P85D Power draw numbers do not add up...

[Zetopan]

I somehow came away from the Elon fuse discussion thinking Tesla was employing active "pyro" fuses which have circuitry on board to monitor instantaneous current (within the sample time frame of the circuit) and "blow" the fuse link in a fault condition.

Firstly, my original posting had some errors in it that I need to correct. While the power dissipated in a passive fuse is I^2*R the actual critical value is I^2*T.
I used I^2*R in some places where I should have been using I^2*T. That could just add to the confusion, sorry about that.

The resistance obviously has a tolerance, but that can be pretty well controlled in manufacturing by using what is known as a Kelvin connection to accurately monitor the resistance as a QC check. What has the greatest source of error is the I^2*T, a measure of how long it takes the fuse to blow.

As a simple example, take two fuses that have identical resistances but slightly different geometries. One fuse is slightly thinner and shorter, with the other obviously being slightly thicker and longer. The thicker longer fuse will be able to dissipate slightly more power before self destructing since it has more mass and a larger cross sectional surface area to conduct heat to the non-fusing regions as well as slightly more surface area to the air. These small differences can cause relatively large changes in the time that the fuse takes to do its intended job.

To get a tighter tolerance on the I^2*T Tesla employes active circuitry to monitor the voltage drop across the relatively accurate resistance of the fuse. When the desired I^2*T has occurred the pyro is activated to blow the fuse, which could have otherwise taken too long or too short of time to blow due to the thermal characteristics of the material and geometry variations. You get the T component in the measurement by integrating, also known a filtering. Computing I^2 is not a difficult task these days since the circuitry measures a current proxy (the voltage) and squares it. The result is then filtered and then thresholded to establish the I^2*T limit that is desired. By tightening the I^2*T tolerance in this way the actual current threshold can be increased much closer to the desired limit without suffering from premature fusing.

[QUOTEDo you have a reference for the fuses still being set to 1300 amps?[/QUOTE]
These are not 1,300 amp fuses, they are much lower than that when subjected to a continuous current. The 1,300 amps is the fusing current when subjected to higher currents for relatively shorter time intervals. Musk has publically stated that the original (I^2*T) fusing was set to 1,300 amps because of the tolerance problems and by using pyro with active circuitry the limit could be raised to 1,500 amps (without risk of premature destruction).

 

[Zetopan]

And please note, 600 HP is from the battery, then it goes to the inverter to change into AC power and then to the motor, then a transmission and out to the tires.
It`s anyone`s guess as to losses combined there. 15% minimum. 20+% in reality.

Let's not wander too far into irrelevancies here. The manufacturer's HP rating of their internal combustion engines are done at the crankshaft and do not take into account the power transmission losses at all. Change the elevation and the output power varies, likewise for a less than pristine air filter, different fuels, etc. When someone has their vehicle power measured on a rolling dyno, the dynamometer measures the wheel torque and computes the power by including the RPM, and then they use a lot of fudge factors to estimate the crankshaft torque and power based on assumed losses in the transmission and differential gearing. That is why dynos rarely agree even when the same vehicle is measured, even at similar altitudes. (HP = torque in pounds-feet times RPM divided by 5252, in archaic units. This is why HP = torque at 5252 RPM.)

As previously stated, technically speaking the Tesla does not have a transmission, it has a gearbox. For those who are unfamiliar with the difference: a gearbox has a fixed gear ratio that cannot be changed without dismantling the gearbox. A transmission has multiple gear ratios available that can be selected from. When multiple ratios are required most people seem to prefer having the latter rather that the former and having to stop and dismantle it to change the ratios. The latter has more gears in mesh and thus greater losses that the former, with everything else being equal. Correctly made straight cut gears could lose about 2% per meshing pair. Oftentimes the gears are not straight cut to reduce meshing noise and this is essentially always the case for the large ratio reduction in a differential. However, non-straight cut gears are also less efficient.

 

[jpet]

These are not 1,300 amp fuses, they are much lower than that when subjected to a continuous current. The 1,300 amps is the fusing current when subjected to higher currents for relatively shorter time intervals. Musk has publically stated that the original (I^2*T) fusing was set to 1,300 amps because of the tolerance problems and by using pyro with active circuitry the limit could be raised to 1,500 amps (without risk of premature destruction).

First, a big thank you for educating us on this subject!

Does the above mean that the core fuse is essentially the same as before but by adding "active" measuring circuitry, it is now possible to safely get closer to the short burst limit of the fuse?

Before your write up, I thought they had introduced a completely new type of fuse that could be programmed to blow at a variable amp rate, allowing to go beyond 1500A in the future via a simple software update.

 

[lolachampcar]

I do not know anything about pyro fuses nor do I play one on TV.

That being said, if I were designing one, I would use a much lower resistance value to lower I2R losses in use then use my active I monitoring to integrate energy and blow the fuse according to a look up table (Energy v. Time). Why keep a thinner piece of copper with losses when you can have a thicker piece with less loss and a much wider operation/blow range set in software? I'd manufacture a few, each of which covered a very wide range of amperages.

 

[jpet]

That being said, if I were designing one, I would use a much lower resistance value to lower I2R losses in use then use my active I monitoring to integrate energy and blow the fuse according to a look up table (Energy v. Time). Why keep a thinner piece of copper with losses when you can have a thicker piece with less loss and a much wider operation/blow range set in software? I'd manufacture a few, each of which covered a very wide range of amperages.

Indeed, that's what I thought Tesla did but I haven't got a clue if it is "that simple"... It sort of gave me hope there would be a chance Tesla could be increasing the 1500A to 1600A etc... in the future so that the 691 figure would be achievable given the motor power of our P85D.

 

[lolachampcar]

Even if they did do as I suspect, I can not imagine that fuse ratings would be anything other than established as a one time programmable function at manufacture. The cost and complexity of having the battery's fuses sitting on the CAN bus allowing for field updating would seem prohibitive.

 

[Torben_E]

I am unsure who your "we" is, but I am pretty sure that it does not include me. Firstly, as was quite correctly previously stated, fuse ratings actually allow for relatively short term overloads that far exceed their nominal ratings. Hence the fuses that Tesla is using on the P85D allows for up to about 1,300 amps for many seconds without self destructing (see message #22). Tesla had to limit the maximum current draw to 1,300 amps on the P85D because most fuses have a rather wide resistance and I^2*R tolerances. The "typical" fuse would handle higher current than 1,300 amps, but for reliability the worst case has to be used in the design. If that were not the case some Tesla owners could find themselves stranded with blown fuses even though the "typical" fuse would have easily survived under the exact same conditions. Tesla did not replace a 1,300 amp fuse with a 1,500 amp fuse. They replaced a wide tolerance fuse (having a much lower nominal rating than 1,300 amps) with a much tighter tolerance fuse. The latter fuse, because of its tighter tolerances, lets Tesla increase the current and still avoid blowing the fuse with worst case fuse resistance and I^2*R tolerances.



Secondly, lets do a little simple math. 691 HP = 515486 KW (I am using a slightly higher wattage per HP than reality due to rounding). Now we can divide that wattage by 1,500 amps since that is the maximum current draw. The result is 344 volts (rounded). That assumes 100% efficiency for the inverters and motors. Induction AC motors with copper rotors (what Tesla uses) are more efficient that most industrial motors and this is why Tesla actually manufactures their own motors. A very good industrial induction motor can be 97% efficient at its maximum load. I don't know the efficiency of the Tesla inverters but I suspect that it is very good. Lets assume for the sake of argument that it is only 90% (it is likely better than that). Given those numbers, the battery voltage would need to be 394VDC (rounded) to deliver
691 HP.
If I instead use 95% efficiency for the motors and 95% for the inverter, the battery voltage would have to be 381 VDC. Note that the nominal terminal voltage of the P85D battery is 400 VDC. At that voltage and 1,500 amps the battery output power is 600 KW but it will drop pretty rapidly from there to a slightly lower voltage under load. That electrical power translates to 804 HP. Since the P90D "Ludicrous Mode" rating is 762 HP, that allows for only a 2% efficiency loss, unless the terminal voltage has increased somewhat on the higher capacity battery.
Manufacturer internal combustion engine HP and torque ratings are measured at the crankshaft, these do not include the losses due to the transmission and drive axel gearbox (usually a locked up differential). Since Teslas have a simple gearbox rather than a complex transmission the motor gearbox transmission losses will be lower than your common automotive internal combustion engine power transmission losses. Hence the Tesla will be placing more of the generated power to the automobile axels than an internal combustion engine.

Hi Zetopan...

I am not quite sure I can follow you with the explanation that the battery can deliver the 691 hk to the two motors. Please see this link with our own investigation about the subject.

http://teslaforum.dk/p85d/battery-technology/

http://teslaforum.dk/p85d/

It basically suggest that when you have an internal resistanse of the battery pack og 0,051 Ohm, an 70 V drop will occur on the battery terminals and thus it is not 400 VDC x 1.300 A but only 330 VDC x 1.300 A.

This is also in line with what se see from the REST data.

Thus I suggest that when the internal resistance is taken into account the battery pack cannot deliver the 691 hk to the two motors.

Torben_E

 

[Zetopan]

That being said, if I were designing one, I would use a much lower resistance value to lower I2R losses in use then use my active I monitoring to integrate energy and blow the fuse according to a look up table (Energy v. Time). Why keep a thinner piece of copper with losses when you can have a thicker piece with less loss and a much wider operation/blow range set in software?

The original fuse in the Tesla was a standard passive fuse that relied on I^2*R heating to blow. The new Tesla "smart fuse" operates essentially as you described. Of course I don't know if it uses a lookup table or if the I^2*T can be dynamically changed via software, but ignoring that aspect the new fuse is essentially what you have described.

 

[AWDtsla]

Just a technical note.

I don't think you, like many others, understand how voltage drop works in the battery. Under a 1500A load the voltage of the pack as a whole is going to plummet due to the internal resistance of the cells. It's not "slightly lower under load," it's MUCH lower under this load, nearly 100V lower by my calculations.

I would redo this with kW as the measure, but we all know 1 HP = 746W, so here is the graph I made that takes into account voltage drop of the battery pack:

First thing you'll notice is that this curve is not V*A linear. You'll see that at a full charge, under 1500A load, the voltage drop is going to cause the power output to be around 600 HP max, or around 450kW. You'll never see 691 HP out of the pack because that would require amperage beyond what the cell level fuses can handle, literally destroying the entire pack as a whole almost instantly. Once one pops it increases the load on the other 74 in the group and a chain reaction would ensue in every group until they were all popped.

This is data from the 85 kWh pack cells, so the 90 kWh pack would be about 5% better, in theory. They may also have beefed up the cell level fuses in the 90 pack as well, I don't know. However the 1500A cap puts the power cap around 600HP/450kW from the battery, and I'll say +/- 5% to account for potential differences in the 90 kWh pack cell chemistry.

I should probably do up a technical thread on battery voltage drop.

This graph doesn't line up with Tesla's claims that the lowest cell resistance is NOT at 100% SOC, along with member's results. Voltage sag should be higher and power output should be lower at 100% SOC.

- - - Updated - - -

Even if they did do as I suspect, I can not imagine that fuse ratings would be anything other than established as a one time programmable function at manufacture. The cost and complexity of having the battery's fuses sitting on the CAN bus allowing for field updating would seem prohibitive.

New fuse is microprocessor controlled anyway. I don't see a company like tesla making anything in the car that has software not be remotely programmable.

 

[kennybobby]

And don't forget the #29 AWG aluminum bond wires on each cell that add a resistance of ~7 mOhms per cell, so that at 1300 A they would drop 11.6 Volts off of the pack voltage.

It's unbelievable to me that those tiny bond wires can survive 17.5 A (1300A pack), but evidently they do since there have been no reports of packs replaced for opened cell fuses. And maybe they went to #28 bond wires to increase margin at the cells for 1500 A.

Why do you suppose that they would want to go to an active main fuse and add the risk of a false positive that fires the pyro and disables the car and probably damages the contactors in the process? i'm thinking it might be a pyro wire cutter with inconel anvils such as is used in aerospace launch vehicles. oops getting off topic...

 

[wk057]

This graph doesn't line up with Tesla's claims that the lowest cell resistance is NOT at 100% SOC, along with member's results. Voltage sag should be higher and power output should be lower at 100% SOC.

Using the loose cells I have I was unable to verify Tesla's claim that 100% SoC isn't the best. The internal resistance is virtually unchanged through the entire SoC range. At most it moves +/-0.5% in either direction vs average, low enough to simply ignore. The graph I made is based on data from the actual cells from the 85 kWh pack.

 

[Matias]

This graph doesn't line up with Tesla's claims that the lowest cell resistance is NOT at 100% SOC...

Can you please provide a source for that claim.

 

[AWDtsla]

Can you please provide a source for that claim.

I think it was Elon video, can't find it now. Claimed best performance was at ~90% SOC

- - - Updated - - -

The internal resistance is virtually unchanged through the entire SoC range. At most it moves +/-0.5% in either direction vs average, low enough to simply ignore. The graph I made is based on data from the actual cells from the 85 kWh pack.

Under load?

 

[stopcrazypp]

I think it was Elon video, can't find it now. Claimed best performance was at ~90% SOC

Might be Straubel more likely. I remember someone posting about an engineer saying that, not from a video. Might be able to dig it up when I have free time.

 

[wk057]

Can you please provide a source for that claim.

I think it was Elon video, can't find it now. Claimed best performance was at ~90% SOC

Might be Straubel more likely. I remember someone posting about an engineer saying that, not from a video. Might be able to dig it up when I have free time.

It was Elon Musk during some Q&A call if I recall correctly. Might even have been the "Three Dog Day" call. Not sure, but I did hear the claim that around 90% SoC was optimal.

I was unable to replicate this though.

Under load?

Yes, I've tested the cells at loads as high as 8C (equivalent to a 2000A draw at the pack level). As I said, the internal resistance is pretty constant across the SoC range, regardless of load. Perhaps this is different with the new cells in the 90 kWh pack, but I don't know because I don't have any of those to torture... er, test.

 

[JRP3]

It makes sense, since I can't think of any changes in the cell related to SOC that would change the effective internal resistance.

 

[Zetopan]

And don't forget the #29 AWG aluminum bond wires on each cell that add a resistance of ~7 mOhms per cell, so that at 1300 A they would drop 11.6 Volts off of the pack voltage.

Could you please post pointers to where you are getting your numbers from? I am working relatively blind here since I have not seen some of the numbers that are being quoted and I would like access to the relevant Tesla data. For example, I know that each cell is separately fused but I have not seen any Tesla data about the bond wire details, and likewise for the internal resistance of the cells vs their state of charge and temperature. Thank you.

 

[lolachampcar]

search on wk posts using the battery for home storage....

 

[arg]

Yes, I've tested the cells at loads as high as 8C (equivalent to a 2000A draw at the pack level). As I said, the internal resistance is pretty constant across the SoC range, regardless of load. Perhaps this is different with the new cells in the 90 kWh pack, but I don't know because I don't have any of those to torture... er, test.

Maybe there's another factor reducing the performance above 90% compared to the theoretical limit - ie. they aren't prepared to draw the full 1300A above 90%.

One possibility is that the temperature rise from a 1300A discharge would be considered unacceptable at 100% charge (admittedly it wouldn't still be at 100% afterwards). Or something around this - as I understand it, the cells need to be warm to hit the numbers at 90%, maybe it won't let the cells get that warm at 100%.

 

[Matias]

Could you please post pointers to where you are getting your numbers from? I am working relatively blind here since I have not seen some of the numbers that are being quoted and I would like access to the relevant Tesla data. For example, I know that each cell is separately fused but I have not seen any Tesla data about the bond wire details, and likewise for the internal resistance of the cells vs their state of charge and temperature. Thank you.

Plan: Off grid solar with a Model S battery pack at the heart