Powering older locos with PWM

[passenger1955]

I am experimenting with running various conventional HO locomotives with PWM (instead of variable direct current). I notice relatively new locomotives run pretty well with PWM. The older locomotives are hit and miss. Some do not run so well when I give them PWM power (yet run ok when operated with DC). I have a 20 year old Bachmann 0-4-0, and it runs ok on 450 Hz PWM, but runs poorly when given 16 kHz PWM. I have a 40 year old SD-70 Blue/Yellow Santa-Fe (made in Yugoslavia) that jiggles with both 450 hz or 16 kHz PWM, but runs so-so with DC power.

I am trying to understand if there are certain classes or types of motors that run poorly with pulse-width modulation. Are there any guidelines to help me understand "that type of motor probably won't run well with PWM" as opposed to what type will?

[rrinker]

That old Bachmann, does it have one of those flat pancake motors? 3 pole motors will PROBABLY work better with lower PWM frequency, whereas can motors and 5 pole and up (not very common - but back in the day there were a few 7 pole motors available) will probably work better with higher frequency PWM.

                         --Randy

[gregc]

passenger1955

Some do not run so well when I give them PWM power (yet run ok when operated with DC).

can you describe what you mean?   do they run well enough at high throttle settings?   are they jerky?  don't run well a low throttle settings?  don't run at all?   tend to stall?   does it jiggle at all throttle settings?

i had thought, based on the Tat-IV articles, that adding pulses to low voltage DC overcomes some mechanical resistance of some motors, improving performance.  Do some motors, want to rest at certain positions and need a pulse of energy to get them "over this hump"?   Would this explain why longer pulses, even at lower duty cycles at lower frequency, are better in this case?

[rrinker]

 Think about something like a 3 pole motor vs a 5 pole. Especially with the pancake type, each pole is a rather alrge part of the arc of motor rotation. At high PWM frequencies, this will result in several pulses before the next commutator segment even engages. And since at low speed the pulses are very narrow, you aren't imparting a lot of energy - many DCC decoders have a 'torque compensation' mode that allows extra pulses at the low end to compensate for this lack of power at low speeds. As speed increases, the shorter, faster pulses are fine to swing the motor between poles. Sort of the opposite takes place on a motor with more poles. The segments are closer together, so the low frequency pulses at slow speed are enough to turn the motor, but at high speed, the now widened pulse at low frequency may be too long.

 The inductance of the motor also need to be taken into account, and with an infinitely variable pulse frequency you just might be able to tune in a resonant frequency and block most of the power. Sort of like when putting a good DCC decoder in a Bachmann loco without taking out the capacitors. The LC circuit formed is a low pass filter hoping to block out RF, but it also attenuates the pulse frequency of the good 'silent' type decoders. Remove the caps (easiest, since they parallel the motor) and they run much better.

 All motors need something to "get over the hump" - think about how they work, at least a DC permag motor. You energize a given winding and it is attracted to the opposite magnetic pole - then you need to cut off the power to that coil or else the motor will just stall with the energized pole as close to the permanent magnet pole as it can get.  There are other factors like residual magnetism and the decay time of the pole field that affect just how the commutator is timed. For optimum results across a wide range of speeds, ideally the timing needs to vary, but since that's not practical, it's optimized for some RPM. At lower RPM, it's cutting out a bit too soon and the cogging action of the armature passing the pole slows it back down. A strong pulse introduced on top of the smooth DC give the pole a higher power 'kick' to make it past the permag pole. Properly tuned, instead of just simple 60Hz pulses, can make that pulse behave as if the armature pole had power close to the crossing point. Each motor is differnet, unless you have all the same brnad and model motor in every loco, in which case it just comes down to manufacturing tolerances. That's why a throttle like the TAT-IV had all the pulse adjustments, and I think it was the TAT-V that took it a step further and had a DIP header in which you soldered resistors equating to the control potentiometer positions so instead of fiddling around each time you operated a different loco, you could just plug in the header with the settings already configured for the particular loco.

                         --Randy

 

[passenger1955]

Great information. I'm blown away what a great resource it is to be able to post questions and have people explain it to you. Ten years ago this never would have happened. Thanks so much all. Someone pointed out my diesel couldn't be an SD70 because of its age. I guess it's not an SD70 - I just called it that because of its length.

The "poor performance" I got on certain locos is jerky motion (as if the PID was incorrectly tuned on a modern loco). Like I said, this only occured on 2 of my locos, and switching to 450 PWM cleared the jerkiness up on one of them.

Another issue you get on all the locos (even if they run well with pwm) is the cab light's flicker as they are turning on at low speed (which I guess should be expected). And when you slow down and your speed hits 0 the lights turn off abruptly. It seems more natural when you use a DC transformer and the lights fade on and off.

 

When you factor in how the lights act, I'm getting the impression that running a conventional loco by applying pulse-width-modulation to the track is generally less desireable than giving them variable DC.

Is that a reasonable generality?

[rrinker]

 With a proper pulse frequency and amplitude, there should be better low speed performance than with pure DC, at least when it comes to smooth starts.

 The peak pulse is something you cn take advantage of to get constantly lighting. Many years ago I built a ditch light circuit for a fellow club member that used a capcitor to store up the pulses and keep the circuit operating - it worked on our one power supply which used pulses, it did not work on the other one which used smooth DC. Same thing could be used to make a constantly headlight - crack the throttle but before the motor can turn, the lights come on, and not blinking (they blink because the PWM frequency you are using is slow enough to see, despite the persistence of an incandescent bulb). I don't have a schematic, I'd have to dig around, but if the equivalent circuit exists anywhere it would probably be on Rob Paisley's site.

                  --Randy

 

[DigitalGriffin]

oh boy...well...hmmm how to explain this.

Some engines work better with premium gas, some work better with regular.

Edit: oh and +1 on what randy said.  He always beats me to it! :-)

Those older motors tend to suck up a lot of juice before they would get moving.  And it took a lot of juice to overcome the motors internal resistence.  (Magnetic field flux change (results in heat) and inertial resistence)  And some just had plain old low resistence which creates it's own issues.  So older motors not only ate up more current and moved less, but generated more heat.  (Especially true of open motors or 3 pole pancake motors)

And to make matters worse, the higher the frequency, the worse the low speed response is.

Of all the decoders out there I firmly believe Lenz gold series is the best when handling "older" motors.  They have a CV which allows you to specify what type of motor you have installed so it will adjust it's feedback!