It depends on what vintage selector you have, but usually worn/stretched/stiff chain, pawl magnet core, too much or not enough oil on the slide, compounding, air gap on u and d and h relays, leveling speeds, temp, moisture, load, could be any or all of those things.

matthew, of all the things that greenpants has indicated, compounding is the most important, and the first thing that would need to be addressed before continuing with the other situations. I usually put car on inspection, and run the car by simulating the car running off the pie plate, in level speed. You just block out the associated relays in the generator leveling field circuit, and run the car up and down and make sure the level speed is within a couple of feet per minute. if they are way out, say 15 fpm up and 5 fpm down, then you need to recompound generator. if they are nearly equal, then i would look at making sure SF relay is picking up during leveling. if that is good then your lv ring brushes would be next, put new leveling brushes on if they are severely worn. as green pants indicates, load could also be a problem, if you have load weighing switches that are not indicating proper load, the the car would not approach the floor at the appropriate speed. If you have an otis controller that uses compensating fields, ie, it has HC,LC,RC,TC relays then you must set car speed at each load increment to obtain proper leveling. If you dont have a compensating field, then load weighing switches usually only affect hi speed up and down, usually the relays are PL, or NL, or ECU, ECD. Positive load, negative load, empty car up, empty car down. these wont usually affect leveling, unless they are way out of tolerance, say more then 50 or so fpm. usually when this circuit fails, you will trip the governor. any way after all that rambling, i would bet your compounding is off so start there.

One area people always forget about, check and tighten all your connections.

Check brush spring tension and verify that brush spring arms are not bottoming out on brush holder.

If it has a tape selector, make sure you have not skipped a hole or that the selector is properly set in the correct holes.

All good recommends thanks guys. I bet its probably gotta be recompounded. Thats probably the only thing that hasn't been checked.

A dry slide could throw out the leveling?

If this is the vintage that the selector tape is driven by the sheave it is very important to make sure the sheave has some lubracation. As the tape rides higher on the sheave than the cables it will travel farther than the car if no slippage is present. And wait till you do a inspection with a safety set and want to see loss of traction. the spring on the counterweight side can streach several feet!!!!

just a side note i would also check that the proper grade of brush is in the genny as some gennys are very sensitive to this

Hello Fella's... New guy here

Another thing you might check. (easier if you have the print) Some Otis relay controllers have a soft set brake. When the car lose direction the relays holding the brake up drop out. There is a capacitor on the board that will discharge into the brake coil allowing it to set softly about an inch later. The sympton is a car stopping off level an inch, or a half inch in both directions. It may look like two worn brush's on the selector.

Good Luck
Spuk!

Before you start doing a bunch of adjustments trying to get your compounding straight , check your loop circuit for grounds. Verify your generator and motor are not severely grounded. It would be a good time to shut it down and meg everything out. Compounding is a severe factor, but you'll never get it stable with a grounded loop circuit.

Lot of ideas here...

You might compare your motor field voltage, and generator field voltage, and armature voltage with a good running car, to see if anything's amiss.

Is it always long in the up, short in the down? (empty car) Could be weak compounding. Or overly strong compounding, if opposite. (Long down, short up)

Does this selector have an advancer panel, +SLS cams, to drop out the speed switches on slowdown? If there is no consistent pattern, and it's just "all over the place", it may be suffering from years of people shooting from the hip, bending pie plates, moving LV ring brushes around indiscriminately, etc. You might uncouple the tape sheave drive, and rotate the gear drive into the middle floor in both directions, counting the degrees on the vernier, as each SLS switch opens up. The actuation of switches, brushes in each direction should be symeterical in both the up, and the down, generaly. The floorbars themselves could even be in the wrong position, affecting the GK wing and the motor field strengthening in slowdown. I always like to compare to a good car.