precision turntable optical alignment?

[RR_Mel]

The hardest part of getting my turntable to work was the hole alignment.  Drilling seven 1/16” holes (super happy I didn’t have more tracks) with the precision alignment in the pit wall wasn’t easy.  Next was prefect alignment with the sensors on the bridge.  Both had to be perfectly centered between the rails as well as the vertical position to match the pit wall holes.

 

Because I paralleled the sensors both needed to be a matched pair.  Just that took a good amount of experimenting.  I used a Rob Paisley circuit for the optical detection, the sensitivity is adjustable and once set many years ago I’ve never had to readjust it.  I had experienced some problems earlier on using IR detection from interference with the fluorescent lighting so I put a Styrene tube shield in front of the sensors and painted the tube and sensor flat black to prevent leaks.

 

I used P&B 17 series 12 volt 4PDT relays because I had bunches of them with sockets.  I came across a full box (50) of them new in the 70s.

 

Originally I used a 12 position rotary switch to select the track emitter but last year I bought a 24 position rotary switch to replace it.  Both have a stop washer for 8 positions, off and seven tracks.  The 24 position switch looks better on my control panel, 15° spacing.

 

It took what seemed like for ever to get the bridge to cooperate; many take it out and tweak something.  I was determined to have a perfect operating turntable.  This wasn’t my first attempt at building a turntable. I scratch built many over 30 or so years, all had problems.  The only reason I succeeded with this one was that the CMR Kit is the very best kit I ever put together.  The kit came with the Dayton Gear Motor and had I known how nice that motor worked earlier one of my scratch built may have worked.

 

It’s been quite awhile since I put everything together (2008) but I do remember being very frustrated and doing constant redos.  Another tuffy was the sliprings.  I went through a dozen or more 1/16” bits drilling the holes for the wires in the copper washers.  The bits would grab the copper and break.  I made a jig to hold the washers and used my Craftsman bench top drill press on the slowest speed.

 

Once the slipring was assembled it worked perfect first time and I’ve never had a single problem with conductivity to my bridge.

 

The CMR Kit came with a very nice pit rail power pickup system but because of my previous problem trying to use the pit rail for power I built my own system.  I do have to be conscious about track polarity but it is worth having to pop a toggle now and then to have perfect power to the rails.

 

Looking back I’m sure glad I didn’t wait to complete my turntable, there is now way I could do that now.  It is the nicest project I’ve ever done in my 68 year HO model railroading career. 

 

 

 

 

Mel

 

 

My Model Railroad   

http://melvineperry.blogspot.com/

 

Bakersfield, California

 

I'm beginning to realize that aging is not for wimps.

 

[rrinker]

 The response curve of the photsensor is distinctly non-linear. That's how you can get an accuracy greater than the physical size of the light tube. The cut out circuit looks for a threshold, not just "stop as soon as I see ANY light". If it did that, then indeed the best you could achieve would be the width of the tube, but if the circuit discriminates more, it's not just the tube diameter, but where within the tube diameter the sensor is. A peak detector comes to mind but then it has to see at least one drop from absolute peak before it could tell it has passed the peak point. But even that would be very consistent. Accuracy would be based on the tolerances of the electronic components and the response curve of the detector. You can see in the chart how steeply the response drops off as you move off the 0 angle. And that is just for that one specific part. Check data sheets, choose wisely.

 At least as far as the circuit commanding the motor to stop - there's still the mechnical tolerances in the drive train.

 The decidely non-scientific answer is - it works, we know because of empirical proof. No reason to believe Mel is not telling the truth that his works, but there are also commercial systems that clearly work. I may be way off with the idea that they might use peak detection, but I think that's one way it COULD work.

                                      --Randy

 

                               

[gregc]

RR_Mel

The accuracy is less than 1/64” from either direction every time.

i'm not doubting this, but i don't exactly understand how using small 3mm (1/8") components can be this accurate.

the beam from an LED emitter must spread and a phototransistor must be sensitive to a beam at a slight angle.  The angular displacement chart below indicates that a phototransistor sensititvity is still 90% at 5 degree angle.   I assume similar for an LED viewed at an angle

seems that the aperature to both the LED and phototransistor need to be restricted.

why is the accuracy using the 1/16" tubes < 1/16"?

[RR_Mel]

The Dayton motor isn’t a stepper motor.  I use optical positioning for my bridge.  I select the emitter in the pit wall for the track I want the bridge to stop at.  I use a rotary switch to select the emitter and press the operate toggle (momentary), that latches the motor relay.  The direction of rotation is done with a DPDT reversing toggle.  The IR sensor trips a second (stop) relay releasing the latched motor relay shorting the motor (break).  The track alignment is close enough that I didn’t have to bevel the inside of the rails.

 

The Dayton motor is super powerful!  50 in-lbs torque.  The 12 volt motor will turn my bridge down to 6 volts for super slow rotation.  I run the motor at 12 volts for the max rotation of a ½ RPM.

 

I made a slip clutch shaft connector between the motor shaft and the bridge shaft.  Without the slip clutch if something was to stop the bridge from turning there would be heavy damage to something.

 

I’ve been building turntables for well over 50 years and this 135’ CRM turntable is the best I’ve ever run into.  The Dayton gear motor is what makes the difference between a puny turntable and the very best.

 

Doing my thing with the turntable wasn’t an easy task.  I went all out in the construction.  I had problems with every turntable I built from the mid 60s to this one in 2008.  I actually started building the kit in 2000, my workload took its toll on the project but over the eight years I made a lot of Mel mods to it before I installed it in 2008.  The bridge has 1½ volt 1mm bulbs for lighting and the two IR sensors for indexing.  That is connected through a Mel slipring on the shaft.  The rail power is done through an overhead scale power drop for one rail and the other rail is powered from the motor shaft.  I do not use the pit rail for power, that has always been the biggest problem with all of my turntables in the past.

 

 

 

 

Mel

 

 

My Model Railroad   

http://melvineperry.blogspot.com/

 

Bakersfield, California

 

I'm beginning to realize that aging is not for wimps.

 

[gregc]

when calibrating to a zero position, there's no need to stop the motor.      Calibration can occur whenever the bridge passes across the calibration point during normal operation, should there be any slippage.   

This isn't what Mel did.   But his motor turns at 1/2 rpm (~0.5"/sec for 130' bridge). 

Gearing and load on the bridge may provide sufficient friction to stop a DC motor quickly.

[rrinker]

 That's another key element - the shorting of the motor when power is cut. Otherwise there may be considerable drift. At least if a motor with encoder or other drive like that is used. A stepper, if you stop stepping, will just stop.

 I had a great loco to demonstrate that efffect - the Bowser Baldwin switchers with their Cannon can motors. On DC< if the power was just disconnected from full throttle, they would travel quite a distance on jus thte tiny flywheel momentum, and the motor BEMF would keep the headlight LED lit. Short the track and it woudl stop like you threw a chain around it. Some varying resistance and you could control how fast it stopped. The only thing I've ever seen coast futher are old Lionels with spur gear drive and no worms.

                                  --Randy

 

[RR_Mel]

I did the tubing thing to prevent other IR sources from interfering with it.

 

 

You can tweak the sensitivity of the sensor to adjust the stop.  I ordered a handful of sensors and parallel them, I piddled around and matched a pair pretty close so tweaking the sensitivity works pretty good.

 

I’m really impressed with the Arduino FC-51 IR package.  I pulled the emitter on one board and remoted it’s good for 6’ for break beam detection.  If I was starting over I would use a FC-51 for each inbound track and put the emitters on the bridge.  Then I could tweak each track sensor individually.

 

But like I said it’s working great so I’m not changing anything on mine.  

 

 

 

Mel

 

 

My Model Railroad   

http://melvineperry.blogspot.com/

 

Bakersfield, California

 

I'm beginning to realize that aging is not for wimps.

 

[gregc]

RR_Mel

IR sensors centered under each end of the bridge with a 1/16” ID  x ⅜” long piece of Styrene tubing in front of the sensor.

i think the tubing is a good idea.   what made you think of it?

if the tubing were mounted at a slight angle, it could restrict the effective opening to much less than 1/16"

[RR_Mel]

I use 1/16” holes in the wall of the pit with an IR emitter behind it.  I positioned a pair of IR sensors centered under each end of the bridge with a 1/16” ID  x ⅜” long piece of Styrene tubing in front of the sensor.  I painted the tubing and sensors flat black to prevent leakage.

 

To drill the holes in the pit wall I scored the wall at 1” from the bottom of the pit then drilled the holes centered under each incoming track.  After assembly I stopped the bridge at each track hole then aligned each track to mate with the bridge track.

 

The accuracy is less than 1/64” from either direction every time.

 

When the sensor sees the emitter it drops the motor relay stopping the bridge.  I wired the Dayton 2L003 12 volt DC motor DPDT relay to short the motor in the off state as a dynamic break stopping the ½ RPM (7189:1) gear motor instantly.

 

If I was to do it over I would reverse the emitter and sensor using the cheapo Arduino FC-51 modules.  It has worked flawlessly as is for almost 10 years so the “if it ain’t broke don’t fix it” rule works for me.

 

 

 

 

Mel

 

 

My Model Railroad   

http://melvineperry.blogspot.com/

 

Bakersfield, California

 

I'm beginning to realize that aging is not for wimps.

 

[rrinker]

 Seems to work precise enough to Lenz, who makes the electronics in the Walthers turntables. A narrow aperature, coupled with fast enough peak detection, should be plenty accurate.

 I'm wondering if approaching the cal point from both sides, adn then taking an average would better compensate for any slop in the drive mechanism. If you only go from one side, all the accumulated error will be against one side of each gear in the drive, if gears are used. The fewer the better, I think. If you cal from one side, then when the table is turned the opposite direction the bridge position would lag the motor count by whatever the total abount of slop in the gears is. Fewer gears, less slop, less need to cal both directions. Or maybe instead of gears, a cogged belt drive - if not a direct drive. That would be for more accurate, or at least easier to set up and be accurate, than a multiple gear train.

                                           --Randy

 

[gregc]

an indexed turntable system using either a geared stepper or dc motor with encoder needs a reference point on the turntable that indentifies position zero.   I believe this can be an opening in the turntable wall with a phototransistor that detects a light from an LED on the turntable bridge.

i've been wondering how precise that can/needs to be to be accurate to some fraction of the width of a rail.

i'm wondering if the opening in the wall needs to be a thin slot.   is a slot also needed between the led on the bridge.

while the semiconductor junction sensitive to light on the phototransistor must be pretty narrow, i wonder if that and a thin slot are enough to detect the LED at a very precise and repeatable position.

do they always do such calibration by approaching from a specific direction?