Clutch - leave in or not?

Have you made or bought a converted vehicle if so this is for you
User avatar
EVguru
Posts: 305
Joined: Fri May 25, 2007 5:17 pm
Location: Luton
Contact:

Postby EVguru » Tue Aug 19, 2008 11:49 am

Out of curiosity what is considered high voltage?

Well, my definition for an EV starts at about 144 volt.

An awful lof of cars have been converted in the US and experience has shown that 96volt is the minimum to get adequate performance (with series wound motors) from something of VW Rabbit (MK1 Golf), Ford Escort, etc size/weight. Going to 120 volt starts to give performance more like an ICE and people pushed on to 156v nominal (matches quite well to 200v components; fully charged battery + spikes) first with controllers from Auburn Scientific and later from DCP. Apart from Otmar Ebenhoech who was making controllers for up to 336v, only Auburn Scientific produced successful 192 volt controllers, the DCP models proved rather unreliable.

On the racing Porsche we were eventually running 288 volt with a 9" ADC motor. This was an early motor and bore a rating plate stating 72 volt (exactly the same motor is now plated at 144 volt). Since 12 volt in second gear was enough to move the car at walking pace, it's quite a challenge for a controller to step 300+volt (24 time 12.7v charged batteries) down to less than 12v. A clutch was absolutely essential (and also just to get into gear with the awful Porsche transaxle and shift linkage).
Paul

http://www.compton.vispa.com/scirocco/
http://www.morini-mania.co.uk
http://www.compton.vispa.com/the_named

User avatar
qdos
Posts: 2089
Joined: Fri Apr 27, 2007 10:26 pm
Location: Dorset
Contact:

Postby qdos » Tue Aug 19, 2008 2:38 pm

I think the car refered to here in this thread that kicked off the topic though is a Cinquecento for commuting in outer London and isn't really a Racing Porsche. Though it may be the Rover I can't remember now.

Most of the folk here are generally looking for regular road use EVs.

User avatar
Jeremy
Posts: 472
Joined: Thu Jul 31, 2008 6:35 pm
Location: Salisbury

Postby Jeremy » Tue Aug 19, 2008 6:14 pm

I cheated slightly, in that the vehicle in question is not a pure EV, but a Toyota Prius hybrid. However all models of the Prius use pure EV to pull away, as the ICE is not ever directly mechanically coupled to the output shaft, so it's relevant to this discussion. In the Prius, the ICE is connected to the front wheels via the in-line differential device that Toyota have named the "power split device". This is always torque moderated by MG1 (the combined starter motor, battery charging generator and torque compensation/adjustment motor generator) which rotates backwards or forwards as required to allow the ICE to contribute torque to the output shaft at speeds above a standstill. MG2 drives the front wheels directly and is a 3 phase brushless motor running on around 500V in the current model Prius. Total electrical power available is around 60kW, so it's reasonably powerful in EV terms.

More importantly, the controller works well from a standstill with no clutch, operates at high voltage, doesn't produce high EMC levels and uses motor phase current monitoring (albeit on only two of the phases of each motor), showing conclusively that Toyota cracked this problem over ten years ago (the Prius was developed during the early nineties).

Jeremy

User avatar
EVguru
Posts: 305
Joined: Fri May 25, 2007 5:17 pm
Location: Luton
Contact:

Postby EVguru » Tue Aug 19, 2008 7:42 pm

If you remember Jeremy, we were discussing the problems of obtaining sufficiently low duty cycles to get good low speed control.

I assumed you had an incomplete knowledge of EMC becuase every article I've read on power switching design says stay away from hard switching if you can. I have friends who have built power converters into the 100Kw range and they've said the same. Why in this application would you make the EMC requirements harder to deal with, make the gate drive more expensive, and increase the voltage spikes from the parasitic inductance, when psudo-random pulse skipping will get the job done with no real downside?

When you mentioned One million EV's, I thought 'He must mean the Prius!?'.

I really couldn't beleive it though, it's totally irrelevent to this discussion.

The problem of delivering very low duty cycles simply doesn't exist in the Prius. Firstly the PWM frequency is fairly low at just 12KHz or 10KHz on the earlier model. Secondly, the motor is a Synchronous AC machine. It's speed is controlled by frequency, not voltage.

Perhaps you could stop accusing me of underestimating your knowledge, just because I will question anything that doesn't appear to make sense to me.

The "motor current control" approach needs an additional limiting control loop to prevent speed runaway, plus some time-bound filtering probably, to give a smooth feel, but that should be possible easily enough with a simple microcontroller.


The Brusa PM controllers (analogue) are current mode, but their highest voltage model is 48 volt. There is a 72 volt version, only available from ASMO as a 'Race Only' model. Cedric has long been a fan because it's very good for economy and gives very smooth control with his motor.The sudden increase in speed after climbing over an obstruction is only really a problem if you aren't expecting it, but it does need sorting. Both the Curtis and the Brusa are measuring peak FET current (Curtis with Vds-on, Brusa with a current mirror), rather than real battery or motor loop amps, so the measured figures don't quite match up with specs.

I've tracked down the throttle response modifier for the Alltrax;

http://www.maxdon.com/

I've been looking at pulse by pulse current measurement for a 15kHz PWM brush motor controller and it seems quite straightforward, although it's complicated by needing to account for the actual installed motor/wiring inductance, so would be more difficult to incorporate into a universal product. If designing a universal controller along these lines then using a simple shaft torque sensor might be a better approach, especially as the same sensor would automatically give motor shaft rpm.


I don't see why you'd need to know the motor and wiring inductance. I'd only normally think of getting into pulse by pulse measurment if trying to save money (like Curtis and Brusa) or of designing an independent overcurrent protection circuit for the power devices.

The average motor current is easily measured with a shunt or hall effect device and you don't really need to know what the motor torque is, so long as you can hold it at a given level with the throttle. It doesn't even matter if it's a series wound motor with a non linear current/torque curve. Drivers/rider are used to a non-linear throttle response!
Paul

http://www.compton.vispa.com/scirocco/
http://www.morini-mania.co.uk
http://www.compton.vispa.com/the_named

User avatar
Jeremy
Posts: 472
Joined: Thu Jul 31, 2008 6:35 pm
Location: Salisbury

Postby Jeremy » Tue Aug 19, 2008 8:13 pm

EVguru wrote:When you mentioned One million EV's, I thought 'He must mean the Prius!?'.

I really couldn't beleive it though, it's totally irrelevent to this discussion.


With respect, it's not, as the Prius is a pure EV when pulling away from a standstill, or reversing/slow speed manoeuvring, which is precisely when good low speed motor control is needed. We were discussing low speed control, weren't we?

The problem of delivering very low duty cycles simply doesn't exist in the Prius. Firstly the PWM frequency is fairly low at just 12KHz or 10KHz on the earlier model. Secondly, the motor is a Synchronous AC machine. It's speed is controlled by frequency, not voltage.


The problem still exists, as the Prius AC controller is still a switched mode device. It derives motor power control using a switched mode, three phase IGBT inverter, so is still subject to the same pulse width limits for low power. The fact that it then turns this into a variable frequency signal to drive the motor doesn't remove the need for sensitive power control.

12kHz is just about fast enough for pretty much any motor available to us, although it's perhaps on the edge with regard to ripple current for a very low inductance PM design, like one of Cedric's motors. For all practical purposes, using commonly available series wound motors, 12kHz should be OK, as I'm sure you know.



Perhaps you could stop accusing me of underestimating your knowledge, just because I will question anything that doesn't appear to make sense to me.


It's not the questioning, Paul, but the superior tone that causes annoyance. I'd encourage anyone to question anything, but I take exception to people assuming that I'm ignorant by default, without bothering to find out if their assumption is valid.


The Brusa PM controllers (analogue) are current mode, but their highest voltage model is 48 volt. There is a 72 volt version, only available from ASMO as a 'Race Only' model. Cedric has long been a fan because it's very good for economy and gives very smooth control with his motor.The sudden increase in speed after climbing over an obstruction is only really a problem if you aren't expecting it, but it does need sorting. Both the Curtis and the Brusa are measuring peak FET current (Curtis with Vds-on, Brusa with a current mirror), rather than real battery or motor loop amps, so the measured figures don't quite match up with specs.


I suppose battery current throttle control would be best for economy, but might well wind up other road users as your speed goes up and down like a yo yo...........



I don't see why you'd need to know the motor and wiring inductance. I'd only normally think of getting into pulse by pulse measurment if trying to save money (like Curtis and Brusa) or of designing an independent overcurrent protection circuit for the power devices.

The average motor current is easily measured with a shunt or hall effect device and you don't really need to know what the motor torque is, so long as you can hold it at a given level with the throttle. It doesn't even matter if it's a series wound motor with a non linear current/torque curve. Drivers/rider are used to a non-linear throttle response!


Just being a perfectionist. Average motor current is the result of switching device peak current and motor/wiring inductance which obviously smooths the current to a near continuous level between pulses, so I was looking to have an accurate measure of the true average current, measured near instantaneously, flowing through the motor windings. One issue with an indirect measurement is the rate of rise of motor current, and the subsequent speed of the control loop to arrest it, in the event of the motor suddenly being slowed right down by a heavy load.

It most probably isn't an issue, although it's interesting to note that Toyota opted to measure individual motor phase current at high speed for their controller, using independent sensors directly on the phase connections.

I wholly agree about driver/rider adaptation to highly non-linear throttle response. It surprised me when I was playing about with it as to just how little effect changing the control linearity had in practice. I've always believed that human beings are better closed loop devices than most engineers can build - this just helped confirm that view......................

Jeremy

Tom Thomson
Posts: 64
Joined: Fri Aug 31, 2007 12:48 am

clutch or no

Postby Tom Thomson » Tue Sep 09, 2008 1:25 am

In my conversion, I eliminated the clutch and have no problems up or down shifting. I do not, however, bash it from gear to gear rather I give the synchro rings time to do their thing. Porsche 924 5 speed with a 6.7" motor.
tommyt

User avatar
arnolde
Posts: 23
Joined: Mon Sep 08, 2008 9:11 am
Location: Germany

Postby arnolde » Tue Sep 09, 2008 10:59 am

Very interesting topic, I am just now at the stage of having a connector made to connect my motor to my gearbox (Opel Corsa), and I had not even considered leaving out the clutch. Drat, now another decision to make :roll:

I'm all in favour of the simplest and lightest solution... so I think I'll do it without at first, but leaving everything in original condition so I can put the clutch in later if I need to.

Alan Ward
Posts: 28
Joined: Wed Dec 13, 2006 3:31 pm
Location: mid Dorset

clutch - leave in or not

Postby Alan Ward » Sat Oct 11, 2008 9:49 pm

I have read with interest Christopher’s 9 July question : Clutch – leave it in or not . And I have read the many good comments and advice which followed. ChrisB described his Fiesta EVwith no clutch, which he kindly sold me, total now 20,000 miles. I can only add one strange detail. In this type of EV the electric motor/s connect by a toothed rubber belt to the driven pulley on the splined steel shaft where the clutch used to be. It is of course essential to have a new end plate and bearing to take the pull of the toothed belt. ChrisB had machined this accurately, and everything lined up perfectly. BUT the ball bearing assembly in this end plate suffered a strange problem. Chris had made a careful brass bush as a push fit into the inner of the ballbearing race, and initially this brass bush was a tight fit over the end of the splined steel shaft. After a few thousand miles of hilly driving, the splined shaft then started turning in the brass bush, even though the ballbearing race was perfect. I was puzzled, and replaced the original brass bush with a new one which was an even tighter push onto the splined shaft, and fitted a new ballbearing race ‘just in case’. But the same fault recurred. I tried several different metal glues, but after a few thousand miles the splined shaft started again turning in the brass bush. Totally inexplicable, as the ballbearing race turns perfectly freely. Finally, in desperation, I got an elaborate brass bush which fastened securely to the driven pulley itself , and this has been successful. Alan Ward

User avatar
Jeremy
Posts: 472
Joined: Thu Jul 31, 2008 6:35 pm
Location: Salisbury

Postby Jeremy » Sun Oct 12, 2008 9:17 am

My guess is that it's the material, brass, causing the problem. Brass has a fairly high linear thermal expansion coefficient (18.7 versus 13.0 for steel).

The problem may well be that the hole in the brass bush expands to give a clearance fit when warm, then the inherently good bearing properties of brass then allow the bush to spin on the shaft. The fact that the shaft is of a much smaller diameter than the ball race reduces the torque needed to make it spin, hence the reason it spins in preference to the bearing once it gets warm.

Making the bush from steel would most probably eliminate the problem, provided that it was a good interference fit on the shaft.

Jeremy

Alan Ward
Posts: 28
Joined: Wed Dec 13, 2006 3:31 pm
Location: mid Dorset

Clutch - leave it in or not.

Postby Alan Ward » Sun Oct 12, 2008 4:12 pm

Thanks Jeremy. I'm sure you are correct, I had not thought of the larger coefficienct of thermal expansion for brass. So fellow EV friends - take notice! Alan Ward


Return to “Conversions”

Who is online

Users browsing this forum: No registered users and 13 guests