Model S Gen2 Charger Efficiency Testing

[wk057]

(Been a while since I started a thread here! )

I recently finished some efficiency testing on the Model S Gen2 chargers (pre-refresh, but VIN > ~20k). Tests were done at 240V single phase.

• 5 amps, 82.77%
• 10 amps, 89.53%
• 15 amps, 89.06%
• 20 amps, 90.04%
• 25 amps, 91.03%
• 30 amps, 90.37%
• 35 amps, 90.75%
• 40 amps, 91.75%
• 45 amps, 90.81%
• 50 amps, 90.88%
• 55 amps, 90.55%
• 60 amps, 90.32%
• 65 amps, 90.21%
• 70 amps, 90.55%
• 75 amps, 90.77%
• 80 amps, 91.63%

Keep in mind that the gen2 chargers are actually three smaller chargers in one, and that dual chargers would be six smaller chargers. So, during the current ramp up different numbers of these are online which explains the dips and jumps at say, for example, 25->30A.

The data was collected over several days from two different vehicles with dual gen2 chargers and averaged. The number represents overall system efficiency (actual power into the battery pack while charging vs actual power into the charge port) which would include coolant pumps and other items needed for charging. Tests done at room temperature (~70F/21C)

This doesn't take into account losses through the EVSE cabling and such, which will vary depending on design and other factors, but shouldn't be appreciable enough to matter when pitted against charger system efficiency.

TLDR: For highest charger efficiency, charge at 40 or 80A.

 

[msnow]

(Been a while since I started a thread here! )

I recently finished some efficiency testing on the Model S Gen2 chargers (pre-refresh, but VIN > ~20k). Tests were done at 240V single phase.

• 5 amps, 82.77%
• 10 amps, 89.53%
• 15 amps, 89.06%
• 20 amps, 90.04%
• 25 amps, 91.03%
• 30 amps, 90.37%
• 35 amps, 90.75%
• 40 amps, 91.75%
• 45 amps, 90.81%
• 50 amps, 90.88%
• 55 amps, 90.55%
• 60 amps, 90.32%
• 65 amps, 90.21%
• 70 amps, 90.55%
• 75 amps, 90.77%
• 80 amps, 91.63%

Keep in mind that the gen2 chargers are actually three smaller chargers in one, and that dual chargers would be six smaller chargers. So, during the current ramp up different numbers of these are online which explains the dips and jumps at say, for example, 25->30A.

The data was collected over several days from two different vehicles with dual gen2 chargers and averaged. The number represents overall system efficiency (actual power into the battery pack while charging vs actual power into the charge port) which would include coolant pumps and other items needed for charging. Tests done at room temperature (~70F/21C)

This doesn't take into account losses through the EVSE cabling and such, which will vary depending on design and other factors, but shouldn't be appreciable enough to matter when pitted against charger system efficiency.

TLDR: For highest charger efficiency, charge at 40 or 80A.

So you're saying that car's above VIN ~20k came with gen2 mobile chargers and 80A HPWC's purchased in that timeframe are also gen2?

 

[wk057]

I'm referring to the chargers that are built into the car under the rear seats which actually do the AC to DC work, not the EVSE (HPWC/mobile connector) which is basically just a glorified "smart" extension cord.

 

[callmesam]

So are the onboard AC to DC chargers the same as the ones in the Supercharger cabinets?

 

[Cosmacelf]

It would be interesting to contrast these results with the newest 48A/72A chargers in the current cars.

A methodology question. Are these numbers instantaneous numbers gathered after a set time after charging starts, like 5 minutes, or are they session averages, or session averages after a set time, or ??

 

[wk057]

So are the onboard AC to DC chargers the same as the ones in the Supercharger cabinets?

Yes. The superchargers use 12 of these. The newest ones may use something newer, but I'm not sure.

It would be interesting to contrast these results with the newest 48A/72A chargers in the current cars.

A methodology question. Are these numbers instantaneous numbers gathered after a set time after charging starts, like 5 minutes, or are they session averages, or session averages after a set time, or ??

I had a script do the following:

• Car parked/off/locked/untouched for 15 minutes.
• Reading taken of idle/non-charging DC power usage average over 30 second period.
• Charge input current set to current test setting and charge started.
• 5 minutes into charging a 5 second average taken on input and output power as measured by the chargers and BMS. Repeated a 5 second average every minute for 25 minutes.
• Charge stopped. Subtract initial idle/non-charging power usage from all DC side numbers.
• Repeat for different amperages.
• Repeat entire routine again.
• Repeat on second vehicle.

Initial preliminary findings revealed output current had virtually no impact on the efficiency, so SoC was not a sufficient factor to bother controlling for testing, although the tests were performed between ~20% and ~90%.

Lots of things end up being factored in due the methodology, which is why I say it is a system efficiency. For example, 80A charging is slightly less efficient than 40A charging because pumps/fans would run slightly more at the higher rate, resulting in slightly less power getting to the battery per unit of input power.

 

[AntronX]

Very nice. Are you noticing any degradation in capacity of your tesla modules after one year of solar operation?

 

[Cyclone]

Model S Gen2 chargers (pre-refresh, but VIN > ~20k)

gen2 chargers are actually three smaller chargers in one

Interesting. Does anyone have any pointers to where I could learn more about the differences between the gen 1 vs. gen 2? I imagine my 78xx car w/ dual chargers has gen 1, but I would learn for sure after reading more.

 

[emir-t]

Very nice work. Thanks for sharing as always. Speaking from a European framework;

Initial euro Model S' (with the gen2 charger in question) would not have charge single phase 32 amps, 7kW because a charger used 3 different charging modules as you said and each could carry 16 amps per phase. So they shipped every Model S with dual chargers, except the second one was software locked if you didn't buy the option. Second charger was used when charging 32amps (7kW). To make this possible they designed the UMC blue connector (single phase 32amps) so that CP of the Type 2 connector would signal 32amps total available to the car, only just a total of 32amps. 16 on both L1s of each charger.

Someone plugged this UMC blue connector into a ZOE and ZOE interpreted this as having 32 amps available each phase, meaning 96 amps on the neutral wire. This fried up both ZOE and UMC.

Later on they, via software, made it split the 32 amps to L1 - L2 - L3 of a single charger so it was resolved. Why they went with the complicated way first is beyond me.

Moreover, later after the facelift, why they went for a single 72 amp charger with the initial option locked to 48 amps via software is beyond me. They reduced max AC charge from 22kW to 16.5kW. This doesn't matter that much for markets with a lot of Superchargers but for under developed Tesla markets in europe this is a bad thing.

 

[CSFTN]

I am one of those readers who is very interested by this, has limited understanding of engineering, and missed this when initially posted. I too miss what used to be the preponderance of intelligent conversations.

A simple question: shouldn't the most efficient charging amperage be evenly divisible by 3 (assuming the 3 divisions of each charger are essentially the same)? Feel like trying this again at 39, 42, 45, 48, and 78 amps? Someone (maybe FlasherZ?) posted a similar but less thorough analysis back in ~ 2013-2014, and suggested that 48 amps was the most efficient charging level, IIRC.

 

[mspohr]

Thanks for this, wk057. I missed it the first time around. This is very good information.
I've been trying to understand the overhead losses from cooling pumps, fans, etc when charging and it seems there might be a clue here. If you look at the 5 amp efficiency of 82.77 and the jump to 10 amp efficiency of 89.53, would it be reasonable to assume that the 7% difference in efficiency (about 90 watts) is due to the pumps, fans, etc. which do not change much between 5 and 10 amps?

 

[wk057]

I am one of those readers who is very interested by this, has limited understanding of engineering, and missed this when initially posted. I too miss what used to be the preponderance of intelligent conversations.

A simple question: shouldn't the most efficient charging amperage be evenly divisible by 3 (assuming the 3 divisions of each charger are essentially the same)? Feel like trying this again at 39, 42, 45, 48, and 78 amps? Someone (maybe FlasherZ?) posted a similar but less thorough analysis back in ~ 2013-2014, and suggested that 48 amps was the most efficient charging level, IIRC.

You know, I had thought to try this with increments of 3 but decided against it for simplicity. If I have a schedule that fits with doing this testing again some time in the near future (it takes a significant amount of time, actually, and I need to have driven a decent amount prior) I will rerun from 6A to 78A by 3 and see if there are any noticeable differences. Edit: Actually, the differences shouldn't be too noticeable. Once all three charger modules are online the load is split among them evenly at 0.1A increments if I recall correctly. I should try to force debug output and log the operation of the individual modules when I test again.

Thanks for this, wk057. I missed it the first time around. This is very good information.
I've been trying to understand the overhead losses from cooling pumps, fans, etc when charging and it seems there might be a clue here. If you look at the 5 amp efficiency of 82.77 and the jump to 10 amp efficiency of 89.53, would it be reasonable to assume that the 7% difference in efficiency (about 90 watts) is due to the pumps, fans, etc. which do not change much between 5 and 10 amps?

That is a valid estimate, yeah. The MCU, all modules, charger itself, BMS/contactors, pumps, fans, etc. The overhead definitely varies quite a bit during a charge, though, and you're calculating what would be the average overhead.

 

[Zapped]

Yes. The superchargers use 12 of these. The newest ones may use something newer, but I'm not sure.



I had a script do the following:

• Car parked/off/locked/untouched for 15 minutes.
• Reading taken of idle/non-charging DC power usage average over 30 second period.
• Charge input current set to current test setting and charge started.
• 5 minutes into charging a 5 second average taken on input and output power as measured by the chargers and BMS. Repeated a 5 second average every minute for 25 minutes.
• Charge stopped. Subtract initial idle/non-charging power usage from all DC side numbers.
• Repeat for different amperages.
• Repeat entire routine again.
• Repeat on second vehicle.

Initial preliminary findings revealed output current had virtually no impact on the efficiency, so SoC was not a sufficient factor to bother controlling for testing, although the tests were performed between ~20% and ~90%.

What device did you use to meter the power usage? I'm looking at a TED device ( The Energy Detective ) to monitor the power consumption on an EVSE

 

[hacer]

It would be interesting to contrast these results with the newest 48A/72A chargers in the current cars.

A methodology question. Are these numbers instantaneous numbers gathered after a set time after charging starts, like 5 minutes, or are they session averages, or session averages after a set time, or ??

I did some work on this some time ago. My methodology is similar, but rather than taking a series of 5 minute averages, I used the integrated total energies into the battery and from the AC service over the complete charge cycle. I did it for many different charge levels and I wasn't able to control for temperature. All of the test data was gathered during the summer, but mostly moderate temperatures inside my garage (usually < 80 F). The new charges are more efficient than the old ones, and seem to have a sweet spot at 25A, at least for the 75kWh battery in my car. The results can be seen here Ideal Charge Rate?? the raw data is here: Charger Efficiency.

 

[wk057]

What device did you use to meter the power usage? I'm looking at a TED device ( The Energy Detective ) to monitor the power consumption on an EVSE

I mainly used data from the car itself and confirmed that data with TED.

I did some work on this some time ago. My methodology is similar, but rather than taking a series of 5 minute averages, I used the integrated total energies into the battery and from the AC service over the complete charge cycle. I did it for many different charge levels and I wasn't able to control for temperature. All of the test data was gathered during the summer, but mostly moderate temperatures inside my garage (usually < 80 F). The new charges are more efficient than the old ones, and seem to have a sweet spot at 25A, at least for the 75kWh battery in my car. The results can be seen here Ideal Charge Rate?? the raw data is here: Charger Efficiency.

Interesting. I have not been able to perform tests on the Gen3 charger yet. However, I now do own a Model X with a 72A charger and may do so in the future.

 

[n2mb_racing]

How efficient would charging at 240V 12A or 16A be? It seems like it would be a lot higher efficiency vs 120V @ 12 or 16A. From your data, it looks like it would be just under 90%.

Looking at my own car with Gen 2 chargers and data from Teslafi.com, it seems I get about 92-93% efficiency when charging at 24A 240V (NEMA 14-30 dryer outlet) but only ~73% efficiency when charging off 120V @ 12A.

I'm thinking about long-term charging off a NEMA 6-20 outlet, since it might be the easiest to add to my existing 100A panel that probably doesn't have enough room for another 50A circuit. It looks like there'd be less than a 1% hit in efficiency, as long as I'd be willing to live with charging at 11 MPH, vs 24-28 MPH. Thanks!

 

[Dbox95]

This is really useful. I want to ask if anyone has had a chance to do a similar analysis for vehicles equipped with the current high capacity (72A) chargers? I'm guessing some of the findings might carry over but curious to see how the newer charger performs.

 

[skrenes]

Fascinating thread. I may gather similar results for my car in the future with what I think is Gen1 chargers. I think its important to note these numbers include a few things:

1. idle losses during charging (active systems, pumps, etc)
2. losses increase as more chargers are required
3. each charger has a worst case efficiency (low utilization) and peak efficiency (usually around 50% load)
4. As the wall load increases, the voltage usually sags (I've seen as much as 20v sags from 5A to 80A), signifying energy losses in the cable and likely efficiency losses in the charger, especially near peak utilization.

This is multivariable, but I took the data points that @wk057 supplied and put it into a spreadsheet with guesses to the wall voltage (240v at 5A to 236v at 80A), charger efficiency (94% to 92%), and then played with static values to match his overall efficiency results. It looks like there's about 140-170 watts of idle losses from 5A to 40A and then about 280 - 310 watts of idle losses from 45A to 80A. So the best charging routine is actually to charge at 40A, if you don't need your car in a rush.

The reason the 5A values look so much less efficient is that the idle losses of having the car active with pumps and such is relatively high compared to the amount of energy going into the battery. But it's important to note that as you charge at a higher rate, those idle losses increase, just not proportionally to the increased amount of power that goes into the battery.

The reason I note this is that I saw that the TWCManager project has a minimum charge current of 12 A to mitigate efficiency losses due to what I think is a misunderstanding of the situation. In reality, if my solar panels are producing say 2.6 kW (just below the 12A threshold) and that's going back to my utility, I lose at least 20% of the cost of the energy to distribution fees, which is about 4x more than the idle losses of charging at 5A. Another way of saying this is, if your distribution fees are about 20% of your energy cost, it's worth charging your car at 140W / 20% / 240v = 2.92 A or greater. Depending on how your distribution fees are, you may have different values. It's a bit abstract but I thought I'd share anyway.

 

[rejedejo]

I did some work on this some time ago. My methodology is similar, but rather than taking a series of 5 minute averages, I used the integrated total energies into the battery and from the AC service over the complete charge cycle. I did it for many different charge levels and I wasn't able to control for temperature. All of the test data was gathered during the summer, but mostly moderate temperatures inside my garage (usually < 80 F). The new charges are more efficient than the old ones, and seem to have a sweet spot at 25A, at least for the 75kWh battery in my car. The results can be seen here Ideal Charge Rate?? the raw data is here: Charger Efficiency.

How are you getting the BMS data?

 

[hacer]

How are you getting the BMS data?

I got it through the JSON API, but it can also be had from the CAN bus in the car.