Here is a article that I wrote years ago for Home Power magazine.
I do not recommend doing this conversion, it tends to blow up controllers and also can overheat motor brushes if not additionally regulated.
I am posting it here just for the educational value of it.

Regenerative Braking a DC Series Motor
by Otmar Ebenhoech

It takes energy to accelerate a vehicle. Any vehicle moving down
the road stores kinetic energy in its motion. Applying the brakes
in your car, whether it is electric or engine-powered, converts that
kinetic energy of motion into heat. This heat energy is lost.

What if we could convert some of that kinetic energy back into
electricity instead of heat? In an electric car, couldn't we put some
of the energy used to accelerate our car back into the batteries
when we slow down? The answer is yes! It's known as
regenerative braking, or REGEN for short.

In a vehicle equipped with REGEN, the motor works as a
generator, trading the motion of the vehicle (momentum) for a
charge in the battery (electricity). Where the vehicle normally
experiences a lot of braking in its driving cycle, REGEN helps to
increase overall vehicle range. As well, it conserves the brakes.

Electric vehicles equipped with AC motor drive systems all have
this feature built in. Unfortunately, AC drive systems are still
very expensive. As well, those that are only triple the price of a
DC system are quite under-powered for the average conversion.

My interest has been to build a REGEN system to work with the
components that most people have in their EVs. This involves
series-wound motors and Curtis PMC mosfet controllers.

Hooked on REGEN

I got the REGEN bug from the first EV I worked on. It was a 1981
Ford Fairmont station wagon with a noisy SCR controller and a
series wound motor. We called it the Lead Sled. It was heavy,
slow and a bit power hungry -- but it did have REGEN. There is
something so satisfying about coming up to a stop knowing you
are not totally wasting all the energy that was consumed to get up
to speed in the first place. I didn't even mind those ridiculous LA
stop signs as much as I used to.

My next EV was a nice new Honda CRX with brand new
components but -- no REGEN. Something was wrong. Every time I
was forced to a stop, I thought of all the energy I had just wasted.
This car could have rolled three quarters of a mile for every city
stop I made from 35 mph! The search for REGEN was on!

What's Real?

Everywhere I went I asked about REGEN. Everyone had different
ideas. "It can't be done ... it requires a shunt motor ... you can do it
but it's likely to ruin your motor, controller, or batteries ... it isn't
worth the trouble." Rarely could anyone tell me why.

Many people told me of experiments that had ended in fried
motors. It seems that the motors would experience commutator
meltdown. It was time to talk to a manufacturer. I called Eric
Dieroff at Advanced DC Motors for some help.

Eric educated me on characteristics of series DC motors and REGEN
issues. It was important to maintain equal currents in the
armature and field windings of the motor. There is an area on the
commutator where adjoining commutator bars have very little
voltage potential between them. The motor's brushes are located
here, to minimize arcing and heating, and meltdown or brush
failure. Failure to keep equal currents in the two windings causes
this critical dead zone to move. The brushes now short out bars
with a higher differential voltage. Sparking results. Excessive
arcing can extend all the way to the opposite polarity brush,
creating a fireball effect. Since all the REGEN circuits I had seen so
far separately excited the field with a high current, I thought I
could see the cause of meltdowns and short-lived brushes.

Note: By comparison, shunt wound motors have a separate
miniature field winding called a compensation winding that
creates a dead spot for good commutation. This mitigates the
need for equal current balance.


I eventually found a circuit design that used a boost/flyback
converter. Although it was intended for a PM (permanent
magnet) motor, I thought that it might be the right way to go.
Shortly thereafter I read a Design News article describing REGEN
on an old Cableform SCR controller. This system, which was similar
to the old station wagon system that I never had traced out, was
also of the boost converter type.

Boost converters can easily develop enough voltage to ensure that
charging current can flow. Older designs had to switch the
batteries into a parallel configuration so that battery pack voltage
was lower than the generator voltage. Without such complexities,
current wouldn't flow. The circuit designed around the boost
converter was simpler, eliminating three contactors and a bunch
of wiring! It's main feature was that it kept the field and
armature in series, assuring equal currents in the field and
armature except for a short period during field excitation. As well,
I found that I could eliminate half of one of the field-reversing
contactors by connecting the REGEN diode directly to the motor.
This eliminated one contact on the reversing contactors. It also
eliminated the possibility of a faulty circuit forcing the vehicle to
take off in reverse! I felt my confidence growing. I was ready to
risk a test.

In a boost converter, a high current at low voltage is fed through
an inductor. This builds up a large magnetic field (see Using
Magnetic Field to Change Voltages, Dr. Kluge, HP #37 for basics on
buck/boost circuits). My circuit does this by shunting the
motor/generator through the controller and the REGEN diode (Fig.
1a and b). The motor itself is the inductor. When the controller
reaches its current limit, it turns off. This causes the magnetic
field to collapse, in turn creating a large voltage spike. Current
flows in the same direction as the original current build up. This
surge of a high voltage and proportionately lower average current
flows through the controller's freewheel diode and -- back into
the batteries.

Crude Testing

The first test was rather crude but served well as a proof of
concept. The vehicle was a lightweight EV called a Freeway with a
72-volt system, a compound wound Prestolite motor with a
manual field reversing switch, and a Curtis PMC 1221B controller.
A paint stick wrapped with heating wire served as a field
excitement resistor that was controlled by a dash switch.

The system worked! It did lack control, though. Motor current
was limited by the controller at 400 amps. By the time I pulled
back the sticky accelerator pedal, toggled the field switch on the
motor behind me, waited for the controller to ramp up, and
switched on the field excitation circuit, the STOP sign had gone by!


Some automation was needed. I modified the circuit to include a
changeover contactor on the field reversal. I also added a "delay"
relay on the excitation circuit.

It was clear that I also needed to reduce the level of the current
limiting feature of the Curtis controller when I was in REGEN
mode. This unit is pretty well sealed. I wanted to avoid breaking
the seal (and the warranty) on the controller. I worked around
this dilemma by contacting the wiper terminal on the current-
limit potentiometer through the hole in the case provided for this
adjustment. I connected a wire from this point to ground through
a signal diode and a 10K potentiometer. The amount of current
limiting was now under my control.

You don't want REGEN to work when the vehicle is fully charged,
or it will overcharge the batteries. For this reason, I added a
"comparator" to the circuit. When it sensed full battery voltage, it
reduced the level of current limit, and regulated overcharging.

A road test confirmed a working circuit. The ability to vary the
current limit allowed quick or gentle slowing almost to a full stop
without downshifting (gears). But the controller's delay circuitry
was very annoying. You see, for REGEN to work, it requires that
the controller be full on. However, it takes about two seconds to
ramp up and half that to come down (or turn off) again. For this
reason, I had to install a one-second delay in the main contactor
just to give the controller time to ramp down before driving again.

Aside from the controller delays, I was happy with the
improvements to the circuit. The vehicle was a pleasure to drive.
As planned, REGEN was limited for the first couple of blocks on a
full charge until the batteries could accept a charge without
exceeding 14 Volts per 12 Volt block. Nighttime tests at 400
motor amps of charge current showed no unusual arcing of the
commutator and brushes. The motor and controller did get hotter
than during normal driving. This was to be expected. After all,
they were now working hard in both acceleration AND braking

Into the Samurai

The system was now ready for full size tests on a more normal
vehicle. The test bed was an electric conversion on a 4WD (four
wheel drive) Suzuki Samurai that I was building with some Lopez
High School students in Washington.

At this point, to circumvent the annoying delays built into the
controller itself, I decided to void the warranty on the controller
by opening it up. In addition to bypassing the internal delays, I
also mounted some of the REGEN circuit's control electronics inside
the controller housing. The result was a cleaner installation!

Standard Disclaimer

The Curtis PMC controller is the standard in the industry today. It
is very reliable and simple to use. The electronic guts of these
controllers are VERY complex. High currents and their related
magnetic fields can cause unexpected problems. Curtis does not
supply circuit diagrams. They don't want anyone to mess with it.
They are generally not interested in what an experimenter might
want to do with their product. And they don't want to be liable
for anything that happens to you. Hence, opening the box voids
the warranty. We are on our own when we start modifying
things. At best, it's a high risk adventure -- even if you do
consider yourself an electronic engineer.

The REGEN circuit itself introduces an element of danger into an
EV's circuitry. What happens if both contactors were to get stuck
in the ON position when the controller was in REGEN mode? The
EV would accelerate on the application of REGEN, with the rate of
acceleration proportional to the REGEN current. The addition of
lock-out microswitches on both contactors can reduce the chance
of this happening.

Getting Into the Controller

The first step in modifying my controller was to remove the
potting compound from the end and the screws on the bottom. To
do this, I put the controller in a large pot of boiling water until it
was hot. This softens the potting compound. I used a nail (any
sharp object will do) to remove the potting material from the
location of the six screws in the bottom and removed these
screws. The potting compound on the terminal end can be broken
free from the case with a punch and a hammer. By tapping the
potting carefully around the edge where it sticks to the aluminum,
I broke this seal. I removed the electronics from the case by
putting a screwdriver through the M- and A2 terminals and
pulling. The heat sink and mounted electronics slid out. I
removed the rest of the potting compound from the terminals
with pliers. I saved the foam piece for the re-potting process.

The Modification
The control board comes off the power board by desoldering
the seven interconnects. After extensive analysis and testing, I
identified the function of the components and took the following

C9 controls the acceleration delay. It's below the base terminal
of the large heat sunk transistor. I removed it completely.

D7 affects acceleration input scaling, making the controller
more sensitive at low speeds. D7 is the second diode from the
large transistor toward the current limit pot. I removed D7 to
disable it.

R72 and R73 control the over-voltage shutdown and they are
factory set for 155 volts. I wanted it set at 140 volts (to avoid
overcharging the batteries). To do this, I replaced the 10KO value
of R72 with a metal film 10K (for more temperature stability) and
replaced R73 with a metal 86.6KO resistor. (Note: The Samurai
conversion uses a 120 V battery pack. R73 would have to be a
smaller value for a lower voltage system. A variac with a bridge
rectifier works well to test and adjust this. Be sure to bypass the
ground lead on any test equipment to avoid ground loops through
the non-isolated variac.)

External connections are needed to reduce the current limit and to
hold the controller in full throttle state during REGEN (Fig 2).

The current limit can be reduced by tapping into the current
limit adj. pot wiper through a diode, a 1.5KO resistor and a 4KO
Pot to terminal B-. This gives a motor current range of about 100
to 200 amps. (These resistors may need to be adjusted to obtain
reasonable motor currents.)

The controller can be held at full throttle (during REGEN) by
pulling pin 5 of IC 5 high to VS through a 100kO resistor. Older
models (board 98627C) needed a 47KO resistor. The pull-up
resistor should be the same value as R62. IC 5 is the fifth IC from
the end with KSI and pot terminals on it. VS is the only square
pad on the test terminals on the top of the board. I connected both
the current limit adj and the pull-up resistor through a DPST relay
with shielded wire inside the controller. The DPST relay needs a
small delay (two/tenths of a second) to prevent the REGEN from
turning on before the contactors have finished switching (Fig 3a).
Switching the contactors under load will cause them to melt down
and could ruin the controller with high voltage spikes.

The field excitation circuit, also known as a tickler, is necessary to
insure that there is enough field strength available to start the
motor generating (Fig 3b and c). I found that 10 Amps at a 1/2
second delay was plenty. Next time, I'll try lower amps for 1/4
second to reduce arcing of the brushes during REGEN initiation.

After testing, I re-potted the controller with about 100 grams of
potting compound. It doesn't take much corrosion to cause
problems at these voltages, so don't skip this step.

Not Finished Yet

Testing was performed around Lopez Island which has many
small hills. The system seemed to work very well. I showed a
10% increase in range with REGEN . Motor current was set at 230
amps maximum and battery current would go as high as 90 amps
usually sitting around 60 amps.

Unfortunately, all was not well at the brushes. The motor brush
temperatures were getting much too hot. 180°C is normal. During
REGEN, we would exceed the 230°C maximum (measured from a
thermocouple inserted 1/2 inch into a positive brush). With the
vehicle on a lift, we could watch the motor during REGEN. Sparks
flew 6 inches off the brushes! We tried zeroing the advance on the
brushes, as suggested by the folks at Advanced DC. There was no
perceived benefit. The project was shelved for a couple months
while I worked on other projects and thought about possible

Too Conservative

One night, as I reviewed the wiring diagrams for possible clues, I
noticed that I had hooked up the A2 terminal on the controller
(diode to B+) to the S2 motor terminal. This had not been shown
in the Design News-Cableform article. Instead, I had copied this
connection from the Lead Sled's controller circuit. Up to this point,
I had assumed it was there to protect the controller from high
voltage spikes that might be caused by "contact bounce" at the
contactors. It had also worked on my original test vehicle with no
problems. With the higher voltages of the Samurai's system, I
wondered if the diode was turning on and diverting current past
the field during flyback. This would result in a weaker field and,
hence, extreme arcing at the brushes. I decided to test this
hypothesis by removing the diode and reverting back to the
simpler Cableform circuit. The final circuit high-voltage circuit
layout is shown in Fig 4.

More Testing

Back in Washington, I hooked up the oscilloscope and brush
temperature meter, removed the A2 connection, and tried it out.
The Samurai EV's owner lives on a smaller island than the original
testing area, so I was limited to alternating between full throttle
and full REGEN around the 3- mile main road. This kind of
"testing" maximizes the current through the motor and controller.
After one loop around the island, the controller was overheating
despite its cooling fan. It had worked! The commutator no
longer turned into a fireball. The highest brush temperatures I
could obtain were 200°c. The brush block was still set at zero
advance but I suspect that even this is not necessary. After the
excitation circuit turned off, I noticed no sign of unusual arcing.

Still To Do

REGEN on the Samurai works well now. Still, I feel that the circuit
is not finished. The inherent roughness of the controller's current
limit causes the vehicle to stutter during the transition from a
large pulse width to continuous operation at very low speeds. For
this reason, I have been activating REGEN with a pull trigger on
the shift lever. Eventually, I would like the system to work
"transparently". That is, there should be no odd little quirks or
special procedures to do. The circuit could be designed to work
off the accelerator pedal or brake pedals. This might be smoothed
out inside the controller itself. If that doesn't prove easy, I'm
thinking of building a dedicated REGEN current limit with a shunt
outside the controller.

I'm guessing that conversion efficiencies are about 50%. This gets
worse at high motor amps. I suspect the inductor component of
the motor is too small and may require another inductor in series.
Higher switching frequencies are another avenue of research. I'll
need a better oscilloscope for this.


The system described adds about $350 in parts to the price of an
EV. The SPDT contactor, at $150, is the largest expense. At $50, the
diode comes in second . Labor is difficult to judge because the
system is still experimental and requires adjustment. I estimate
that the cost to convert an EV to include REGEN should be under
$1500 parts and labor.


REGEN is possible with series motors. Range increases from 8 to 15
percent have been reported. Savings in brake work alone could
pay for the cost of adding it to a conversion, especially if you
regularly drive in hilly terrain. At this time the system is still
somewhat experimental. The next step is more testing on a
variety of vehicles to verify its reliability. It would help to look at
the motor current waveforms during these tests. Please contact
me with any experiences you have on this subject.


Many people helped to make this project possible. Bob Schneevies,
for the use of his Freeway EV for initial tests. Marilyn Anderson
and Rachel Adams, for sponsoring the further research on their
Samurai. The students and staff at Lopez High School, for the great
experience of doing the electric conversion with them. The folks at
Advanced DC for their excellent support and for making a great
motor. And all the other people who taught me so much about
controllers and REGEN along the way.

Curtis PMC is a registered trademark of Curtis Instruments.
Advanced DC is a registered trademark of Advanced DC Motors

Potting compound Part # 2419 Call For local distributor
Techform Laboratories Inc.
2021 N. Glassell street
Orange, CA 92665
(714) 921 9054

SPDT Contactors, Use half of a reversing set or special order a
single one.
KTA Services, Ken Koch,
944 W 21st ST.
Upland, CA 91786
(909) 949 7914

DPST relay with magnetic blowout for field excitement. Also good
for heater switching. Contacts: 125 VDC at 20 amps. 12 VDC coil
#RL9101 $7.95 each.
C&H sales company
(800) 325 9465
Out of stock

REGEN Diode 1N4052 275 amps 600 volts
I bought the last surplus one at HDB. They may be able to special
order a new one or try other surplus houses.
2860 Spring street
Redwood City, CA 94063
(650) 368 1388