'More on Driving Motors'
Well, I spent the weekend doing what I probably should have been doing right
from the start, trying to understand how Motors work under Digital Control
and understanding how the transistors work when they are driving motors.
Going right from the start, I measured two AA Cells in series at 2.9 Volts
(which is what I want to use for driving the motors). Maximum (Short
Circuit) current was 750 mA for Alkiline Cells and 7.6 Amps for NiCads.
I then looked at the motors and found that they run with no load at 160 mA
with a stalled current of 450 mA. When the motors are in the Tamiya "Tank"
application, they require between 250 mA and 400 mA when running (according
to my DMM).
I tried to characterize the noise produced by the motors, but didn't have a
lot of luck, I will be putting a Resistor and Capacitor (in Series) across
the motor terminals, does anybody have any recommendations on what values
should be used? The PIC does seem to run fine with the motors running with
the negative voltage common to the PIC and the motors.
In terms of the LM1111 Step-Up Power supply, I found that it produced a 200
mV noise spike on Gnd and a 500 mV spike ( very fast rising edge with
something that looks like a R/C decay on the downward side). As I've
indicated, the noise produced by both the Step-Up Power Supply and the
motors does not seem to bother the PIC when the negative voltages (grounds)
In characterizing the Transistor controls, I found that using 2N7000
N-Channel FETs (in a TO-92 package) would give me a 2.7 Volts for the Motor.
Using the 2N4403/2N4401 as suggested in the article gave me 1.8Volts for the
Now on a proto-board, I then tried running a motor. With the Bipolar
(2N4403/2N4401) combination, I could *never* get the motor to run. I found
that the PNP 2N4403s could source 300 mA of current, but the NPN 2N4401s
that I had couldn't source more than 100 mA of current. The 2N4403s could
also drive the motor when it was put back into the Tamiya "Tank" kit (with
Gearbox and drive motors). On both devices, when tested on my DMM, I got a
Hfe of approximately 150.
Next, I then used the 2N7000 N-Channel FETs. In the catalog and supplier I
ordered the parts from (Electrosonic, here in Toronto), the FETs are rated
at 500 mA. Using two FETS in a totem-pole arrangement to drive the unloaded
motor worked fine, but when I put the Motor back into the "Tank" and drove
the application, the motor worked for a few seconds until the transistor at
the "top" (between the motor and the positive voltage) literally exploded.
I tried it again and terminated the experiment when the replacement
transistor got hot (about a second after 5 Volts was applied to the gate).
Looking in the Digi-Key catalog, I see that these devices are rated for only
One interesting thing with the FETs, when I first put them into the circuit
(with the gates floating), the motor started running (ie current was going
through the FETs) at about half the speed they would be when hooked up to
the batteries directly. When I put the gates on 5 Volts the motor would run
at full speed, and when the gates were at 0 Volts, the motors would stop.
Interestingly enough, when I disconnected the gate from the voltage rails,
the motors would continue operating as if the voltage (or lack of) was still
Where am I now? Well, I think I will be going with the FETs (once I find a
truly 1 Amp capable device) - not requiring the load resistor is an
advantage for me (less wiring in an application that is hand-wired). I'm
curious to find out why the 2N4401s can only source/sink so little current
although I'm not going to knock myself out, if anybody knows why, please
post it here.
I'm also surprised at the number of different parameters different catalogs
give for different devices. For the 2N440x, I have two different catalogs
giving a HFe of "at least" 100 and "nominal" or 200 as well as a current
rating of 600 mA and 500 mA for both. For the 2N7000, I have a current
rating of 500 mA and 300 mA. Does anybody have any comments on this.
As for the original article (the "Runabout Robot"), I realize that I should
have known better (especially in light of some of my previous experiences)
to take it at face value, but I did because of my limited experience in
driving motors (I've now got a whole bunch more). With this project, I have
not been able to locate the specified Step Up Voltage convertor (the Maxim
756), I have not been able to get the code to assemble or run (It has
Microchip mnemonics, but won't run under MPASM without a LOT of
modifications and once I got it to run, I couldn't get it to read I/R data
at all), and now I have had to redesign the motor drive section.
What is that latin phrase? "Caveat Emptor?"
Once I get some 1 Amp capable FETs and try them out I'll let you know what's
Thanx for all the replies and suggestions,
Do you ever feel like an XT Clone caught in the Pentium Pro Zone?
Mark K Sullivan
1 Amp FETS
Try IRF7201 or 7101 and Zetex ZVN4306A, both available from Digi-Key. I've had
good luck with both, driven direct from PIC I/O. I can't suggest complements,
though. I have only used the N-channel ones.
The 2N7000 is probably too small. Your immediate problem wasn't too little
current capacity, though, but too little power handling capability. It exploded
because it did not have enough gate voltage (relative to source). I assume it
saw 4.8V-3V where the bottom one saw 4.8V. With less gate voltage than required
for minimum on resistance (equivalent to saturation of a bipolar), it is a
resistor and heats up according to I*I*R. This is the advantage of a
complementary drive circuit.
Two NiCds are only going to give you 2.4 V nominal, and less in practice. It
will be real hard to make this work. NiCds have flatter discharge, though.
This means less loss of terminal voltage toward end of discharge cycle. You
should be thinking about what happens to the battery voltage as the batteries
run down. By the way, this would all be much easier with 4 NiCd or 3 alkaline
Try 4.7 to 10 ohm in series with .1uF. Keep leads short or it won't do any
good. Use stacked film cap or monolithic ceramic.
200mV on the power supply is not too bad but 500 or 700 (between rails) is
marginal. The type of capacitor you use on the output of the step-up is
important. It should have low ESR or low inductance. Also, paralleling smaller
value capacitors is better than one large one. You may just need fatter and/or
shorter power conductors.
Motor runs with FET gate open.
Unlike bipolar transistors which (more or less) operate from a current into the
base and therefore will (more or less) turn off when unconnected, FETs are
controlled by the voltage on the gate. The gate is like a capacitor. What you
saw is that, once you charge up the capacitor, it stays charged (and therefore
turned on) until the charge can leak away. When you first hooked up the FETs,
they had a residual charge on the gate, mayeb just from being handled.
Martin J. Maney
On Mon, 15 Jul 1996, myke predko wrote:
> I then looked at the motors and found that they run with no load at 160 mA
> with a stalled current of 450 mA. When the motors are in the Tamiya "Tank"
> application, they require between 250 mA and 400 mA when running (according
> to my DMM).
> Using the 2N4403/2N4401 as suggested in the article gave me 1.8Volts for the
Without a source of base current on the order of 1/10 the required
collector current, bipolar transistors are not going to be operating in
saturation. A monolithic darlington would make the design easy but with
somewhat higher voltage drop; OTOH, it wouldn't "waste" the output's base
current. Either approach could be made to work with some small pros &
cons, but with a 450 mA stall current I'd suggest the 4401/03 are a bit
marginal. I'd probably reach for a TIP29/30 for this - they're cheap and
readily available (okay, I almost certainly have a couple in the parts
bin, and this does influsence me <grin>).
> Gearbox and drive motors). On both devices, when tested on my DMM, I got a
> Hfe of approximately 150.
That Hfe measurement is meaningless for use as a power switch - you're
measuring it at a very low current and far from saturation. The actual
beta will be lower when (trying) to drive the motors both because of the
much higher collector current and the lower (ideally under 700mV) C-E
> motor worked fine, but when I put the Motor back into the "Tank" and drove
> the application, the motor worked for a few seconds until the transistor at
> the "top" (between the motor and the positive voltage) literally exploded.
> I tried it again and terminated the experiment when the replacement
> transistor got hot (about a second after 5 Volts was applied to the gate).
> Looking in the Digi-Key catalog, I see that these devices are rated for only
> 300 mA.
It could be that these particular FETs are underrated for this
application, but the immediate cause of the high-side deriver's
destruction is almost certainly that the gate drive voltage wasn't high
enough. This has a similar effect to insufficent base drive in bipolars:
the device isn't solidly turned on, so it drops more voltage and
therefore disspiates more power than it needs to - or, in this case, than
it can stand to!
> I'm also surprised at the number of different parameters different catalogs
> give for different devices. For the 2N440x, I have two different catalogs
> giving a HFe of "at least" 100 and "nominal" or 200 as well as a current
> rating of 600 mA and 500 mA for both. For the 2N7000, I have a current
> rating of 500 mA and 300 mA. Does anybody have any comments on this.
These are loosely-spec'd parts to begin with. The only things that are
actually in disagreement in the above list are the current ratings, and
those could be due to the rating being at different conditions. One
might be an absolute maximum and the other a reccomended design maximum, etc.
Thanx for your very detailed reply, it really helped me understand what was
going on. I'm going to try complementary pair FETs and see how that works
(mostly for my own curiosity).
>200mV on the power supply is not too bad but 500 or 700 (between rails) is
>marginal. The type of capacitor you use on the output of the step-up is
>important. It should have low ESR or low inductance. Also, paralleling
>value capacitors is better than one large one. You may just need fatter and/or
>shorter power conductors.
I don't think I explained what was happening here properly. The Step-Up
Power Supply is providing a solid 5 volts, the spike is going from 5 to 5.5
Volts and looks like:
on the scope (hopefully the ASCII art isn't too bad). At the same time as
the spike, there is a 200 mV spike on the "Ground". I don't know how much
this affects things; as I've said, the PIC currently runs very solidly with
the two I/R Receivers and the motor running (to put noise in the system).
Thanx again for your answers,
Do you ever feel like an XT Clone caught in the Pentium Pro Zone?
At 05:16 PM 15/07/96 EDT, myke predko wrote:
>Power Supply is providing a solid 5 volts, the spike is going from 5 to 5.5
>Volts and looks like:
>on the scope (hopefully the ASCII art isn't too bad). At the same time as
>the spike, there is a 200 mV spike on the "Ground". I don't know how much
>this affects things; as I've said, the PIC currently runs very solidly with
>the two I/R Receivers and the motor running (to put noise in the system).
---- and also----
>The PIC does seem to run fine with the motors running with
>the negative voltage common to the PIC and the motors.
Can you explain how the power supplies to the motors and PIC are organised
and connected? Are both supplies referenced and tied to the same common (-)
rail? It would also be helpfull if you can indicate where in this setup
your scope was placed, specially the ground reference point for the scope.
It appears you may be seeing some induced emf in the ground plane, or I^2xR
losses, caused by the high current drain of the motors.
I haven't tried this, but if you are getting insufficent gate drive on
a FET, this could be increased by use of a small cell (say silver
oxide) to boost the applied voltage in the right direction. The cell
should last for its entire shelf life. The small charge/discharge
spikes as the FET gate is discharged/charged should not be a problem.
Another untested option is to use a charge pump to charge a capacitance
connected between the driver and the gate. This requires an initial
'priming' of the charge at power up, and the occasional transition to
maintain the charge:
C1 --- Isink
| | ||--
| N-1 diodes ||--
--- C2 |
( >| represents a diode with drop of 0.5V)
In general, V(C1) will be Drvr(hi) - Drvr(lo) - N*0.5 where N is the
total diode string length. The FET must be chosen so that the
available gate swing is enough to bias the fet fully off and on.
The capacitance should be chosen so C(FET) << C1 << C2.
A similar network is used for the high side (P-channel) device.
On the subject of charge pumps, if you have a spare PIC pin, you could
use it to drive a charge pump (3 stage, to get 12-14V) or even a tiny
SMPS - much like what your LT1111 is doing. This supply is then used
to drive the gates via a suitable level shifting transistor. Yes, this
_is_ totally off the planet, but no-one ever said motor control was
easy - look at the prices of motor control chips!
Steve Hardy wrote:
I particularly like the ingenuity of this approach - but, with due
deference to Steve, I suspect the above circuit will be slow in turning the
motor OFF and that discharge resistors (R1,R2, below) will be needed, plus
more regular pulsing to maintain the ON voltage. A capacitor from gate to
source *might* also be necessary.
C1 --- Isink
| | ||--
| | \ G||--
--- \ / |S
C2 --- R1/ R2\ |
| \ | |
| | | |
Lots of scope for experimentation here!
Murray McGregor | (o)(o)
Education Department C _) ZZ
University of Otago | ,___| ZZ |\ _,,,---,
Dunedin, New Zealand. / / zz /,`.-'`' -. ;_.
Ph (64)(3)479-8801 /____\ |,4- ) )-,_. ,\ (-.:,_
|> From: Murray McGregor <stonebow.otago.ac.nz>
> Steve Hardy wrote:
> >Another untested option is to use a charge pump to charge a capacitance
> >connected between the driver and the gate. This requires an initial
> >'priming' of the charge at power up, and the occasional transition to
> >maintain the charge:
> I particularly like the ingenuity of this approach - but, with due
> deference to Steve, I suspect the above circuit will be slow in turning the
> motor OFF and that discharge resistors (R1,R2, below) will be needed, plus
> more regular pulsing to maintain the ON voltage. A capacitor from gate to
> source *might* also be necessary.
At the risk of getting a bit off-topic, I should have made it clearer
that the OFF state of the FET is achieved at Drvr(low)+boost voltage
(2V in the example). The FET will switch just as fast as the driver
can drive its gate capacitance, so long as C1 >> C(fet). It does not
depend on leakage through a resistor to switch off, especially since
any FET worth its silicon will retain its gate charge for at least a
few minutes if open circuit. Most VMOS N-ch fets will be off if their
gate voltage is <= 2-3V with respect to source.
A cap from G-S would worsen a bad situation, since the C looking into
the gate is already multiplied by the Miller effect in common source
By the way, a string of 3 silicon diodes could be replaced with one
LED for the purpose of voltage drop.
|> I spent [time] trying to understand how Motors work
> I then looked at the motors and found that they run with no load at
> 160 mA with a stalled current of 450 mA. When the motors are in the
> Tamiya "Tank" application, they require between 250 mA and 400 mA
> when running (according to my DMM).
Good start. When you say "Tamiya Tank", you must mean the little
toy tracked vehicle with a single gear box -- right?o
Tamiya also has 3 different 1/16 tank models (as I recall) for radio
control with (I think) dual motors, seperate gear boxes, and tracks
with individual links. They're about $300 for each model. But these
motors draw 5-20 amps at 7VDC -- in other words, small fractional
hoursepower in a 1" cylinder that's 2" long. Electronic speed controls
(ESC) for them use multiple power FETs. Check the R/C car magazines
for lots of ideas. R/C hobby shops are a convenient source of parts.
> I tried to characterize the noise produced by the motors, but didn't
> have a lot of luck
I've worked with and read about electronic speed controls for R/C cars.
All require capacitors on the motor. Usually the caps are supplied by
the ESC vendor, so I don't know the size. In the one ESC instruction
sheet I have handy, it just states "monolithic capacitors" between each
motor terminal and the metal motor can. From the physical size, 0.1uF
ceramic looks about right.
> 2N4403/2N4401 transistors [...] On both devices, when tested on my
> DMM, I got a Hfe of approximately 150.
At what current? If measured on a handheld DMM, I'd bet it's at very
low amperage. Measure base current and collector current when it's
driving the motor (under load, make it generate some torque), do the
division, and get a real beta. Repeat for several sets for range.
Then derate somewhat to guarantee saturation.
There's been another post on this (transistor gain under load).
> I'm also surprised at the number of different parameters different
> catalogs give for different devices.
Go get the manufacturers data sheets. Then use the worst case data
(not the nominal values).
Jones Computer Communications frumble.claremont.edulee
509 Black Hills Dr, Claremont, CA 91711 voice: 909-621-9008
More... (looser matching)
- Last day of these posts
- In 1996
, 1997 only
- New search...