Weber and Coulomb's MX-5

[Thalass]

What range are you wanting?

My bare minimum is 50km, but I'd like more if possible. The main problem is that everything written online about estimating the required battery capacity is written for 144v DC cars. And I want to use industrial induction motors.

[coulomb]

We wanted to find the temperature coefficient of our Thunder Sky LiFePO4 cell today. We immersed it in almost boiling water, and waited for the temperature at the terminals to be close to the temperature of the water.

This took so long that we put the container of water inside the peltier cooler/heater in heater mode, and spent some time arguing over battery monitoring systems and battery racks.

When we decided it was reasonably equalised, the temperature was some 49 °C, and the open circuit voltage went *up* from 3.335 to 3.337, occasionally 3.338.

We then left it in the fridge for a few hours; the temperature read 6.5 °C. The voltage was 3.335.

Maybe we need to leave it at a new temperature for hours?

Perhaps the internal voltage of LiFePO4 isn't affected by temperature? Certainly its internal resistance is temperature sensitive.

Edit: cold voltage was 3.335, same as ambient voltage.

[antiscab]

Perhaps the internal voltage of LiFePO4 isn't affected by temperature? Certainly its internal resistance is temperature sensitive.

Why do you think the internal voltage is affected by temperature? I haven't witnessed this on any battery (though to tell the truth, i havent looked for it either).

can you repeat the discharge testing for a hot cell for the racers out there?

interesting stuff guys

Matt

[weber]

Why do you think the internal voltage is affected by temperature? I haven't witnessed this on any battery (though to tell the truth, i havent looked for it either).

Lead acid battery chargers sometimes have temperature sensors and they reduce the float voltage by about 30 mV per cell per degree Celsius rise in temperature.

can you repeat the discharge testing for a hot cell for the racers out there?

Happy to oblige. What would you have us do differently? Start with the cell at 30°C? Pull 10C amps?

As I understand it, Thunder Sky's 10C discharge test is at 25^C and goes for 8 seconds and if the voltage doesn't go below 2.0 V it is considered OK.

[weber]

But perhaps the temp coeff of float voltage is not related to the temp coeff of internal or "rested no-load" voltage. This Google Book shows an extremely low coefficient of internal voltage for lead acid too.

[coulomb]

Why do you think the internal voltage is affected by temperature? I haven't witnessed this on any battery...

? With say gel lead acid, figures like the maximum charge voltage of say 14.4 V for a 6-cell battery are supposed to be derated my so many millivolts per degree Celsius. So an uncompensated charger is often overcharging a battery, since it forces the battery to 14.4 even when it is hot, and it should be limiting to 14.3 V or less. In very cold climates, a non temperature compensated charger could undercharge the battery.

My solar charge controller has the option of a temperature probe, and a choice of various compensation profiles.

So for our Lithium BMS, I was going to temperature compensate the threshold voltage against which the three critical voltages are compared (being low volts, high volts, and bypass volts, typically 2.5 V, 4.2 V, and 4.0 volts for Thunder Sky).

But now, it seems that this would be a waste of effort. In fact, I'm thinking that a temperature sensor per cell is probably also not needed.

[antiscab]

[With say gel lead acid, figures like the maximum charge voltage of say 14.4 V for a 6-cell battery are supposed to be derated my so many millivolts per degree Celsius. So an uncompensated charger is often overcharging a battery, since it forces the battery to 14.4 even when it is hot, and it should be limiting to 14.3 V or less. In very cold climates, a non temperature compensated charger could undercharge the battery.

It was my understanding that the temperature compensation was needed due to the secondary reactions inherent in the lead acid chemistry.

from my table of standard reduction half equation potentials:
O2(g) + 2H+ +2e- --> H2O2   +0.68v
H2O2 + 2H+ + 2e- --> 2H2(g) +1.77v

combine the two and you get hydrogen at 2.45v

given that at under charge, the voltage is usually limited to 14.4v (2.4v cell average), it is possible some of the cells in a block are already gassing (by a small amount if the pack is well balanced).

the amount of current this process shunts (steals) from the charging process varies with temperature, and how far above 2.45v the cell is.

The Arrhenius reaction-temperature relationship shows the reaction rate doubes (approx) with every 10 deg C increase in temperature.

Therefore, when operating at above rated temperature, to keep the shunted current below a certain level (so the current applied to the battery actually charges it), the voltage has to be lowered.

operating above rated temperature without temperature compensation especially has to be avoided with Gel and AGM batteries, as the recombiners give off heat when the hydrogen is recombined with oxygen (otherwise you get thermal runaway).

when operating at below rated temperature, the battery internal resistance will be much higher, so charge rate is reduced.
at lower temperature, the voltage can be raised, as the gassing reaction is more subdued anyway.

charging at lower voltage than specified will still allow a lead acid battery to fully charge, it will just take a long time (13.6v at 25 deg C takes 24Hrs from memory).
you only run into trouble if you don't allow the battery long enough to reach full charge.

my chemistry is a tad rusty (it was from all the way back in yr12), so if ive stuffed up, do point it out

back to the original comment, im of the opinion optimum charge voltage only changes for chemistries that have secondary reactions in the course of normal operation. LiFePO4 doesn't have secondary reactions (that i know of).

Matt

EDIT:
In case someone reads this through and gets the wrong idea, i made a mistake.
The voltage at which gassing occurs *does* change with varying conditions.
the voltage above which a reaction occurs is modelled by the Nernst equation:
http://en.wikipedia.org/wiki/Nernst_equation

basically the variable that have greatest effect on cell voltage are concentration of reactants.
The concentration of reactants depends a bit on cell construction. lead acid does have a bit of a wide variation in reactants concentration.
im lead to believe the concentrations vary markedly less in lithium cells.

this is something that is learnt in 1st or 2nd year university if you are studying chemistry (which im not, but i do have friends who are who set me straight).

[coulomb]

Ah. All this time, I actually thought that the open circuit rest voltage varied with temperature, and that's why you temperature compensate.

That makes lead acid even cruder in my eyes, and LiFePO4 relatively more sophisticated. It also explains why there was practically no change in open circuit voltage with our cell. And no temperature compensation is needed.

Thanks for the chemistry lesson!

[coulomb]

With Sky Energy SE-40AHA cells on the way, our thoughts turn to battery racks. When thinking about battery racks, our thoughts turn to welding. So Weber splashed out and bought one of these:



It's a MIG welder. For those not familiar, you usually use it with gas, and instead of a welding rod, there is thin welding wire that is fed automatically through a "nozzle". You can also weld gasless, with different wire that has flux in it. This one is a 165A model, and cost about $1000.

That's a "solar powered" helmet on top that lets you see in normal light, then turns dark in milliseconds when the arc starts. A friend of ours says that it's essential, so we got one.

Yes, that's a Tesla Roadster (Mattel version) in front of the helmet. I think it's self explanatory

We read on the 'net that these Cigweld welders have the "coarse" and "fine" labels round the wrong way. That seemed to agree with our experience, so we measured the open circuit voltage at the tip with a multimeter while changing the coarse and fine settings. Sure enough, they are reversed. The voltage ranged from 18 V to about 44 V (from poor memory). We've amended the front panel:


The machine came with a small reel of gasless wire. Initially, we thought we'd avoid all the expense of gas and just use gasless wire. However, we found the results to be somewhat disappointing, and we definitely needed thinner wire for welding to the thin metal that most of the car chassis is made of. (With the provided 0.9mm wire, we mostly blew holes in the thin metal).

So the decision was made to get some 0.6mm and 0.8mm wire and a small bottle of gas suitable for steel welding. Well, it turned out more economical (assuming we do a fair bit of welding) to get a bigger bottle. This one has some 15 kg of "argoshield light" gas (4.0 sm³) at 15.8 MPa (about 150 atmospheres) (edit: had them swapped).



The welding machine came with a regulator, but no flow meter. We were supposed to adjust the flow to 15 L/min, but how do we know when we get there? Weber came up with the idea of bubbling the gas into an inverted, calibrated container, and we time how long it takes to get to one and later two litres. Here is the setup; the bucket is a "bulldozer" model that is commonly used in cleaning, and was to hand.



There is a small piece of ordinary garden hose on the end of the welding tip (nozzle removed), to keep the welding tip dry. We eventually got it to bubble in two litres of gas in just over 8 seconds, which is about right.

We had very variable results for a while. Weber suspected the wire feed, since the noise wasn't the same as before. Indeed, the wire had almost "knotted itself" and was attempting to feed under another loop of wire. We had to unfeed the wire, untangle, and refeed (twice). It settled down after that. It might have had something to do with the wire unsproinging (TM) when first installed.

Another thing we found was that the metal had to be very clean and completely free of paint to get good results. That stands to reason to me, but we had heard that the paint would just burn off, so there was no need to sand it off first.

[acmotor]

Neat. but I hope the gas bottle was only 15MPa

unsproinging (TM)

[coulomb]

acmotor: edited, thanks.

Re the welding gas: I meant to mention that it seems that you can't own a bottle outright, you buy the gas ($75 in this case) then rent the bottle ($40 per quarter?). That way the supplier is responsible for the upkeep and safety of the bottle, I guess.

[weber]

We figure we need to be able to weld anything from 1 mm thin car panels up to 5 mm thick angle for battery-cage mountings. According to BOC Gasses the "Argoshield Light" makes it possible to weld the thin material by giving a cooler arc. But you also need to use the thinner feed-wire (0.6 mm) and lower voltage as well.

BOC's Argoshield Light has 92% Argon, 5% CO2 and 3% O2. Apparently Supagas have a similar product for welding thin metal called "Supashield O7" with 93% Argon and 7% O2.

Supposedly we will still be able to weld 5 mm angle with the Argoshield Light, provided we use 0.8 mm feed-wire and set the voltage flat out. Haven't tested this yet.

The Transmig 165 is the most powerful in the CIGWELD range that you can still plug into a standard single-phase 10 A socket outlet. But it has to intermittently pull 28 A from that socket to achieve its max output current of 165 A (which it can do with a max 16% duty cycle).

Despite this, the welder comes with only 1 mm^2 wire in its 240 Vac lead! At least they tell you in the manual that you will have to get this upgraded to 2.5 mm^2 to acheive the full rated current of the welder. But why didn't they just install the right cable to start with?

So we got that cable upgraded and got a dedicated circuit with a 15 A outlet put in the switchboard. This will also be useful for EV battery charging.

[woody]

I've had a closer look at the SE40AHA specs on the evcomponents website. Most interesting to me:
4C continuous discharge (20 seconds). 20 Seconds isn't very continuous...
12C peak discharge (5ms). 5 ms isn't very long either. Some others quote 10 seconds peak.
Internal Resistance 0.9mOhm. This is waaaaay lower than thunder sky. If it's right you should still be able to do 4C (160Amps) at 3.06 volts x 208 cells = 101.7 kW for 20 seconds.
I'll be very interested to see the tests
cheers,
Woody

[acmotor]

Waiting for some close up pics of your welding skills !

[weber]

Waiting for some close up pics of your welding skills !

We'll let you know when we've got some. Skills that is.
Some good free video tutorials here:
http://www.mig-welding.co.uk/tutorial.htm

[coulomb]

I've had a closer look at the SE40AHA specs on the evcomponents website. Most interesting to me:
4C continuous discharge (20 seconds). 20 Seconds isn't very continuous...
12C peak discharge (5ms). 5 ms isn't very long either. Some others quote 10 seconds peak.

I totally agree. Continuous is continuous, surely. I wonder where they get that information from; Sky Energy still haven't updated their web site with actual data. The Chinese are so frustrating in that respect.

Internal Resistance 0.9mOhm. This is waaaaay lower than thunder sky. If it's right you should still be able to do 4C (160Amps) at 3.06 volts x 208 cells = 101.7 kW for 20 seconds.
I'll be very interested to see the tests

Us too

[Powered By DC]

I talked to Sky Energy about the specs. They say they put the 20s and 5ms as concervative ratings so people are not tempted to push the cells too hard. They say the the cells can do up to 5C constant until flat but the cycle life will suffer if this is done often . Same with the 5ms for the 12 C rating it can do it longer but the cylcle life will suffer.

The specs on our website came directly from Sky Energy. They are changing their model sizes again and the new sizes and Specs will be on our website within a few days.

Dave Kois
Powered By DC, LLC
EV Components, LLC
http://www.evcomponents.com
253-988-5020
Skype dkoisii

[weber]

I talked to Sky Energy about the specs. They say they put the 20s and 5ms as concervative ratings so people are not tempted to push the cells too hard.

Thanks for the explanation Dave. But I presume you can see that a "continuous" rating that you can only do for 20 seconds is a contradiction in terms. And that 5 milliseconds is no better than zero as far as an electric vehicle is concerned.

What we really care about is the direct comparison with the familiar Thunder Sky cells. i.e. how much current can you pull out of an SE cell continuously until the cell goes flat and have the same shortening of life, or perhaps more measurably, the same cell temperature rise, as a Thunder Sky cell at its continuous rating of 3C?

You've already told us that the SE can do 4C in comparison to the TS 3C so we'll be holding you to that.

[Powered By DC]

I will be testing them side by side TS vs SE and posting the results to our website as soon as I get a chance.

Dave Kois
Powered By DC, LLC
EV Components, LLC
http://evcomponents.com
253-988-5020
Skype dkoisii

[EV4AV]

Thanks for showing me your project, Coulomb and Weber,
It was great to see people taking the electric car seriously, most people I meet think I am nuts when I say I am making an electric car.
I am more than happy to give you guys a hand with this project, Machining your adaptor plate, helping you through some manufacture hurdles etc, there is often an easier way to do stuff which is why its great to brain storm. I can help you sort out your headlight modifications, and help fabricate mounts etc. I definately think that we should investigate other options for battery holsters, Fibreglass/Aramid composites perhaps, they are stronger and lighter and more space conscious, providing you have an engineer that is friendly to the idea. I hope I didn't stop you getting too much done when I came to visit. This will be an awesome car when complete. I look forward helping where I can.
Cheers, Anthony.

[weber]

No worries, EV4AV. We totally enjoyed your visit. You gave us lots of useful ideas. You more than made up for our missed welding practice by telling us the secret is to "grind the galvanising off, guys".

[coulomb]

We have a controller for the MX-5:



It's a monster! 75/90 kW Control Techniques. Many more photos and details to follow.

[coulomb]

Welcome back, AEVA forums!

In the break, we've done what any self-respecting EV'er would do with a $13,000 controller: took it apart


Check out the soup cans! Those little chips on the PCBs on the capacitors' terminals are 68K bleeder resistors; they are something like 4S3P.

Edit: those are 17 mm bolts for the motor terminals. The L1-L3, DC+, and brake terminals on the other side are the same size.

An overall view:


The fan in the upper left appears to be just for the capacitors! The IGBTs are under the main board, with the motor terminals to the left of this photo. The PCB at the right (with the rightmost yellow transformer) is the SCR firing circuit; more on that later.

The connections to the main caps are via these interesting double sided straps (hopefully with very good insulator between them). At the extreme lower right, you can see the corner of one of the filter inductors. The ribbon on the left goes to a sort of mini mother board, where up to three daughter boards can be plugged in. We have one of these, which is a processor that will take over most of the speed and torque control from the main processor.


Here you can see the three large torroidal chokes, along with a hairy hand for comparison. These are directly in series with the L1-L3 input terminals. Instead of getting rid of these and their ~20 kg weight, we may use them on the motor side. If we succeed in using the controller as the battery charger, we'll probably need them anyway.


Here is the honeycomb style main heatsink. We were hoping for fins, so we could make more space by grinding them off here and there. However, there are several places where we can get the size down, and miraculously the controller actually almost fits as it is (see later pictures).

[coulomb]

The fan is actually removable:
   

Removing the SCR firing board. It turns out this was important (so we found out it's not a bridge rectifier, but an SCR/diode hybrid rectifier):


The Semikron SKKH162/16E modules (160 A 1600 V):


These SCRs are used to limit inrush current to the capacitor bank. If we wanted to power up the unit on single phase power, we'd want a voltage doubler circuit, but the upper half of the circuit would not turn on (as the transformer that powers it would not see any potential difference between one phase and the next). So the SCRs would not come on, while the lower half (diodes) would. This could reverse bias the upper capacitors. Good thing we weren't stupid enough to do that!   

At the motor end are the expected current sensors (except 3 of them; often there are just 2), but the devices at the bottom are a bit of a mystery:

[coulomb]

We had need of getting to the bowels of the controller today. There be IGBTs:

The blue boxes are 1uF 1000 V capacitors, on the DC bus. To supply that little blast of energy to the IGBTs while the main capacitor banks get their charge down there via the bus bars (I guess).

We managed to borrow some power suitable to power the controller today. These multimeters were across the two halves of the capacitor banks:

The one on the right is a $10 jobbie, and gave us some cause for concern when it told us that the upper bank was going all over the place (wildly varying voltage) when the controller was working. But it was just that it was not true RMS reading; it didn't like the sharp pulses on the DC bus. When we replaced it with another multimeter, all was well.

So we were able to connect White Suzi's motor, and get it to spin:


We were able to do a rotating autotune where it spinned the motor up to 2/3 rated speed, use the control panel's "joypad" (4 arrow buttons), and also wired up a 10K linear pot for speed control: