DIY Electric Car Forums banner

1 - 20 of 22 Posts

·
Registered
Joined
·
268 Posts
Discussion Starter #1
I have been looking at building a nice low cost DC controller as I have a couple of series wound Fork Truck motors that might like to become part of something cool.

I've built a few 4 quadrant PMDC controllers in the past including one (2 Quadrant) 1200A 24v - so I have a good starting point.

I've looked at the instructions for Zilla controllers an they appear only to control the direction of the motor via contactors - and don't mention regenerative braking. Is this common?

Obviously, if you just reverse the current to a series wound motor, it will turn the same way because both the rotor and stator fields have reversed. So you need to control the two separately.

Do most controllers use contactors to reverse the flow through one?

Do they have one non reversing controller for the Armature and a reversing controller for the field windings - which would give you the ability to reverse and regeneratively brake?

Do they use contactors for reversing then connect the windings in parallel (and limit the current flow electronically) and not bother with regen?

Or is the regen achieved using the magnetic hysteresis in the core to generate current in the windings - possibly using a pulse to energise the field then regen for a short duration afterwards?

The regen on my AC controller is very good - makes a big difference to my range as there are intersections at the bottom of the two big hills I have to climb / decend - so I cannot just coast up the next hill.

Without giving away any 'trade secrets', what strategy is generally employed? Are there any good reasons (including cost / complexity) to avoid any?

I favour the idea of having separate controllers for the armature and field as it gives you the greatest control over the motor - but increases the number of transistors by 50%. Maybe there is a more cunning solution though!

Thanks in advance,

Si
 

·
Registered
Joined
·
2,229 Posts
I'm no expert but from what i understand regen is very difficult on a series motor for a few reasons one of which is that the arm current reverses which would then cause the series field to fight back. I'm looking at regen as i have a 50/50 compound wound fork motor.

Again i believe the popular controllers don't actively implement reversing as most are used with some form of gearbox with a mechanical reverse. I have seen it implemented on a few vehicles on evalbum.

I don't think it is possible within reason to control the series field with a sprerate controller.
 

·
Registered
Joined
·
7,837 Posts
I've looked at the instructions for Zilla controllers an they appear only to control the direction of the motor via contactors - and don't mention regenerative braking. Is this common?
Hi Si,

Yes, almost all series motor controllers are this way. Basically a buck converter where the output inductor is the motor itself. Usually just a switch (transistor), diode and capacitor. Strictly unidirectional.

For a while, Curtis had a regen version, and I think someone else offers one. But these have proven, for the most part, to be unreliable and more trouble than they're worth.

Obviously, if you just reverse the current to a series wound motor, it will turn the same way because both the rotor and stator fields have reversed. So you need to control the two separately
Yes, so the series motor controllers which are applied to reversible applications use a contactor set to reverse the polarity of the field. These controllers typically also have an extra diode across the armature to prevent full field plugging.

Do most controllers use contactors to reverse the flow through one?
Yes, the field. See above.

Do they have one non reversing controller for the Armature and a reversing controller for the field windings - which would give you the ability to reverse and regeneratively brake?
All, to my knowledge, reverse the field only. This can provide plug braking, not regen.

Do they use contactors for reversing then connect the windings in parallel (and limit the current flow electronically) and not bother with regen?
No and yes. They always have the armature and field in series, although the plugging diode allows higher armature current than field current. And yes, they don't bother with regen.

Or is the regen achieved using the magnetic hysteresis in the core to generate current in the windings - possibly using a pulse to energise the field then regen for a short duration afterwards?
Here again, most do not regen. But the few that have tried it will "bump" the field to get it going.

The regen on my AC controller is very good - makes a big difference to my range as there are intersections at the bottom of the two big hills I have to climb / decend - so I cannot just coast up the next hill.
Yeah, regen is nice.

Without giving away any 'trade secrets', what strategy is generally employed? Are there any good reasons (including cost / complexity) to avoid any?
I don't know of any trade secrets. Maybe because they are secret. But like I said, few if any series motor controllers do regen. Why? Yeah, cost and complexity. It is just too much to get a reliable series motor regen system, mainly because the series generator is unstable.

I favour the idea of having separate controllers for the armature and field as it gives you the greatest control over the motor - but increases the number of transistors by 50%. Maybe there is a more cunning solution though!
Actually it increases the number of transistors from one to six. So, the smart approach is to use a separately excited motor instead of a series wound motor. That way four of the switches can be rated at about 1/10th of the armature switches. In other words, use a 500 amp half bridge for the armature control and a 50 amp H bridge for the field.

The downside is that you need the motor field wound for this. Like with about 10 times the number of turns. Another downside is that these series wound motors used in EVs will have a brush advance to get acceptable commutation at the high voltage and current and RPM. When you go into regen mode, the motor becomes a generator and then this advance is in the wrong direction, making commutation unacceptable. The recourse here is to use neutrally timed motors having interpoles for commutation purpose. This is true for SepEx as well as series motors.

You seem like a smart guy :) So, good luck trying to figure out a better way.

major
 

·
Registered
Joined
·
122 Posts
I cannot answer your more technical questions, but I can tell you a reason why there are very few regen controllers for series wound motors.

Zapi has a regen controller for series wound motor that people consistently blow up because they try to use it with their motor with advanced brushes. Also with large series wound motors, you supposedly need a motor with interpoles to regen safely.

Zapi is the only controller for series wound that I know of that can regen, if used properly. Now the new Soliton1 has the capacity to do regen, Tesseract and Qer will likely chime in and answer your tech questions on this. But they have said that they hesitate to offer that function because they worry customers will attempt to turn it on while connected to a motor with advanced brushes, and blow it up.

Another good reason is, there are very few large high voltage motors made that are set neutral timing and have interpoles. You usually need to order them custom, or find an appropriate (and sometimes modify) a big used forklift motor.

Netgain is coming out with an 11 inch motor with interpoles that is set at neutral timing and can handle high voltages and a controller (with regen, supposedly) to go with it, so that will be a new player.
 

·
Registered
Joined
·
122 Posts
Ha ha wow it looks like all three of us answered at almost the same time, and we all made some similar points.
 

·
Registered
Joined
·
268 Posts
Discussion Starter #6 (Edited)
Thanks for the info - that makes the whole thing tremendously easier!

Is there any mileage in using a boost converter with lower voltage higher Ah batteries?

Although it would probably be way too complex, what about this....
If you advance the brushes to optimize running in one direction, I presume the issue when generating is that you disconnect a winding when it is near it's max current rather than when it has trailed off to closer to zero.

Although the amount of braking would be reduced, if you added a rotational encoder, you could ramp down the field current before the commutator disconnects. In fact, if you know the rotational speed and the number of poles , you just detect when current starts flowing back to the batteries, then turn off the field windings after a short delay such that there is no field when the commutator disconnects hence no arc?

Si
 

·
Registered
Joined
·
7,837 Posts
Is there any mileage in using a boost converter with lower voltage higher Ah batteries?
You can't boost energy, so no. But on the other hand, must be something of advantage to boosting battery voltage because Toyota does it on the Prius.

Although it would probably be way too complex, what about this....
If you advance the brushes to optimize running in one direction, I presume the issue when generating is that you disconnect a winding when it is near it's max current rather than when it has trailed off to closer to zero.

Although the amount of braking would be reduced, if you added a rotational encoder, you could ramp down the field current before the commutator disconnects. In fact, if you know the rotational speed and the number of poles , you just detect when current starts flowing back to the batteries, then turn off the field windings after a short delay such that there is no field when the commutator disconnects hence no arc?
Got no idea what you're talking about. There are no disconnected windings and commutators don't disconnect anything, they switch current direction, but the winding is never disconnected.

Regards,

major
 

·
Registered
Joined
·
268 Posts
Discussion Starter #8
You can't boost energy, so no. But on the other hand, must be something of advantage to boosting battery voltage because Toyota does it on the Prius.
To drive more current through a given motor with lower voltage batteries?
Fewer larger batteries costs less and are easier to charge.

Got no idea what you're talking about. There are no disconnected windings and commutators don't disconnect anything, they switch current direction, but the winding is never disconnected.
When is the arc produced then? What happens at the change-over between directions?

Si
 

·
Registered
Joined
·
7,837 Posts
When is the arc produced then? What happens at the change-over between directions?
Hey Si,

If done properly, no arc is produced. And as for the "change over", current is reversed through the armature coils via commutation. Looking at a single armature coil, current flows one direction, then is shorted by the brush and then has current flow in the opposite as the associated comm segment exits from under the brush. All the armature coils are permanently connected in one continuous circuit. The winding pattern and brush contacts on the comm determine the number of current paths and the direction of current in portions of that circuit.

Regards,

major
 

·
Registered
Joined
·
3,219 Posts
Regen and the series dc motor is one of those questions that if you ask 3 engineers about you get 4 answers.

Yes, the Soliton1 already has the hardware inside to ATTEMPT regen, but that's mainly because it didn't cost much in parts to add it. Even IF Qer and I get it to work we still may not release the code to do so because it is hazardous to the health of an advanced motor. Take a stroll through the EVAlbum and you'll see plenty of people with Zapi H2 and H3 (series w/ regen) controllers hooked up to an advanced motor. Here's a prime example:

http://www.evalbum.com/2454

One thing that should be mentioned, regardless of whether it is possible to do regen on the series motor from a hardware perspective, the control algorithm for doing such will be very difficult! Without getting too much into feedback loop theory, the series generator is a "classically unstable" system. Consider this sequence of events:

The torque in a series motor is proportional to field flux (i.e. - field current) while the BEMF from the armature opposes the flow of current through the field proportional to the armature's speed. Thus, as the motor slows down during regeneration the braking torque increases. This in turn slows down the armature even more which further reduces BEMF, etc. and so on.

A runaway condition that either results in the motor briefly locking up (and the whole process repeating) or the current going to near-infinity (cooking off the controller or motor).

THIS is why you don't see regenerative braking as a feature in too many series dc motor controllers - it's simply very difficult to get it to behave well under all conditions and if the driving experience is jerky and/or things get destroyed every time you use it, the slight increase in range ain't even remotely worth it.
 

·
Registered
Joined
·
268 Posts
Discussion Starter #11 (Edited)
I'm rapidly coming to that conclusion!

The way I wondered was if you connect your PWM speed controller to the field windings alone, the armature windings should generate power. Full wave rectify (put it through a bridge rectifier so the polarity of the induced current does not matter) this and feed back in to the batteries and regulate the current flow with the PWM.

The switching would need to be via contactors and it's probably not going to give a nice smooth transition from drive to regen like you get with AC or PMDC) - but it should still deliver the goods.

I reckon the regen is giving me about 10% more range at the moment - but my town although mostly flat has a couple of decent hills that I have to climb (otherwise I'd be cycling to work!).

Si
 

·
Registered
Joined
·
2,229 Posts
My regen idea is to use the shunt field of my compound motor at full bore and feed the output into a boost converter. I'll switch out the series field during regen. I don't know if or how well this will work but time will tell.
 

·
Registered
Joined
·
14 Posts
The way I wondered was if you connect your PWM speed controller to the field windings alone, the armature windings should generate power. Full wave rectify (put it through a bridge rectifier so the polarity of the induced current does not matter) this and feed back in to the batteries and regulate the current flow with the PWM.
Try measuring the resistance of the field windings on a series motor. You'll need a milliohm meter to measure it. Most controllers won't like the duty cycle required to drive a field directly.
 

·
Registered
Joined
·
268 Posts
Discussion Starter #14
Try measuring the resistance of the field windings on a series motor. You'll need a milliohm meter to measure it. Most controllers won't like the duty cycle required to drive a field directly.
However, this thread is about building a custom controller - so it would be made for it. Anyway, it's only going to be half the combined resistance of the field and armature windings - so if it will drive the combination, it should drive individually.
 

·
Registered
Joined
·
3,219 Posts
However, this thread is about building a custom controller - so it would be made for it. Anyway, it's only going to be half the combined resistance of the field and armature windings - so if it will drive the combination, it should drive individually.
A custom controller, indeed. Consider, though, that the field in a "typical" series wound motor consumes around 1% of the total power but field current and armature current are always equal (when wired normally, that is). In other words, if there is 100V across the motor then there will 99V across the armature and 1V across the field. If the input voltage to the controller is, say, 200V, then the controller's duty cycle will be.... yep, 50%. Now, let's split the controller into two so as to drive the armature and field separately... The armature controller will need to supply 99V so it will operate at a duty cycle of 99/200, or 49.5% while the field controller will need to operate at 1/200V, or 0.5% duty. Let's say you want a good balance between minimizing switching losses and audible noise from the motor so you use a frequency of 10kHz (period of 100us). The switch in the field controller will be on for 500ns. One is also advised to keep the switching transition times (either the time it takes to go from fully off to fully on or vice versa) to 1% of the total time to minimize switching losses. So, you'll need to turn the switch on in 5ns and off in 5ns. As you might imagine, this is totally impractical.

Hence, why the rule of thumb for buck (or boost or buck-boost) converters to not change the voltage by more than 10:1 because of practical duty cycle limitations. You can use digital techniques such as dithering (randomly skipping pulses) to increase the practical transformation limit up to say 100:1, but that radically increases the control algorithm complexity (as Qer well knows from creating such code for the Soliton1). If you need to convert a voltage by more than 10:1 you are well advised to use a transformer.

Or you can drop down in frequency a la the Curtis... that's a bit too kludgy for me, but hey - it works.
 

·
Registered
Joined
·
268 Posts
Discussion Starter #17
Thanks chaps - I'd run in to exactly this with my charger as I was testing it on a single 12v battery - so even with a 10 bit PWM resolution the numbers were getting pretty small. Dropping the frequency from 41kHz to 8kHz improved matters a lot.

Dithering / spread spectrum is a good idea! I've built spread spectrum transmitters before using a 7 bit (shift register based) pseudo random number generator with a 128 bit repeating cycle then a multiplier as the modulator. I'll have to give it a go in software.

Have you tried using the Propeller microcontrollers? 8 x 100MHz processors on one chip working concurrently. You can do some pretty amazing stuff bu splitting a task between the processors including direct synthesis radio. I have mine generating vector controlled 3 phase PWM at the moment - it was a fall back position in case I couldn't get my AC controller to work (since it came from a junk yard). It uses 3 processors for the PWM, one for the vector control and the a fifth as an overall supervisor. Worked pretty well on a small 3 phase motor. It is fast enough to generate 400Hz 3 phase sine wave - all written in a high level language. Radio needed assembler though!

Si
 

·
Registered
Joined
·
3,219 Posts
Thanks chaps
See, I'm not just a rapacious and greedy capitalist, no matter what my own mom (that's mum to you) says.


Have you tried using the Propeller microcontrollers?
Nah... my knowledge of digital stuff is pretty thin. If it weren't for Qer the Soliton1 would be a glorified Curtis.
 

·
Registered
Joined
·
1,760 Posts
Have you tried using the Propeller microcontrollers? 8 x 100MHz processors on one chip working concurrently.
I don't like the Propeller out of two reasons:

  1. Even though the 8 cores gives a lot of performance the lack of interrupts and the slow communication between the cores might lead to serious bottlenecks that takes a lot of effort to avoid.
  2. Last time I looked the documentation were still not 100% complete which, for example, means there's no support for the Propeller in gcc. As I do work on many different hardware setups (sh, arm, boadcom, avr, ix86 etc) I prefer to use one set of tools that works on them all instead of one tool per platform.
I'd rather use arm or avr32 if I need serious performance in a microcontroller.

YMMV, of course. :D
 

·
Registered
Joined
·
268 Posts
Discussion Starter #20
There is no one, or one family of processors which is good for everything - but there are definitely some applications which lend themselves to the concurrency (I have a bit of background with Transputers) - but certainly not everything. I agree about the GCC issue - but you can't have everything!

Si
 
1 - 20 of 22 Posts
Top