Tuesday, April 12, 2011

The Graphs is Always Greener....

This week we return to our 1957 Porsche Speedster Replica, the electric rebuild we call REDUX as it is the second version of the same car USING the same car. Recall we had upgraded this vehicle to 57 CALB 180 Ah prismatic LiFePo4 cells, a new version of Netgain's Warp 9 series DC drive motor, and a new to the market Soliton1 Controller by EVnetics.

After a broad misadventure with our first Soliton1 installation, Sebastien Bourgouis came to examine the problem and replaced the controller with a new one. It indeed seemed stronger, first blowing a Stage II Kennedy clutch and on replacement of that, blowing a 400A fuse we had forgotten entirely.

To find out just how much was gained in a numeric sense, we returned to Slingblade Performance in Anna Illinois for another round of dynomometer testing on their Dynojet dynomometer.





The results were interesting from a number of angles. Though young, the Soliton1 proves notably easy to configure, and now confirms a muscular presentation in the arena of the spinning of the tires. We hit about 156 peak horsepower at around 3200 rpm, and a clutch tearing 277 lbs of torque in the lower regions of the rpm band.

We had developed a bit of mistrust of the Soliton1's data log. It was clearly reading 150 rpm off on our test bench, and as it turns out, it's initial difficulties spreading the mustard were also entirely a measurement issue - not ours - but rather internally. It thought it WAS doing 1000 amps when our other best indicators were showing something in the 720 to 750 amp range. That's a noticeable variance that needed to be investigated and adressed.

As it turns out, the measurement itself WAS the problem. But the new Soliton1 that Mr. Bourgois so graciously not only provided, but personally delivered, measured amperes with good attention to detail. All comparisons with the other test equipment we had available looked essentially identical.

Surprisingly, the RPM problem went away as well. We assume the same calibration procedural issues had caused the RPM variance as well. And at that point, the Soliton1 log file function became seriously more interesting. Once we had some confidence the values recorded were real world, the value of this data logging capability quickly became apparent.

Essentially, the major gain was the ability to correlate voltage and current information with the Dynojet data files's horsepower information using RPM as the link as both machines record RPM.

This allows us a heroically finer level of graphing detail, as we can graph each mph of the run instead of every 10 miles per hour, and better we can correlate motor amps, battery volts, and more at this order of magnitude more frequent rate.



The leftmost column of the Soliton1 log is simply a timestamp and it indicates that the Soliton1 confesses its' secrets each 20 milliseconds. This is a useful thing, we actually then have a clock on our car runs.

The first column to the right, let's call it column two, is somewhat less useful. It features the percentage of available CPU cycles that are being used by the program. We have to assume the EVnetics software guru cares.

Column three indicates current in amperes. But it is't real. This is actually the level of current, after your throttle calibration, deadzone, and mapping are all taken into account, that the Soliton1 thinks is commanded by the accelerator input. Picture this as a request for power.

Column four provides a readout of actual motor current in amperes. This value, an important measure of the power in the system, is not nearly as controversial as many of our viewers believe. And I guess I'll have to explain that in more detail below.

Column five lists the duty cycle as a percentage. If you assume the entire period of time for one cycle at 8 khz, performance mode, or 14khz in quiet mode, is 100%, then duty cycle is that portion where the output waveform is positive.

Column six is the voltage of the battery pack. This will vary somewhat with load and indeed we saw the pack voltage sag as much as 22% during our runs.

Column seven is the internal temperature of the Soliton1 in degrees Centigrade. Almost all controllers monitor their internal temperature and when it reaches some higher value, the controller normally LIMITS the output current in an effort to reduce temperature.

Column eight is Revolutions Per Minute or RPM. This requires an external input from a magnetic or optical pickup reading the motor shaft rotation. This is important for our purposes because it allows us to correlate these Soliton1 values with the Dynojet values by finding the same RPM.

Columns 9-11 show the input voltage to IN1 through IN3, the general inputs to the Soliton1 that you can use for various things such as connecting the motor temp switch, etc.

Column 12 is the voltage of your 12v auxiliary system.

And finally the Soliton1 records a STATUS message indicating if it is running normally, is limited by slew rate, is limited by temperature, etc.

This wealth and granularity of information allowed us to graph in a resolution we've not been able to achieve before. We can take the horsepower output at each MPH increment and then use the RPM recorded by the Dynojet to lookup the same moment in time in the Soliton1 log. At that point, we can note battery voltage, motor amperes, temperature, and much more.

We published the graphs and put them on camera. The video wasn't up a couple of hours before we received a note from Steve West of New Zealand that we had it all hosed up.

Indeed, we had used motor amps and battery pack voltage to calculate kilowatts - kind of mixing apples and oranges. And of course we used kilowatts to calculate eHP by dividing by 746. And we used eHP to calculate efficiency by comparing it to the HP output on the Dynojet. In this way, we can compare how much power we are putting into the system from the battery pack, to how much power we are getting OUT on they Dynojet to see our losses.

And this brings us right back to the motor current vs battery current that many of you believe is so very important. For the purposes of determining peak power output of the controller OR the entire drive train, it really really isn't. But for this last calculation I threw in, the difference is significant.

The relationship between the two currents is largely a function of duty cycle. And it is quite proportional.

So to put out 500 motor amps with a 20% duty cycle, we need but 100 battery amps. Battery amperes will never exceed motor amps, as at 100% duty cycle the battery is basically connected to the motor. But motor amps can quite exceed battery amps.

The corrected graphs are depicted below. This grandly changes the calculation of KW. More importantly, if perhaps less grandly, it also changes our efficiency curves.

But while doing it, I might as well graph duty cycle and motor and battery amps separately. This leads to a bit of a busy graph at this point, probably beyond the level of useful. But it does show the relationship between battery and motor amperes.

We originally simply wanted to measure battery current to find our peak power output. A whole little army of people jumped up to try to "splain me" the relationship. I was quite familiar with it. The point was it doesn't matter. If you stomp on the accelerator pedal, they are one and the same very quickly, and we will certainly see full power, and full current, out of any controller very soon. And we were not getting it originally. We were getting something like 720 amps.

You will note that the motor and battery current are quite different right up to the point of peak power, where the duty cycle first reaches 100%. From there on, they are precisely the same. That point is also our peak power point as RPM also builds with duty cycle and current. From that point on, we are starting to be reduced in current and power by the counter electromotive force generated by the motor itself. Notably, you can move this point further right along the RPM scale by INCREASING your pack voltage. And of course, you move it left and down by REDUCING your pack voltage.










There are a couple of things that jump out here. Most of you do NOT have a dynomometer and I would find it astounding if anyone went to the trouble we did to manually extract several hundred data point values out of the Soliton log and type them into an excel graph. Assuming your controller is working properly ITSELF, you really can accelerate your car at maximum speed, and read battery amps and voltage to get a pretty good idea of how much power you are applying. Volts x Amps = kilowatts/746 eq eHP. Multiply that by 85% and that will be pretty close to what you will get on a dynomometer.

Without having a "go by" it is VERY difficult to tell if you are getting full power seat of the pants without such calculation. One of the reasons we have spent so much effort on this is that we have found NUMEROUS instances of people with perfectly operational electric vehicles, that ran WELL. But they were experiencing a FRACTION of the power available in their system - which was artificially limiting their power output.

This one even got US as it was an unusual situation where the controller itself was at fault. The most common problems in achieving full power, in our experience:

1. Battery voltage limits. Note that our pack voltage sags as much as 22% during a full acceleration. Most controllers have a well intended and largely useless function to limit current to protect a low battery pack voltage situation. This is intended as a poor man's "limp mode" and it almost never performs any useful function. But assume we had set this value to 160v. With a 192v pack this would seem logical. The controller would increase power until the pack sagged to 160v and at that point it would limit current. It is microprocessor controlled and can do this at a very fast rate. The result is such a smooth limitation of current, that it will hold your pack voltage at almost EXACTLY 160v, and vary the output current to do so. The result would have been MUCH less power than we see here in sagging to 147.

2. Temperature Limits. This is another very tricky one. First, again the controller can limit current in a very timely fashion to maintain it's temperature limit. You will notice that temperature rises to the limit, but never really exceeds it. This is because the controller is making constant adjustments to output current to maintain this temperature. Again, you won't really know you are current limited. Further, temperature in a controller is really quite a tricky thing. At these high current levels, it can change DRAMATICALLY and INSTANTLY. And so you will see a temperature spike ONLY when requiring maximum power. Let off the accelerator and temperature falls quickly. You won't even know this is happening.

Indeed, in our fourth gear run we can see that late in the run we WERE actually temp limited. Fortunately, it was so late in the run that we were not making much current anyway so it had little effect. But understand, we have a Toyota Prius pump squirting 20 liters per minute of glycol through the Soliton1 and thence through a pretty good sized heat exchanger. True, the car wasn't actually moving and so there was little air flow, but the point is that we intentionally OVERBUILD our cooling systems.

One of the myths of EV's is that they do not have the failures of components like ICE cars. Unfortunately, they have regular component failures and the most common is the controller. You can not only achieve maximum power, but GROSSLY extend the life of your controller if you keep it cool. There is no heat sink to large. We have developed a profound preference for liquid cooling of all controllers. And in the event of a controller such as the Curtis 1238, we actually manufactured a liquid chill plate to mount it to so we could have liquid cooling. And we entirely avoid the cheesy little kits most of the controller guys suggest. We use automotive pumps of at least 20 liters per minute, expansion tank fillers, high efficiency exchangers and quality hoses and fittings. It is always a thousand dollar expense. But it avoids power limitations and more importantly extends the life of the controller.

Finally, temperature is local. There is rarely a lot of heat energy to deal with, but it is very localized. The "case" temperature of your controller isn't the issue. It is at the IGBT itself. It drops about a volt and a half and at 1000 amps that's 1500 watts. Good internal design is important, but you cannot overcool this device. In fact, we suspect if you could cool it to about -400F, the volt and a half might go away.

3. Throttle input. The controller's primary job is to convert your throttle input to motor control. If you have the maximum throttle input set at 5v, but your throttle really can only provide 2.8v full pedal, you're going to be disappointed in your car. The car is fine. The configuration of throttle input is not fine. The controller basically maps the available power across this throttle input range. And 2.8 v is a little past half power. If you set the throttle max value in the controller to 2.7, then you will get full power BEFORE reaching the end of the pedal. Some controllers have byzantine configurations for this with different throttle TYPES and a MAP to alter the curve of applied power across the throttle input. You must become familiar with those configuration items.

Those appear to be the three most common power limiting mistakes. This has been the only event we've ever had where the controller itself was a problem. But if you have one of the Soliton1's, you might want to check. If they do not have now, they will undoubtedly have soon a procedure for you to follow to determine if you have a calibration issue I'm sure. But you might check with EVnetics to see what it is and check it. Ours sure got better when it was replaced. But again, the pernicious thing about this problem was the controller operated PERFECTLY in all respects and the car drove VERY NICELY and without actually measuring the current, we would have NEVER known how much power we were missing, if we had not measured.

Jack Rickard

19 comments:

  1. Jack,

    When discussing the Vicor Megapac, you mentioned that you would probably put a diode in front of it when connecting to your battery pack so that current would not flow from the Megapac capacitors back to your battery / controller. (That was the solution for your air conditioner compressor issue.) Wouldn't the full wave bridge rectifier that is part of the input section of the Vicor Megapac already give you the "diode isolation" from the battery pack? (Just a thought as I watched the show. Nice show by the way. And thank you for the oscilloscope output at the dyno!)

    Steve

    ReplyDelete
  2. Jack what happens when you apply full power when the pack near the end of its discharge?

    ReplyDelete
  3. Jack,
    As always, GREAT JOB !!! Well done.

    Thank you.
    -Youri.:-)

    ReplyDelete
  4. -400F :) Almost laughed out loud. I don't even want to calculate how much battery pack you would need to do that.

    I was very close to getting a sweet deal on a used Warp11 and an Elcon PFC3k charger last week but missed out.

    Excellent show gents, keep it up.

    ReplyDelete
  5. Jack - a question - in the graphs, are the battery current curves calculated from duty cycle and motor current, or were they measured directly? (I'm not disputing your explanation - just interested)

    John Hardy

    ReplyDelete
  6. The last graph should be VERY interesting to those of us in New Zealand. The peak power in both 2nd and 3rd gears falls very close to 50 kph and 100 kph, which are our 2 most common speed limits (urban and highway). 70kph is probably the next most common but this falls into a bit of a "hole" in the graph. I guess 3rd gear would probably be the choice to avoid motor overspeeding and get on the 'rising' side of the curve.

    I checked on Evalbum for the Duh gear ratio's but I can only spot the final drive at 3.44:1. Could you advise what the other gear ratios are (I'm sure you told us once, or even twice before)?

    I will need to re-watch the last show to recap what you said regarding how the graphs will alter with nominal pack voltage and current capability, but these graphs give some pretty awesome information regarding where to 'aim' for with gear ratios on a similar (soliton + 9") build.

    ReplyDelete
  7. John Hardy,

    I think this is what your looking for. If I am not mistaken the gearing is the same for both Speedsters. This is for Duh.

    1st 2.64
    2nd 1.93
    3rd 1.14
    4th 0.93
    Ring & Pinion 3.44

    Welded synkros heavy duty side plates and I think a few other goodies.

    Pete :)

    ReplyDelete
  8. Many thanks Pete - the gear ratio question was Nicks howevere - I was asking about the derivation of battery current - measured or calculated? I guess that is one for Jack

    ReplyDelete
  9. Yes, the battery amps are derived from Soliton's reported motor amps and duty cycle. If you look at the video carefully, you can see the relationship though as we have a meter on both. I think I named them backwards (when don't I).

    But for the graph, they are derivative.

    Jack

    ReplyDelete
  10. Padraic:

    The voltage sag will be more pronounced toward the end of charge. Beyond that, nothing.

    Jack RIckard

    ReplyDelete
  11. "...Yes, the battery amps are derived from Soliton's reported motor amps and duty cycle...."

    Many thanks Jack

    BTW - from the timestamp on your post, I guess you must start kinda early in Missouri

    ReplyDelete
  12. Jack said: "The voltage sag will be more pronounced toward the end of charge. Beyond that, nothing."

    Can this be measured? Will there be a significant drop off if one cell goes into reversal?

    Padraic

    ReplyDelete
  13. Hey Jack!

    Now that's a data dump!

    Nice to see you're straight on PWM volts and amps in and out. You had me worried there for a while.

    I wonder what kind of tractor pull you'd have to set up to verify the 1000 Amps CONTINUOUS claim of the Soliton 1. Maybe a tarmac event during the convention? How about the Escalade vs. any three?

    1500W at 1000 Amps (IGBT), that isn't even enough heat to keep the windshield clear on a cold day.

    With 80% efficiency at 140 kW, 28 kW is being lost to heat, 1.5 kW at the IGBT. That leaves 26.5 kW lost in the motor, transaxle and tires. Any of them feel warm afterward?

    Next time you should open your text file with your spreadsheet program (Excel, or whatever). There's no need to manually input hundreds of data points when your text file is that nicely formatted.

    Mike

    ReplyDelete
  14. Mike,
    80% seems pretty good in these conditions. Most front engined, rear wheel drive cars expect greater losses in transmission to road alone! Yet here the motor is the source of most of the losses - and that excels! It weighs in at 165lbs, mainly steels and copper with a ~3,500rpm fan inside. Takes a lot of heating and gives a lot of blowing.

    Must add with a smile. The voltage drop must have George Hamstra breathing a sign of relief. Imagine if the pack gave 191V all the way under load!

    More kudos to Evnetics for having the forethought to be able to use such a high voltage. The smaller controllers I've been finding for the Lemco/Agni motor are very limited in upper voltage. :(

    Once again, another top blog post by Jack & his fine crew :)

    Now, my knee's are killing me. Shouldn't crawl.

    Andy

    ReplyDelete
  15. Mike/Andy.

    The motor case at the end of a hard run like this is about 97F and the commutator looks like 115F. The motor is hardly breathing hard.

    Second, that IS what 1000 amps continuous looks like. If we actually did 1000 amps continuous over time, we would soon be going 726 mph.

    Finally, getting the data into Excel is not the problem. I have 20 entries in the Soliton log for every one in the Winpep7 data. And they have to be manually interpolated. Undoubtedly I could write a program to do this, but it would take more time than just entering it manually, and how often am I going to do that? So I wind up manually selecting each entry based on RPM, and copying over the temperature, duty cycle, motor amps, and voltage. It is extremely tedious. But there are no "exact' matchups on rpm. They must be interpolated.

    But the results are pretty, and quite informative to me. It's not just the Soliton. I've never been able to find any persuasive data on the Warp9 at these power levels.

    One of the things we are thinking about is expanding our test bench voltage one battery at a time until we GET arcing on the Warp11's with the new brushes, at say 100 amps, to find out just what the voltage is where it becomes a problem. Helwig claims 192 volts will work. We had that pack voltage nearly, but never approached that under current. So my thinking is run the bench at 100 amps and keep increasing the voltage until it sparks. Then decrease by one battery - 12v. And we'll know the actual point where arcing begins - at least generally.

    Jack

    ReplyDelete
  16. Hey Jack!

    The back EMF at 726 mph, even in 4th gear...

    Thus, the TRACTOR PULL! All the power, none of the speed. The smart money is still on the Escalade.

    The persuasive motor data I'd like to see would look like your torque graph (only motor output torque), and plot steady state torque vs. RPM, with horizontal lines that represent constant motor amps (isoamps) and top-left to bottom-right curves that represent constant motor volts (isovolts). With this, you could predict your motor/controller/battery performance by following your controller current limit isoamp until it intersects with your battery pack (sagged) isovolt. You could play with what-if combinations, and simulate (you'd need lots of other info like weight, frontal area, drag coef, trans efficiency, tire diameter, ..., but most of that is easy to measure or is published). Or, you could just bolt in different combinations and trot them out to the dyno to see how they work, which is admittedly much more entertaining.

    But if I'm trying to decide what components to buy, and trying to get the combination right the first time, I'd prefer to do the trade-offs on "paper" (boring, but cheap and fast). The "persuasive motor data" is the missing link.

    Mike

    ReplyDelete
  17. Mike:

    Well, when you get those done, be sure and share em with us.

    Everyone has something different they think is important. I like to see what happens in the wild where the rubber meets the dynomometer drum. If you want a purer form of laboratory data, that's certainly understandable. Like I said, when you get it done, share.

    Jack Rickard

    ReplyDelete
  18. Jack,
    Anything can do 726mph and more providing air is not an issue. But air is great at cooling things. Look at the size of the cooling fins they needed on spaceships like you see on Babylon5 and Avatar!
    I digress.

    On load/velocity/brushes... Cedric Lynch's funny bike thing is streamlined to an amazing ~30WH/mile.
    A must see. Don't laugh, he's English like me..

    http://dai.ly/9VdPNX
    ================================

    Talking commutators, am I not right in saying its the Volts which make the sparks?
    ================================

    Just been looking at the Redux's graph for 2nd gear. I'm sure that car can do low 6's. Just need to allow it to rev out to 60. Changing gear loses a surprising amount of time. The end of rev's will still around 70hp.

    ReplyDelete
  19. Since nobody else mentioned it, I'll bite. That makes two April 1, 2011 titled shows...are we going for a third this week? Haha, just kidding. Was that a belated April fools joke?

    Michael

    ReplyDelete