Bigger Battery Booster

Last December we had a very long stretch of dull days and the tempest didn’t charge at all during that period. Of course the tempest went to the highest power save mode and stopped measuring wind, rain and lightning. In an attempt to prevent this from happening again, I did several experiments, including a very promising reduction of the fastest sampling rates to be no higher than 4 times a minute. Weatherflow probably will come up with better power save settings in the near feature. Another solution is the upcoming power booster from Weatherflow, which basically gives you various options to charge the unit through a wire. As mine is mounted on an acrylic mast, which looks stunning, I didn’t want to add a wire to it.
Yet another solution is to replace the internal battery with a bigger one, and that is what I just did.

This is what arrived yesterday from China.

Four pieces of 2500mAh LTO batteries. ( Those are the same types as currently inside the Tempest, but almost twice as big. Not only bigger in capacity but also bigger in size.

In order to replace the original battery, some destructive work has to be done and you will loose your warranty if you do so. Therefore I don’t recommend this solution, but I still went ahead with this modification.

Start with turning the unit off and opening it up by removing the four screws at the bottom.

Next step is to remove the two blue connectors. Be sure to mark them first so you know which one goes where. Do this by first sliding up the black parts. They move vertically up from the board.

Unfortunately they seemed to be glued in place by some coating and are very fragile. As you can be seen in the little rectangle, mine broke during the operation (the left two). They are supposed to look like the right one, and don’t have to come out of the white part completely, just slide them up 2 mm. Luckily the central part which holds the blue connector in place is still functional, so I had no problems using them again. Note the blue connector on the right was stuck in on a slight angle. More on that later.

There is no need to remove the connectors to the solar panels.
Next up is to remove the three hex 2 screws hidden deep inside the three holes. They connect the part with the solar panels to the rest. The antenna simply slides out of its position.

Remove the white bottom from the printed circuit board and use a soldering iron to remove the original battery. Now comes the really destructive part. As you can see there is a kind of white cylinder that partly covers the battery. That is too small for the bigger battery, but it also doesn’t seem to have any function, so use a Dremel or similar to cut the plastic part away. Be generous with what you cut away as a bigger battery is going to need to fit.
(I remove the three screws mentioned earlier to separate the top part of the unit from the bottom. But perhaps you can remove this plastic cylinder with the Dremel without doing that. If you do, it might take some work to get the debris out of it. Its just easier when the parts are separated. While typing this I realized that if you are able to remove this white cylinder without separating the top and bottom, there is no need to remove the blue connector I mentioned earlier. You could try, if you don’t manage to remove the cylinder, you can still separate the parts).

Here is my result seen from the bottom, with the remainder of the plastic cylinder next to it. (it looks smaller and out of focus, because it is further away from the camera). Note in the top left of the unit, the bigger hole is next to plastic mounting point for a screw. If you mount the battery normally it will touch it and you cannot close the unit later on.

Time to solder the new battery in place. Take care about the right polarity. On the original battery the plus side is marked with a big black band over the side of the battery, the replacement has its minus side marked.

The separation between the pins of the bigger battery isn’t the same as for the original. Bend the one that is closest to the positive side, so the battery moves a little to the center, away, from the plastic mounting point for the screw.
It turns out it is slightly to big to be mounted against the pcb, but the pins are long and sturdy enough to keep it floating above it. (but see my remark below the last picture)

All that is left is to reconnect the blue connectors again and push the black parts in place. (I used a piece of tape to keep the wire temporarily in place while doing that. I needed that, because my black parts were broken off. Once in place even the broken connectors stay in place without problem. You might use some hot glue to make sure they do).

Turn it on for testing. The green led should turn on. That was a scary moment as it didn’t do that initially in my case. I didn’t know what could have gone wrong, until I remembered to that initially the blue connector was slightly angled. The coating on it also was at an angle, so when I inserted it straight, the coating probably blocked the contacts. Inserting it again, but at a slight angle, fixed the problem (of course the connector should have been inserted straight in the first place).

Job done, one Tempest ready to take on the next winter.

So I thought. Turns out the pins of the battery aren’t sturdy enough to prevent the battery from rattling against the plastic in windy conditions. So I’ll have to open it up again and insert a bit of foam to solve that.
Still I’m quite happy with the results.


Great job, @sunny … leave it to you to think outside the box. Even when the box is cylindrical, with a mushroom-shaped top!


Thanks, I really was laughing out loud (lol)


Sunny your my kind of guy! Awesome job!

1 Like

Its function is to hold the battery in place so the solder joints aren’t the support. Solder joints should not be used as a structural weld as they will crack when frequently flexed, causing all sorts of issues.

A dab of hot glue or electronics friendly silicone would hold the cell in place quite well. This is what is frequently done with automotive rated electronics.

It will be interesting to find out how long it will last. Maybe when you see it at full voltage, cover the solar panels and see how many days it runs before reaching the first power save mode.


Good point, but in this case there wasn’t any room to wiggle the original battery, as it was mounted flush against the pcb.
With my new battery I had to insert some foam, once I did, it also has no room to wiggle. Should be ok now.

Sure thing I will measure how long it lasts by putting it up in a place where it only receives diffuse light (that’s more realistic than covering the solar panels up). I’ve just requested to reset my sampling rate to the normal behavior.

1 Like

Covering them completely will give the lower bound on how long it will last. It will also give an indication of how much the diffuse light adds to the energy available.

Given that one of my Tempests makes it about 5 days from 2.8V to 2.48V and your replacement is 2500mAh compared to 1300mAh of the original, it should be able to run 10 days or more before going into a power save mode.

Do you have any way to do a Voltage vs SOC test on one of those cells? The Tempest drains them so slowly that a 1C discharge curve I find online doesn’t really match what happens in the Tempest. I’d like to get a 0.008C test curve.

It’s true that that would give a lower bound, but it only gives an indication of how much diffuse light helps if you measure both. By now I measured 5 discharge curves with the normal battery, all only in diffuse light. The result varies wildly. All with a constant sampling rate of 4 times per minute. Temperature might be a factor, but the last curves (not on this charge) were measured at higher temperatures, but didn’t give clearly better results.
I don’t have a 0.008C test curve, but you and I are kind of measuring those.

1 Like

What we don’t have is what percentage of the battery capacity is actually available for a fully functioning Tempest. It appears that the Tempest goes into power save mode with a significant portion of the battery capacity still left due to the Tempest not operating down to 2V or so. I’m trying to to determine what the effective capacity of the cell is for Tempest use. For example, is the 1300mAh cell in the Tempest more like a 1000mAh cell? I may have to get a LTO cell to test on my PowerLab8 where I can set the current really low.

my guestimate is that the unit stops reporting when the battery is still around 40-50% full. Suppose that is true, what are you going do to with that knowledge?

I am tempted to perform the same upgrade to my field test unit.
This brings the question…is there any hardware difference between the field test and production Tempest?

I don’t know, just open it up and see if you have the same space available. As long as you replace it with a bigger 2.4V LTO battery, it should be fine.
I just finished charging mine and put it in a place where it receives no direct sunlight. My guess is it will last a lot longer than before.
you can follow it here Tempest

1 Like

I was more concerned with circuit differences between the FT and production units. The board and battery layouts are probably the same.

if the battery is the same type, then the circuitry will never know there is a bigger battery.

I know the first part of the discharge curve is very steep but if it is any indication, things are looking very good. The new battery went from 2.79V to 2.71V in slightly more than a day. The original battery would have dropped down to somewhere around 2.61V by now. I’m excited!


Cool Sunny, How long did your batteries take to arrive? I have one on order (different Ali supplier). Suppliers are using extremely slow shipping - 3 months to NZ, I could swim faster!

6 weeks. perhaps these batteries are not shipped by plane as some people consider them dangerous goods (LTO are amazingly save). I could only find one ali shop. Do you have a link to yours?

I think your right, by ship. My listing has disappeared, not holding out much hope of seeing the battery.

I’d hold it close to my heart, of course. :wink: It would satisfy curiosity, for one. Some have claimed that the columbic charge/discharge efficiency is quite low as far as Lithium-ion type batteries go and a slow discharge/charge cycle would also measure that. A low columbic efficiency has a direct consequence to how well diffuse light is able to charge the Tempest battery.

apparently it takes a few cycles to get to columbic efficiency in the 95%+ range. Not something I should worry about. Charging in diffuse light is a problem. The solar cells don’t generate enough voltage. When you would put them in series to increase the voltage, the problem becomes that the voltage drops very dramatically when you draw more current than the panels can supply. Part of the problem might be the power used to charge the battery. I’m not sure if that power would be reduced to zero (or better close to zero), the power consumption would be low enough to keep the panels working at a decent voltage in diffuse lighting conditions. Currently it doesn’t seem to charge at all in dimmed diffuse light. (mid winter conditions).

1 Like