DIY Electric Car Forums banner
301 - 320 of 490 Posts

·
Registered
Joined
·
780 Posts
I sent an email to the owner of the shop that modifies Curtis controllers. Unfortunately, he says he doesn't have any experience with the Curtis AC models and he thinks it would be difficult to increase their output. :(
Doh! I wonder if he's thinking it's 3 times the work (3-phase) or because there are other things like DC/DC on each phase, etc to change.

You can buy a complete 300V, 400A IGBT power module, driver board, caps and current sensors and vector control board for about $1K and an industrial VFD for $300. You can use the AC-50 motor with that and get 80kW input power (at 200V max for AC50). Not much different than the 78kW of the HPEV package and that's why I'm hesitating with a DIY controller. That Curtis controller is a bargain for what you get. When they up the voltage to 144 nominal it will be a killer AC system.

JR

PS: oh yeah, forgot about the 6 months worth of coding and tinkering to get the industrial controller to work nice in a car instead of driving a bandsaw.
 

·
Registered
Joined
·
2,318 Posts
I did some tests to determine base speed (manual page 62), and check at what rpm Modulation_Depth reaches 100% and Current_RMS starts decreasing from the max 550A when accelerating with the throttle “floored”. I also recorded Frequency and Motor_RPM, at controller frequency at 5000 rpm. I used the Curtis PC software and a laptop to view parameters.
For these tests I set things up for maximum acceleration:
Max_Speed = 8000 rpm (max value for motor rpm)
Accel_Rate = 0.1 (min value)
Gear_Soften = 0 (min value)
Drive_Current Limit = 100% (max value)
Typical_Max_Speed = 6000 rpm (max is 8k, but I think this only has an effect in speed control mode)
Base_Speed = 6000 (set to its max for the base speed test described on manual pg 62, this was factory set to 3200)

AC50 motor/1238-7501 controller running in torque control mode, 36 180Ah CALB cells, full charge at start of test. Suzuki Swift, 2250 lb, http://www.evalbum.com/3060

Results

The Base_Speed test was run twice and values of 4517 and 4663 rpm were obtained. I think I may have let up the throttle a bit on the second test, so have more confidence in the first number. Another estimate of base speed is to directly monitor when the controller is using full battery voltage, i.e. when the Modulation_Depth parameter reaches 100%, or equivalently, when the motor current, Current_RMS parameter, starts to decrease from its max of 550A rms, when accelerating at full throttle (I would guess the controller does something like this in the test).

It is unfortunate that monitored values cannot be data logged and/or that the software does not permit viewing of multiple windows simultaneously. It requires 5 or 6 clicks of the UP arrow key to move from the Modulation_Depth parameter to view the Motor_RPM parameter value. This takes far too much time to give even a rough estimate of the rpm at which Modulation_Depth reaches 100% (controller using full battery voltage) when accelerating at full throttle. So to get a rough estimate, I timed how long it took to reach 100% Modulation_Depth (run1), and how long it took for the rms current to start to decrease (run2), then checked the rpm after this elapsed time (run3). The first two runs gave the same result within measurement error as expected, about 5 seconds. The motor rpm at 5 seconds monitored in the third run was about 5200 (approx, it was increasing fast). A 0.6 second, or 12%, overestimate of time, which I easily could have done, would explain most of the difference between this value and the results in the Base_Speed test. After the test I entered a base speed value of 4550.

The controller frequency at about 5200 rpm (+/-100 roughly) was about 174 Hz (wasn’t possible to hold them constant). It agrees well with the speed as a function of frequency for a 4 pole motor: rpm = 120*f/4 = 5220. So at max speed of 8k rpm should be reached at f = 267 Hz. In an earlier test while accelerating with it floored I saw about 280 Hz, so it must have overshot a bit.

I also did some tests monitoring motor and battery current. At lower motor rpm motor current is greater than battery current and voltage modulation depth is small. At 50% modulation depth motor current is about twice battery current, and at 100% modulation depth (I extrapolated) they are about equal (controller rms current ~battery current), as expected. I only have a 500A shunt so could not directly compare rms current and battery current at 100% Modulation depth. I just extrapolated from several observations at lower than 100% values.So it appears controller output power about equals controller input power minus losses. Time to give the little transmission a rest.
 

·
Registered
Joined
·
8,126 Posts
Discussion Starter · #306 ·
Great data as always Tom. You mention that the software does not allow multiple windows, does a single window display more than one parameter at a time? For example the hand held programmer will display a few different values of a sub menu at once, say motor temp, current, voltage, rpms, or something like that as I remember.
 

·
Registered
Joined
·
780 Posts
Hi. Agreed, good report.

Does anyone have the protocol detail for Curtis controllers of this type? Or at least a captured stream of data from the serial port? It should be possible to write a program to capture and display those parameters at once.

JR
 

·
Registered
Joined
·
4,064 Posts
hmmm, might be fun to set up some virtual ports on a computer and then "peak" at them with a serial data analyzer.

I know the HPEVS don't stream, they're set to work with the dumb spyglass, so they only spit out one variable at a time.
 

·
Registered
Joined
·
2,318 Posts
You mention that the software does not allow multiple windows, does a single window display more than one parameter at a time? For example the hand held programmer will display a few different values of a sub menu at once, say motor temp, current, voltage, rpms, or something like that as I remember.
The motor parameters are all displayed in real time simultaneously in one window, but the controller parameters are not, the value of only one is displayed at a time. And the menu lists Motor as a major heading, then each of its parameters is in an indented list under that heading. Same thing for the controller, and they both use the same window. So you have to select say, Modulation Depth, to view the value of that controller parameter in the window. Then to see a motor parameter such as rpm, you must scroll up the menu through the controller items listed above Modulation Depth, then through the listed motor parameters until you come to RPM, or scroll through them to the major heading Motor, and it will then display all of the motor parameters at once. It is really nice to be able to see all those parameters and watch them as you drive, but it would have been nicer to view both controller and motor parameters at the same time, or even more than one controller parameter at the same time. Datalogging would be even better.
 

·
Registered
Joined
·
2,318 Posts
I think it is interesting that the base speed came out higher than what you would predict from using the rpm for peak power on the Curtis dyno data, which I estimated at about 3100 rpm, then multiplying it by 1 plus the fractional increase in pack voltage between my (sagged) pack and the one used in the dyno test: 3100*[1*(110-86)/86] = 3965. I don't understand why. The test result was dependent on how fast I accelerated, and the road was slightly uphill, so maybe level ground would give a result closer to 4k.
 

·
Registered
Joined
·
2,318 Posts
Someone recently contacted me regarding performance of the AC50 motor. In an effort to quantify it a bit more I put together a graph comparing the available wheel torque in my car with it compared to an Advanced DC FB1-4001 9" series DC motor, both with 115V pack. It occurred to me it might be useful to others to compare the AC50 to an 8" ADC motor as well as the 9" ADC, and also to the AC31. So I put together charts for them below.

Few comments on the charts:
1) They are all for my Suzuki Swift (http://www.evalbum.com/3060)
which weighs a bit under 2300 lb. Gear ratios are given in the evalbum. Tire diameter (loaded) is about 20.75".

2) All are for a nominal 115V pack

3) The curve at the bottom in all the graphs is the "required wheel torque". This is the wheel torque required to move the vehicle at a constant speed. "Available wheel torque" is the maximum wheel torque obtainable at a given speed if you "floor it". The flat part of these latter curves are determined by the maximum controller current which is 550A for the Curtis controllers for the AC50 (rms) and ADC 8" motors, and 500A for the ADC 9" motor (FB1-4001). The decreasing part of these curves are determined from a second order fit to the torque-speed (motor rpm) curves for the ADC 8" and 9" motors, and the Curtis dyno data for the AC50 (adjusted for 115V since the original data is for 96V). Several torque-speed curves were used for each of the ADC motors to get a fit over as wide of pack voltage range as possible, but of course data doesn't exist on these for very low voltages or very low rpm so the curves are extrapolated there and subject to more error. They seem fairly reasonable though, at least for comparison purposes. All the fits were good, with R squared values over 0.98. The "bumps" at the knee of some of the curves are artifacts of the curve fitting routine.

4) The wheels apply a force, F = ma to the road to accelerate the car. The acceleration is a = F/m = T/rm, where T is the wheel torque, r the "loaded" tire radius, and m the vehicle mass. The wheel torque is the product of the overall gear ratio, transmission and rear end, and the motor torque.

5) Drive train efficiency is assumed to be 90% and is factored into the available wheel torque (available wheel torque = motor torque*dt efficiency*gear ratio). No accounting is made of inertia of rotating drive train parts (I figured the torque for it was negligible compared to the required torque to move the car). Edit: I discovered some time after posting this that actually a good bit of the torque in first gear goes to rotational inertia, affecting acceleration in that gear significantly, not as much in the higher gears.

6) The difference in available wheel torque and required wheel torque is the net wheel torque available to accelerate the vehicle, so comparing available wheel torque for different motor/controller combinations indicates relative available acceleration at a given vehicle speed.

So here is how I interpret the curves. The AC50 and AC31 are fairly close, with the AC31 having significantly higher torque at lower rpms, but the higher power (higher torque at higher rpms) of the AC50 gives a bit better acceleration to 60 mph, and higher top speed (defined as where the available and required torque curves intersect for a given gear).

The 9" ADC has higher torque than the AC50 at lower motor rpms (so lower vehicle speed in a given gear) but it doesn't extend out to as high of rpm or vehicle speed as the AC50. So it looks to me like the AC50 might do better. The AC50 clearly has more available torque and acceleration than the ADC 8" motor.

The AC31 and ADC 9" are fairly similar due to the lower base speed of the AC31 compared to the AC50.

As has been stated on this site many times "amps is torque" (or something like that:D), so if you use the ADC 9" motor with a 1000A controller the peak torque roughly doubles from what is show for it here with the 500A controller. Then we are in another league altogether.

Also, the "knee" of the available wheel torque curves can be extended to higher motor rpm and higher vehicle speed if a higher voltage pack is used to compensate the increasing motor input impedance at higher rpm, providing the controller can handle it. The knee moves roughly proportional to the fractional increase in voltage. The highest voltage torque-speed curve I could find for the 8" ADC was 132V. This is (132-115)/115 = 0.15 higher, so doesn't change the knee that much - about 15% higher motor rpm and hence vehicle speed. I don't know how high a voltage is safe to use with it. My understanding is that the Warp9 gives similar torque per amp as the ADC 9", but has a number of changes like larger brushes to make it more robust. Considerably higher voltages can be run with the 9" motors, moving the "knee" of the curves to higher speeds, and giving overall higher acceleration.
Edit: According to this post: http://www.diyelectriccar.com/forums/showpost.php?p=19647&postcount=2, and post #4 in the same thread, the max motor input voltage Jim recommends is 160V. This is is about 39% greater than 115V, so for example, it would move the "knee" of the curve for the 9" motor in second gear out to about 35 mph rather than 25 mph. That would be about the same as the knee of the second gear curve for the AC50, but with about 700 ft-lb wheel torque compared to about 590 for the AC50.

So keep in mind this is for the same 115V pack for all motor/controller combinations, and for 500A (9") or 550A (8", AC50) maximum controller current. The idea is that because there are many more ev's around with DC systems, there are more examples of performance for these. This gives some way to compare that performance to what you might expect from an AC50 or AC31 with similar pack V and max controller amps.

Edit: I thought I should add a comment on how I adjusted the AC50 data for 115V. The dyno data is for a 96V pack which sagged to 86V. My nominal 115V 180Ah CALB pack sags to about 110V when I floor it (3C current, would be more sag for lower Ah cells), so (110 - 86)/86 = 0.28. The "knee" of the torque curve in the dyno data is at about 3075 rpm, so I estimated the knee would occur at 3075*(1+0.28) = 3936 rpm, which I rounded up to 4k. I rounded up since when I performed the base speed test as described in the Curtis 1238-7501 manual I initially got values over 4.5k rpm - but there was a slight slope to the road. After repeating the test on level ground I obtained values of about 4400 and 4500 rpm. So I thought 4k was a reasonable value to use. I adjusted the AC31 knee to 1800*(1 + 0.28) = 2300 rpm based on the 96V dyno data for it. I used 90 ft-lb peak torque for the AC50 and 117 ft-lb for the AC31.
View attachment AC50, ADC 8 inch, ADC 9 inch, wheel torque.pdf
 

·
Banned
Joined
·
497 Posts
Someone recently contacted me regarding performance of the AC50 motor. In an effort to quantify it a bit more I put together a graph comparing the available wheel torque in my car with it compared to an Advanced DC FB1-4001 9" series DC motor, both with 115V pack. It occurred to me it might be useful to others to compare the AC50 to an 8" ADC motor as well as the 9" ADC, and also to the AC31. So I put together charts for them below.

Few comments on the charts:
1) They are all for my Suzuki Swift (http://www.evalbum.com/3060)
which weighs a bit under 2300 lb. Gear ratios are given in the evalbum. Tire diameter (loaded) is about 20.75".

2) All are for a nominal 115V pack



3) The curve at the bottom in all the graphs is the "required wheel torque". This is the wheel torque required to move the vehicle at a constant speed. "Available wheel torque" is the maximum wheel torque obtainable at a given speed if you "floor it". The flat part of these latter curves are determined by the maximum controller current which is 550A for the Curtis controllers for the AC50 (rms) and ADC 8" motors, and 500A for the ADC 9" motor (FB1-4001). The decreasing part of these curves are determined from a second order fit to the torque-speed (motor rpm) curves for the ADC 8" and 9" motors, and the Curtis dyno data for the AC50. Several torque-speed curves were used for each of the ADC motors to get a fit over as wide of pack voltage range as possible, but of course data doesn't exist on these for very low voltages or very low rpm so the curves are extrapolated there and subject to more error. They seem fairly reasonable though, at least for comparison purposes. All the fits were good, with R squared values over 0.98. The "bumps" at the knee of some of the curves are artifacts of the curve fitting routine.

4) The wheels apply a force, F = ma to the road to accelerate the car. The acceleration is a = F/m = T/rm, where T is the wheel torque, r the "loaded" tire radius, and m the vehicle mass. The wheel torque is the product of the overall gear ratio, transmission and rear end, and the motor torque.

5) Drive train efficiency is assumed to be 90% and is factored into the available wheel torque (available wheel torque = motor torque*dt efficiency*gear ratio). No accounting is made of inertia of rotating drive train parts (I figured the torque for it was negligible compared to the required torque to move the car).

6) The difference in available wheel torque and required wheel torque is the net wheel torque available to accelerate the vehicle, so comparing available wheel torque for different motor/controller combinations indicates relative available acceleration at a given vehicle speed.

So here is how I interpret the curves. The AC50 and AC31 are fairly close, with the AC31 having significantly higher torque at lower rpms, but the higher power (higher torque at higher rpms) of the AC50 gives a bit better acceleration to 60 mph, and higher top speed (defined as where the available and required torque curves intersect for a given gear).

The 9" ADC has higher torque than the AC50 at lower motor rpms (so lower vehicle speed in a given gear) but it doesn't extend out to as high of rpm or vehicle speed as the AC50. So it looks to me like the AC50 might do better. The AC50 clearly has more available torque and acceleration than the ADC 8" motor.

The AC31 and ADC 9" are fairly similar due to the lower base speed of the AC31 compared to the AC50.

As has been stated on this site many times "amps is torque" (or something like that:D), so if you use the ADC 9" motor with a 1000A controller the torque roughly doubles from what is show for it here with the 500A controller. Then we are in another league altogether.

Also, the "knee" of the available wheel torque curves can be extended to higher motor rpm and higher vehicle speed if a higher voltage pack is used to compensate the increasing motor input impedance at higher rpm, providing the controller can handle it. The knee moves roughly proportional to the fractional increase in voltage. The highest voltage torque-speed curve I could find for the 8" ADC was 132V. This is (132-115)/115 = 0.15 higher, so doesn't change the knee that much - about 15% higher motor rpm and hence vehicle speed. I don't know how high a voltage is safe to use with it. My understanding is that the Warp9 gives similar torque per amp as the ADC 9", but has a number of changes like larger brushes to make it more robust. Considerably higher voltages can be run with the 9" motors, moving the "knee" of the curves to higher speeds, and giving overall higher acceleration.

So keep in mind this is for the same 115V pack for all motor/controller combinations, and for 500A (9") or 550A (8", AC50) maximum controller current. The idea is that because there are many more ev's around with DC systems, there are more examples of performance for these. This gives some way to compare that performance to what you might expect from an AC50 or AC31 with similar pack V and max controller amps.
View attachment 10569
GOOD JOB MAN, this should clear up some questions and make it easier for those on the fence trying to decide what to do.
 

·
Banned
Joined
·
497 Posts
I don't have it in front of me right now but when going though the parameters there was an adjustment available for Throttle Frequency. I'll have to check the values later.
The throttle parameters that are programmable using a 2 or 3 wire pot is the starting point and the ending point of the 0-5v or 0-5k range. You can also program the torque curve by adjusting the knee point. I dont recall a adjustment for frequency. I you find a reference to that, point me in that direction.
 

·
Registered
Joined
·
8,126 Posts
Discussion Starter · #318 ·
Well I won't be finding anything for a while since I just fried a bunch of stuff :mad: I'm not sure exactly what caused it but I was charging the car, with the car off, and I plugged in the PC cable to the molex plug, with the computer also off, but plugged into the same circuit as the Manzanita charger, and sparks and smoke started coming out of the IOGear adapter. :eek: DOH! I quickly unplugged but the damage was done. The controller still seems to work fine but now when I plug in the 840 display all I get is the ******* symbols. So I fried my IOGear, fried one of the USB ports on my computer, and fried something in the communication circuit to the 840 display. I really should have put some fuses in the cable I made up. I'll get another IOGear and see if I can connect with a PC again but I'm not hopeful
 

·
Banned
Joined
·
497 Posts
Well I won't be finding anything for a while since I just fried a bunch of stuff :mad: I'm not sure exactly what caused it but I was charging the car, with the car off, and I plugged in the PC cable to the molex plug, with the computer also off, but plugged into the same circuit as the Manzanita charger, and sparks and smoke started coming out of the IOGear adapter. :eek: DOH! I quickly unplugged but the damage was done. The controller still seems to work fine but now when I plug in the 840 display all I get is the ******* symbols. So I fried my IOGear, fried one of the USB ports on my computer, and fried something in the communication circuit to the 840 display. I really should have put some fuses in the cable I made up. I'll get another IOGear and see if I can connect with a PC again but I'm not hopeful
OUCH! I did something like that as well with a Kelly controller. I plugged in the programming cable from computer while charging and took out all the RS232 stuff including the CPU ports. I believe it is the ground path that took it all out. Advice, when charging, dont connect anything. When this happened, you probably didnt have the spyglass connected as that is where the programming cable goes, so that should not be a problem. If the spyglass doesnt work, I would suspect that there is damage to the controller in the I/O circuit which should not effect the rest of the controller.
 

·
Registered
Joined
·
780 Posts
Yikes sorry to hear JRP. Was that a laptop you were using? If so, running plugged in to AC outlet with a three prong connector? Could it be that your charger is not isolated? Can you measure a potential between your charger's negative pole and your AC outlet ground leg?

If the Curtis' serial port is opto-isolated, it shouldn't be too hard to fix. Best of luck and keep us posted as you can.

JR
 
301 - 320 of 490 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top