cannot be any real advantages gained.

John, whether Winthrop frosted up their panel or stuffed up their electronics setting last year, there was certainly a dramatic loss in performance in their quarter final when they got knocked out against the eventual winner NFG. Had this not of occurred, they could quite possibly have won the event.

Im trying to keep my increments large, because between the time i take a temp measurement, and run the test, the panels temp would of already changed by a few degrees, especially if its far from ambient.

Plus we need to take into consideration the accuracy of our measuring equipment.

Just received this from Ian yesterday with regards to panel power variation with temperature.


As can be seen, he has tested a number of different panels.
All produce results close to what is expected when going by the power vs temperature correction formula in the regulations.

Tony, John tells me that you found that your initial test results at below zero degrees were indeed affected.

That seems to clear a few things up then.

miseli

I found my results to be linear like Ians, in higher temperature ranges, however, in cooler temps like 0° or less they lost linearity.
Ive got my data in excel, ill put it up as soon as i get home from work.

Please do.

Have you taken any measurements at something a little above zero? Say 5 or 10 degrees?

Here are the NanoMETER output I/V curves for the JJ Solar Cells.
They have been plotted for various temperatures from -3° to 51°.

I have a feeling the slight dips in the curves at around the 10volt mark are due to non-uniform temperatures across the panel surface. I had variations of +/-4° across the panel.

As can be seen, on the temperature vs power curve, there is a slight non-linearity at lower temperatures, but that could be due to various things like non-matched cells, non-uniform temps.



Ill the upload data for the other modules in the next few days.

Just had a look at the file and most measurements appear not to deviate too far from both what is expected when going by the power vs temperature correction formula and what Ian was getting with his tests.

I think the formula is close enough, but it would probably still be a good idea to limit the temperature range for panel testing in the regulations just to stop teams from even attempting to play around with ridiculous panel cooling/heating for scrutineering.

It'll be interesting to see what kind of results the other modules will produce.