regenerative braking

[duncan webster 1Participant @duncanwebster1]

[duncan webster 1Participant @duncanwebster1]

Can anyone explain how regenerative braking works on a battery loco with a PWM controller? My simple mind says that the back EMF from the motor acting as a dynamo must be less than battery volts or you wouldn't be able to drive any current through the motor. If back EMF is less than battery volts how can it charge the battery? I'm obviously missing something somewhere

[Peter G. ShawParticipant @peterg-shaw75338]

I'll have a go, simply because no-one else has attempted it. I expect to be shot down in flames.

Right, starting from first principles, the motor is being pulsed with maximum voltage pulses of varying width, and during the "off" period the motor "free-wheels" and due to its inherent inductance produces the so-called back-EMF. In a rather simplistic explanation, this was described to me many, many years ago as the motor, or rather the inductor therein, attempting to maintain the status quo. Which of course, it can't.

Now for some funny stuff. For back EMF to develop, there has to be a resistance to develop it across, and as we all know, I=V/R, simplistically that is. But the energy stored in the inductor is, shall we say fixed and was dependant on the current in the first place, therefore we can say that if R is large, then V must also be large to maintain I. And vice versa. So, it is perfectly possible for a inductor, having been fed from say 12v, to generate a back-EMF of many hundreds of volts.

Slight diversion here. In a previous life, used to work for BT as a technician and was once asked to somehow devise a circuit which would determine if a specific spark quench circuit was working correctly. As part of my investigations, I managed to see what looked like a 900v spike from an electro-magnet energised from 50V. The method of monitoring was a bit crude so it might not have been 900v, but nevertheless it darned high, and I didn't fancy my fingers anywhere near it.

Back to business. Assuming then that we can get this high voltage spike, then perhaps all we need to do, and I admit this now pure theory for me, is to connect the back-EMF via a suitably arranged diode to the battery. Hey presto, excess voltage should end up being shunted back into the battery and at the same time clamping the back-EMF to a suitable safe level.

I look forward to being told I'm talking a right load of old cobblers.

Actually, it is possible to use this back-EMF for speed control to ensure constant speed at a particular setting. I have a pair of books by Roger Amos which goes into this in respect of OO gauge model railways. I've had a 0-6-0 tank take about 15 minutes to travel 1 metre. I also used to have a mains operated circuit comprising a thyristor, diac, diode, capacitor & variable resistor for use with my mains drill. It was quite interesting feeling the drill skip-cycling on no-load, and then turning to a deep growl whilst still turning under full load.

Hope this helps,

Peter G. Shaw

[duncan webster 1Participant @duncanwebster1]

Sorry Peter, I'm still confused.

I recognise that if the motor is being powered from a PWM you will get an inductive spike when the current is switched off (if you don't have a freewheel diode), but when the motor is not being powered, ie mark/space = 0, which is when you would want braking, there are no inductive spikes, all you have is the voltage generated by the motor as a dynamo.

Does it somehow short circuit the motor for a very short period to allow the magnetic field to build up, then break the circuit to create inductive spikes?

[Bob LilleyParticipant @boblilley18824]

Hi Duncan. In simple terms when volts are applied to a permenant magnet motor sufficient current flows into the motor such that the back emf generated equals the volts applied. When no volts are applied to ta permenant magnet motor and it is driven on overrun it generates a voltage at it's terminals which can be well in excess of the battery volts. If the controller allows it, this voltage will produce a current flow into the battery which will brake the motor and hence the loco. To protect the controller electronics, some controllers limit the voltage to say 60 volts on a nominal 24 volt system. If you want more detailed information I suggest you look at the 4QD web site which has an excellent article on this subject. Regards Bob Lilley.

[Neil WyattModerator @neilwyatt]

I think most (all) model locos use back-emf braking rather than regenerative braking (the latter is rather complex than it seems as things have to be switched so the motor back EMF exceeds the battery voltage as Peter points out).

Back EMF braking is much simpler, you just short the motor which (at least in the example of my loco) virtually locks it solid.

Try it with a small electric motor with a gear fitted, try spinning it with and without the terminals shorted.

Neil

[Gas_mantle.Participant @gas_mantle]

The class 76 locos used on the Woodhead route between Manchester – Sheffield were 1500v DC with regenerative braking.

The idea was that because of the summit in the middle of the route locos would draw current whilst climbing but once over the top the regenerative braking of the loco put power back into the overhead wires.

It used to be said that a loaded coal train braking downhill generated enough current to power an empty train climbing uphill in the opposite direction. Personally I find that hard to believe but I've heard a few people say it was true.

Peter.

[Peter G. ShawParticipant @peterg-shaw75338]

Duncan,

Sorry about that, I did indeed miss that it was regenerative braking you were asking about. Nevertheless, in a simplistic manner ('cos that's all I can do) the principle must still apply given that if the controller is fully closed, then the motor is not perhaps freewheeling, but being driven in the manner of a car dynamo (in the days when cars had them). (In fact, a good few years ago, some people used to look out for a particular Lucas dynamo as it made a very good motor for electric locos.)

Now any dynamo (or motor working as one) will produce a voltage (and current) and just as I said before, the actual voltage will be dependant on the load into which it is feeding: low load – low voltage, high load – high voltage. Now obviously regenerative braking means that the generated power is fed into something which absorbs the power & dissipates it elsewhere. So really, on a basic level, which is all I can talk about, it's just the same. And if I was having a go at designing same, I'd certainly be starting with a diode biased such that the back-EMF would have to be greater than a specific value before current could be taken.

It's an unfortunate fact of life that any inductive component will produce a back-EMF the value of which is dependant on how the designer chooses to restrict it. Remember car ignition points? And how they used to create spikes and craters? That's the same effect. Sometimes you can see a faint electric glow from a switch when you switch something off. Again, the same effect only compounded by the fact that it's ac. The value of the back-EMF is directly dependant on the availability of any load into which it may feed: no load, ie open circuit, and the sky's the limit. Remember, to jump a gap requires a voltage of 10Kv/cm or thereabouts. We in BT had problems in the days of Strowger where selector magnets which drew 1A at 50V would have caused sparking and burning of contacts were it not for the fact that every magnet had a 2uF capacitor in series with a 200 ohm resistor connected across the coil.

Anyway, enough of that, I'm supposed to be in bed recovering from a dreaded lurgy!

Cheers,

Peter G. Shaw

[AjohnwParticipant @ajohnw51620]

Basically one aspect – there is current running about in a the motor when it's driving something. When it's turned off the current is still there and must be given some where to go. A resistor might be used to do that and the time taken to get rid of the current that is stored is proportional to the voltage across it. Often it's simply a reversed diode but if a resistance is added to that the voltage across the motor when it's turned off will be added to the diode voltage so it will decay more quickly.

2nd aspect if a DC motor is being driven by something it will generate power and in simple terms that has to go some where

That's ever so basics but back emf and regenative braking are one and the same thing really. It's just a case of where the power goes. The average voltage on field windings on the motor have an effect on the average voltage the armature develops when it's being driven by some force and also the power the motor can generate. The power it generates that is then absorbed by something determines the braking force that is developed. On electric vehicles it's generally merged in with the normal brakes so that the user is unaware of it. I used the term average because it's usually actually varied mark space ratio pulses of the supply voltage.

John

–

[Neil WyattModerator @neilwyatt]

The key thing with regenerative braking is that the motor MUST be driven faster than the applied voltage would make it rotate, not easy to do with a simple battery and diode – one way is to switch the batteries from series to parallel.

With PWM it's easier you just use a mark-pulse ratio that would drive the motor at a significantly lower speed, and if your drive transistors are MOSFETS (bidirectional) pulses of back EMF build up and then are switched to 'feed' the battery.

Of course it's never that simple and you can fry your control circuitry if things don't quite work…

Neil

[Anonymous]

Posted by Neil Wyatt on 20/09/2015 21:37:27:

and if your drive transistors are MOSFETS (bidirectional)

That's a new one on me, I've only been using MOSFETs to switch power in one direction.

On electric/hybrid vehicles the regen energy is ideally put back into the battery. The amount of energy that can be absorbed by the battery without over-voltaging is heavily dependent on the SoC of the battery. Controlling the SoC is thus an important function of the battery management system. When testing batteries the parameter of interest that can be measured is charge acceptance.

Andrew

[AjohnwParticipant @ajohnw51620]

A dc motors armature voltage forms a back emf against the field Neil so when used as a dynamo the field control can be used to set the voltage it can generate. Field control can also extend the speed range the the motor can run over.

It's usually called separate excitation in relationship to DC electric motor control. The same principles can be used to control regenerative braking – and have been. I did work on Lucas EV's.

John

–

[AjohnwParticipant @ajohnw51620]

I was in a bit of a rush with the other post. In some ways it's easier to explain from motor speed and torque control using dc motors and the sep ex I mentioned. The usual way of controlling motor speed is some form of pulse width modulation with the field windings connected to the supply. I hope I get this the right way round – fairly sure I will but it's been a long time.

As the speed of a motor increases the back emf it's armature generates increases and at some point it gets close to the field voltage and at that point it can't go any faster. These are average voltages when things are driven with pulses even when it's the field. They are actually setting mean currents, usually across a diode when the power is off so that the current decay rate is slow. So say the field normally has half of the supply voltage on it. That will limit the speed it can produce and also limit the voltage that it can generate when the armature is driven by say braking but the field can also be changed so the motor can be made to go faster or slower purely on that. This is how the motor is made to produce more volts than the battery that is driving it for regen braking. In practice the whole thing is regulated by armature current both ways – providing power or regenerative braking. Most of the regen will be applied when the accelerator isn't depressed but it can also be blended in with the brakes.

Sep ex is not exactly a safe thing to play around with. Do the wrong thing with the field and the motor may rev so fast that it explodes due to centrifugal force. The other aspect is that field currents have an effect on torque.

I aught to remember which way round things go as I replaced a fair bit of a controller with a micro processor. The states were interrupt driven and noise caused some problems. If I hadn't fitted a switch to make 100% sure thing shut down properly a rather large motor would have exploded showering bits all over the place. That had happened in the past while the original controller was being worked on but the guy working on it ran out of the room.

The armature emf thing I mentioned is a model Motorola have used to describe speed control circuits. I aught to get the book out and check that but it's time for bed. They don't consider sep ex at all. Few do. In this case the motor was designed to be driven like that from a 200 odd voltage battery pack.

John

–

[duncan webster 1Participant @duncanwebster1]

I've looked at the 4QD site, very interesting. It appears to be as I suggested in my 16:42 post, the motor is repeatedly shorted, allowing the dynamo back EMF to drive a current, and when this current is switched off the inductive spike pushes charge back into the battery. It must be a bit clever to know how long to short the motor to allow full current to build up without burning anything out.

As energy was taken out of the battery to accelerate the train, it seems that you can't overcharge the battery by regen braking as there isn't enough energy available. That is unless you start at the top of a long hill!

With a permanent magnet motor you can't increase the field to get more dynamo back EMF. Perhaps if you had a loco with 2 motors in parallel you could connect them in series for braking, or as Neil says switch batteries from series to parallel, but it's all getting a bit complicated.

I tried field weakening on a parallel wound motor once (Lucas dynamo, still got a couple under the bench). It worked a treat until I weakened the field too much when it stopped dead and drew a large current through the armature. Never understood that either. It didn't damage the motor as I switched it off quick time. Dead easy to reverse, just swap the filed, which is low current so toggle switch. All done with op-amps and a big FET. I'd use an Arduino now, but the dynamo is pretty big compared with modern permanent magnet motors

I think I've finally got it, (shades of Prof Higgins in My Fair Lady). Thanks for all your contributions

[Neil WyattModerator @neilwyatt]

Andrew, check out:

http://www.quora.com/Is-MOSFET-a-bidirectional-device-or-unidirectional-device-What-is-a-clear-explanation-about-this

Neil

[AjohnwParticipant @ajohnw51620]

Recharge isn't 100% efficient Duncan. There are losses so in principle it's always possible to put back in some of what is taken out.

I can't reliably remember how the field control manages regen but field weakening rings a bell and that sort of makes sense. As some one mentioned permanent magnet motors don't make good generators. With zero field current all that would be left would be residual magnetism. I've worked on too many different things to remember everything so that could be entirely incorrect.

Motors in the extreme are a very specialised subject. I had thoughts about a variable speed lathe a long time ago and talked to a specialist. He happened to have an aircraft generator about that he had been bought from a scrap yard. He gave it to me along with suggestion about armature and field voltages. They were way higher than the volts it was meant to generate. I doubt if this was for a laugh at my expense. As it had a decent sized spindle and bearings I intended to try and use it as a head stock but stuck it in a cupboard at work while using power supplies to control it to see what it would do and some one nicked it. It just goes to show that the optimums for generation aren't optimums for drive as from the power supply trials the numbers he gave me were sensible.

John

–

[Bob LilleyParticipant @boblilley18824]

Nearly there Duncan but it is even simpler. When regen braking the PWM controller is not in operation and in effect is replaced by a diode which allows current to pass from the motor/generator into the battery, but prevents the battery feeding the motor. Therefore during regen you have a simple dc generator feeding the battery.

You are also correct that the battery would normally be partially discharged thus allowing further charging except in the case of Michigan, described in this month's ME magazine, which has an alternator continuously charging the battery. In this case the battery is fully charged and a high voltage can be generated during regenerative braking.

Trust this answers your query.

Regards Bob Lilley

 

Edited By Bob Lilley on 21/09/2015 14:37:34

[AjohnwParticipant @ajohnw51620]

That would really be good for some types of battery Bob. It gets a lot more complicated than that to do it properly even on wet lead acid batteries.

John

–

[Anonymous]

Neil: Thanks for the link, although I would admit to being slightly disappointed. I thought there was some clever way of getting 'reverse power' through a MOSFET that I wasn't aware of. There is an arrangement of two MOSFETs that allows bi-directional current to flow, while paralleling the gates so that only one drive circuit is needed but still blocking current when the gate drive is off. I used the circuit as a connector/disconnector on parallel battery strings for a Formula1 KERS system.

Andrew

[AjohnwParticipant @ajohnw51620]

It might have been based around 2 mosfets opposite polarities maybe with drive built in as they are diodes when reversed so if both are off should block ???????? I'd need several cups of tea to be sure.

There is very little about on separate excitation only power, speed and torque curves also showing that they can run at over the motors base speed. Also golf cart motors for separate excitation – those probably do regen sensibly but it's not to difficult on wet lead acid traction batteries because they don't mind being overcharged as much as some other types. Some others are a nightmare when fully charged as they more or less go open circuit which causes problems when cells are in series. Others just don't like it full stop. People were trying to figure out by pass methods when I was involved. The old drives in things like milk floats were really crude.

Some years ago people were doing strange things with old fashioned starter motors and dynamos. Trouble is that they run at rather large currents and low voltages. I'd guess this was pre web but their might be some info somewhere. There was this wheel chair about at work that was way too fast to ever go on the market controlled with a joystick and sep ex.

John

–

[Neil WyattModerator @neilwyatt]

Posted by Andrew Johnston on 21/09/2015 16:29:40:

Neil: Thanks for the link, although I would admit to being slightly disappointed. I thought there was some clever way of getting 'reverse power' through a MOSFET that I wasn't aware of. There is an arrangement of two MOSFETs that allows bi-directional current to flow, while paralleling the gates so that only one drive circuit is needed but still blocking current when the gate drive is off. I used the circuit as a connector/disconnector on parallel battery strings for a Formula1 KERS system.

Andrew

If it wasn't possible, then the little chips full of 'analogue switches' wouldn't work…

But my bad MOSFETs have a body diode which means they aren't reversible (at least above Vf for the body diode)…

However most jFETs don't give a (significant) hoot which way round source and drain are. This datasheet clearly states 'note drain and source are interchangeable'.

What I can't see working is a diode alone – that will only work with the motor running ~faster than if it was unloaded and connected to the battery i.e. over-running, which would happen anyway with or without the diode. All the diode would do is stop the battery running the motor at lower speeds..

Neil

[Anonymous]

Good lord, a JFET! That's a blast from the past. Neil is correct regarding the operation of JFET analog switches. However, I suspect that the majority of switches fabricated using CMOS need to use two MOSFETs per switch.

As far as I am aware the drain-source diode is inherent in all MOSFETs due to the way that they are fabricated. So one might as well characterise the diode and use it for something useful. Of course if you need a MOSFET to block bi-directionally you need two of them; as in those solid state relays that use MOSFETs.

Andrew

[MuzzerParticipant @muzzer]

Ah yes, the 2N3819…

As Andrew says, you have to put a pair of MOSFETs back to back if you want bidirectional blocking. If you want to keep the losses the same as you would have had with a single unidirectional device, you will need to quadruple the number of devices and hence the cost.

But for a motor drive, you don't need to worry about that. You invariably design the motor so that the maximum back EMF is less than the DC bus. You can control the motor in field weakened operation above this point but this requires clever software and you need to have a plan for when you lose control that doesn't involve popping the switches.

For best operation, you use a half bridge on each phase of the motor, whether brushed, brushless, 3-phase or whatever. This allows you complete control in all 4 quadrants ie speeds and torques in either direction and in any combination. 

For easiest and most versatile control of a brushed DC motor, the best solution is a full "H-bridge" controller. This is 2 half bridges and gives 4-quadrant control, which is what the OP was after. You can buy integrated H-bridge driver power devices with all sorts of in-built protection features or doubtless you can get a ropey equivalent from an auction site.

Murray

Edited By Muzzer on 22/09/2015 13:56:29

[MuzzerParticipant @muzzer]

Something like this would be almost indestructible if you mounted it suitably.Can be controlled with PWM and DIR(ection) inputs. Requires an handful of caps etc to make it complete.

[Tim StevensParticipant @timstevens64731]

In the old days, cars had dynamos which would act as motors. Connect to a battery as normal (for them) – one side to earth on the casing, and the other side to both Field and Armature connections. The dynamo will then turn at a modest and steady speed, and can be used as a motor. Now drive the 'motor' faster in the same direction and it starts to generate, rather than consume. No fancy switching, no change in connections, as long as the rpm is rather more than motoring speed it generates, and in doing so absorbs power from the driver and charges the battery – ie regenerative braking. All that stops the dynamo acting as a motor at low rpm in 'real life' is the cut-out, which disconnects the output at (about) the cross-over point when output changes to input.

Perhaps this will help to understand what is going on?

Cheers, Tim

[Author]

Posts