Smoothing the Light Dim/Ripple When Throwing O22 Switches

Use an LED and a capacitor and diode to drive the lamp.

How many switches on your layout?  And to be clear, you're referring all the lamps dimming when one switch is thrown?

If the aux power wiring to multiple switches is a daisy-chain, then it seems switches closer to the transformer will see less drop and might actually get brighter if some kind of boost method is used!

Is conversion to LED a viable option?  Note that this would likely require more than swapping out the existing filament bulb for a basic LED equivalent.  The LED must have some kind of voltage regulation or energy storage (capacitor).  Why?  Because you will still see the same voltage drop from the large momentary solenoid current!  So there must be circuitry at each switch which I suppose could be considered a bother to hook up.

If attempting to "boost" the voltage at the transformer (rather than regulate/smooth the voltage at each switch), my guess is converting to all DC operation might make for an easier circuit whether staying with lamp or converting to LED.

stan2004 posted:

How many switches on your layout?  And to be clear, you're referring all the lamps dimming when one switch is thrown?

If the aux power wiring to multiple switches is a daisy-chain, then it seems switches closer to the transformer will see less drop and might actually get brighter if some kind of boost method is used!

Is conversion to LED a viable option?  Note that this would likely require more than swapping out the existing filament bulb for a basic LED equivalent.  The LED must have some kind of voltage regulation or energy storage (capacitor).  Why?  Because you will still see the same voltage drop from the large momentary solenoid current!  So there must be circuitry at each switch which I suppose could be considered a bother to hook up.

If attempting to "boost" the voltage at the transformer (rather than regulate/smooth the voltage at each switch), my guess is converting to all DC operation might make for an easier circuit whether staying with lamp or converting to LED.

Thanks for the thoughts.  With my current "concrete floor empire" I have about a dozen switches - and they are all illuminated with incandescent lamps (although the switch panel uses LEDs).  So the transformer sees about a 1A load in steady-state.  The switches are powered in a semi-daisy chain configuration with a couple of low-gauge feeders which then branch out so the effect is by-my-eyes about equal among all of the switches.

As you point out just converting to drop-in LED lamps might not make much of a difference as the voltage drop upon solenoid activation would still exist - although, perhaps, not as noticeably depending upon where in the LED's operating range the steady-state voltage resides (which is about 13V, at least at the panel).

The situation is certainly not a show-stopper by any means, so I'd probably say converting to DC switch operation and having localized filtering is hunting flies with a bazooka.  I might end up doing a test with the DC boost to see if it makes any real difference - else, it will just remain one of those charms of PW Lionel layouts. 

If you added local buffering with the capacitor for each LED, the lights would be steady.

JTrains posted:

 

...The situation is certainly not a show-stopper by any means, so I'd probably say converting to DC switch operation and having localized filtering is hunting flies with a bazooka.  I might end up doing a test with the DC boost to see if it makes any real difference - else, it will just remain one of those charms of PW Lionel layouts. 

Well, this being a discussion forum, some additional thoughts.  I think DC is the way to go.  I also appreciate the idea that fussing with a dozen plus switches (converting to LED, adding filtering, etc.) can be a hassle.  So for the sake of discussion, we can contrive a requirement that the switches themselves are not to be modified.

One tricky part about staying AC is how to sense the momentary solenoid current surge in a timely manner.  AC sensors typically average the voltage or current over many 60 Hz line-cycles to get an accurate reading.  As you know, a switch solenoid can do its thing in a fraction of a second.  So without timely sensing of the solenoid current, the boost circuit could be late to apply the boost, and late to remove the boost.   I suppose you can forego averaging and instantaneously apply the boost whenever the current exceeds +2 Amps (some threshold above the peak of the 1 Amp peak steady-state current).  Then the trick becomes the circuit that swaps in the boost voltage 60 times a second - probably want to go solid-state vs. a chattering electro-mechanical relay.

Obviously, sensing the solenoid current increase (and decrease) in a DC configuration can be instantaneous - no delay "waiting" for the voltage or current to cycle.  Plus, it stands to reason that inserting a DC boost is "easier" if adding to a steady-state DC than to a steady-state AC.  Again, while an electro-mechanical relay might be fast enough, it can never be as fast to pick-up and release as a transistor/solid-state switch.  And transistors really appreciate working in an all DC environment!

Another idea easier to implement in DC is remote-sensing - a standard power-supply regulation method.  Since all your switches apparently see the same voltage drop, you can monitor the voltage way out at a switch and feed back to the control circuit near the transformer.  As a sense wire, this would carry negligible current so thin wire is fine.  The controller circuit would boost the voltage at the transformer to maintain a constant voltage (e.g., 13V DC) at the switch irrespective of solenoid activity.

A possible gotcha with DC operation is where you are getting the DC in the first place.  If deriving from AC accessory voltage thru a bridge-rectifier, this would make the DC- "ground" voltage different from the AC common.  This may have implications if using anti-derail operation.  Of course you can get inexpensive DC supplies (e.g. a 90 Watt DC-output laptop charger for $10) that would be an isolated source so can share DC "ground" with AC common but does add to the wiring jungle!

Town & Country Hobbies, who usually has a booth between Charles Ro & Bachman at York, sells direct replacement bulbs for 022 switches, both bayonet or screw base types.  When I built a layout for some grandkids, I replaced all switch and controller bulbs with LEDs so they wouldn't burn fingers.

What transformer are you using?

A heavier capacity transformer dedicated to just the switches will not dim the bulbs.

Larry

TrainLarry posted:

What transformer are you using?

A heavier capacity transformer dedicated to just the switches will not dim the bulbs.

I'm using a PW ZW-R and the situation occurs even with nothing else drawing current.  

stan2004 posted:

JTrains posted:

 

...The situation is certainly not a show-stopper by any means, so I'd probably say converting to DC switch operation and having localized filtering is hunting flies with a bazooka.  I might end up doing a test with the DC boost to see if it makes any real difference - else, it will just remain one of those charms of PW Lionel layouts. 

Well, this being a discussion forum, some additional thoughts.  I think DC is the way to go.  I also appreciate the idea that fussing with a dozen plus switches (converting to LED, adding filtering, etc.) can be a hassle.  So for the sake of discussion, we can contrive a requirement that the switches themselves are not to be modified.

One tricky part about staying AC is how to sense the momentary solenoid current surge in a timely manner.  AC sensors typically average the voltage or current over many 60 Hz line-cycles to get an accurate reading.  As you know, a switch solenoid can do its thing in a fraction of a second.  So without timely sensing of the solenoid current, the boost circuit could be late to apply the boost, and late to remove the boost.   I suppose you can forego averaging and instantaneously apply the boost whenever the current exceeds +2 Amps (some threshold above the peak of the 1 Amp peak steady-state current).  Then the trick becomes the circuit that swaps in the boost voltage 60 times a second - probably want to go solid-state vs. a chattering electro-mechanical relay.

Obviously, sensing the solenoid current increase (and decrease) in a DC configuration can be instantaneous - no delay "waiting" for the voltage or current to cycle.  Plus, it stands to reason that inserting a DC boost is "easier" if adding to a steady-state DC than to a steady-state AC.  Again, while an electro-mechanical relay might be fast enough, it can never be as fast to pick-up and release as a transistor/solid-state switch.  And transistors really appreciate working in an all DC environment!

Another idea easier to implement in DC is remote-sensing - a standard power-supply regulation method.  Since all your switches apparently see the same voltage drop, you can monitor the voltage way out at a switch and feed back to the control circuit near the transformer.  As a sense wire, this would carry negligible current so thin wire is fine.  The controller circuit would boost the voltage at the transformer to maintain a constant voltage (e.g., 13V DC) at the switch irrespective of solenoid activity.

A possible gotcha with DC operation is where you are getting the DC in the first place.  If deriving from AC accessory voltage thru a bridge-rectifier, this would make the DC- "ground" voltage different from the AC common.  This may have implications if using anti-derail operation.  Of course you can get inexpensive DC supplies (e.g. a 90 Watt DC-output laptop charger for $10) that would be an isolated source so can share DC "ground" with AC common but does add to the wiring jungle!

Discussions are indeed good.  Let’s continue with this a bit more.

When I said “switch in” a DC boost, my thinking was literally to switch it in: my control panel is built with DPDTs with the second pole unused.  So my thought was to wire up the second pole with a static DC boost of a couple of volts and when the switch was physically actuated to bridge that over into circuit.  This would mechanically remove the sensing part of the equation and keep all of the complexity back at the switch panel.  Your point that the DC source has to be beefy is a good one, as it would need to provide sufficient near-instantaneous power which itself could present challenge else it was all for naught.  Sounds like a job for some filter supercaps!

However, over my first cup of coffee…

…your comments got me thinking: how do we normally increase the power of a weak AC signal?  Looking down next to my bench at one of my other projects (a Fender-ish F51 FrankenChamp), it occurred to me: an amplifier! Feed in the steady-state AC voltage, amplify the voltage and do a bit of transformer magic, make sure the output is in phase, and voila.  Need to think this through a bit more – but talk about overkill!  Did I mention that the 5F1 is a vacuum tube amp?  Adding some superfluous NIXIE tubes would complete the paradigm, methinks - and give me some hipster/steampunk cred at the same time.

Taking a step back for a second: does everyone have this dimming effect occur that has not taken some specific measures to alleviate it?

TrainLarry posted:

A heavier capacity transformer dedicated to just the switches will not dim the bulbs.

What is also needed is also heavier wire going to the switches, the voltage drop is just as likely to be drop in the power wiring to the switch as the transformer.  Both elements figure into any voltage drop.

gunrunnerjohn posted:

TrainLarry posted:

A heavier capacity transformer dedicated to just the switches will not dim the bulbs.

What is also needed is also heavier wire going to the switches, the voltage drop is just as likely to be drop in the power wiring to the switch as the transformer.  Both elements figure into any voltage drop.

As a side note I use 14AWG for the switch feeders and 16AWG for the "star" connections that emanate from each one, so I would think there's plenty of ampacity in my configuration.  Also I'm using a PW ZW-R with everything else turned off and still seeing the effect.

JTrains posted:

...

When I said “switch in” a DC boost, my thinking was literally to switch it in: my control panel is built with DPDTs with the second pole unused.  So my thought was to wire up the second pole with a static DC boost of a couple of volts and when the switch was physically actuated to bridge that over into circuit.  This would mechanically remove the sensing part of the equation and keep all of the complexity back at the switch panel.  Your point that the DC source has to be beefy is a good one, as it would need to provide sufficient near-instantaneous power which itself could present challenge else it was all for naught.  Sounds like a job for some filter supercaps!

But doesn't the switch itself "automatically" disable the solenoid when it successfully toggles position?  In other words, the solenoid current stops flowing by itself irrespective of the controller.  So if the boost is slaved to the controller switch, it seems the boost might still be present after the turnout changes position and the lamps would momentarily brighten until you release the controller switch.   Unless you can exactly time the lever switch closure interval.  

stan2004 posted:

JTrains posted:

...

When I said “switch in” a DC boost, my thinking was literally to switch it in: my control panel is built with DPDTs with the second pole unused.  So my thought was to wire up the second pole with a static DC boost of a couple of volts and when the switch was physically actuated to bridge that over into circuit.  This would mechanically remove the sensing part of the equation and keep all of the complexity back at the switch panel.  Your point that the DC source has to be beefy is a good one, as it would need to provide sufficient near-instantaneous power which itself could present challenge else it was all for naught.  Sounds like a job for some filter supercaps!

But doesn't the switch itself "automatically" disable the solenoid when it successfully toggles position?  In other words, the solenoid current stops flowing by itself irrespective of the controller.  So if the boost is slaved to the controller switch, it seems the boost might still be present after the turnout changes position and the lamps would momentarily brighten until you release the controller switch.   Unless you can exactly time the lever switch closure interval.  

Oh, well heck - you're absolutely right.  What I proposed succeeded in creating the inverse of the problem I was trying to solve!  Which, thinking about it, is also the effect found in the PW transformer "whistle boost" circuit in that engines often speed up when the whistle is engaged on a non-motor-driven whistle.

All this is a lot more work than just ignoring the momentary dimming of the lights, like every other operator has done for the last 100 years - unless OCD is a factor.

Amen, RoyBoy.

Well, this is a discussion forum so maybe it's only me but I appreciate the exchange of ALL ideas as a part-and-parcel aspect of the hobby.

For example, I've always meant to look further into the use of audio amplifiers as a low-cost source of clean AC track power.  The invention of the "triac" AC transformer controller and its chopped-ugly sinewave might have been the worst "invention" for our hobby!  From the home-theater amplifiers available today, you get hundreds of Watts of clean variable sinusoidal 60 Hz track power at a fraction of what you pay for a typical chopped train transformer.   It was this about lamp brightness in turnouts that led the Jtrains to brainstorm the idea!

So back to the matter at hand, I nevertheless think all-DC would be the way to go.  For example, I'd look into a 50 cent FET (transistor) power switch that instantaneously turns on to boost voltage when >2 Amps (i.e., something more than the steady-state 1 Amp DC lamp-current is detected. 

But if staying with AC power, I suppose if high power audio amp were available, it would have a volume-control potentiometer (variable resistor).  A FET power switch would then command a higher "volume" by shorting some fraction of the potentiometer resistance when the solenoid coil current is detected.  This would boost the volume (voltage) of the AC output.

stan2004 posted:

Well, this is a discussion forum so maybe it's only me but I appreciate the exchange of ALL ideas as a part-and-parcel aspect of the hobby.

For example, I've always meant to look further into the use of audio amplifiers as a low-cost source of clean AC track power.  The invention of the "triac" AC transformer controller and its chopped-ugly sinewave might have been the worst "invention" for our hobby!  From the home-theater amplifiers available today, you get hundreds of Watts of clean variable sinusoidal 60 Hz track power at a fraction of what you pay for a typical chopped train transformer.   It was this about lamp brightness in turnouts that led the Jtrains to brainstorm the idea! 

Indeed Stan - as this is, after all, the "Electrical Forum" and I thought someone might have a really simple AC solution that I'd overlooked.  As RoyBoy suggests I'll probably just continue to suspend disbelief - as we do with the third rail, the random scale-ness of cars, the gateman who stands 20+ scale feet tall, etc. The fun, at least for me, is in the discussion of ideas as much as the solving of the problem.

Unfortunately my Frankenstein guitar amp (or a derivative of it) won't cut the mustard as it is single-ended and thus can only output about 15W max of power.  I guess I'll just have to build my first push/pull tube amp. 

gunrunnerjohn posted:

If you added local buffering with the capacitor for each LED, the lights would be steady.

John, that’s what I thought you saidl

JTrains posted:

Unfortunately my Frankenstein guitar amp (or a derivative of it) won't cut the mustard as it is single-ended and thus can only output about 15W max of power.  I guess I'll just have to build my first push/pull tube amp. 

Why would you select tubes for this job?  A solid state amplifier will be far more efficient, and cheaper to boot.  Did I mention it will be a lot smaller as well?  A really high power tube amp for 60hz is going to require some huge transformers!

gunrunnerjohn posted:

JTrains posted:

Unfortunately my Frankenstein guitar amp (or a derivative of it) won't cut the mustard as it is single-ended and thus can only output about 15W max of power.  I guess I'll just have to build my first push/pull tube amp. 

Why would you select tubes for this job?...

Maybe the light emitted by the tube filaments could boost switch bulb brightness.  

I forgot about the Photon Effect Stan, I'll have to keep that in mind.

gunrunnerjohn posted:

JTrains posted:

Unfortunately my Frankenstein guitar amp (or a derivative of it) won't cut the mustard as it is single-ended and thus can only output about 15W max of power.  I guess I'll just have to build my first push/pull tube amp. 

Why would you select tubes for this job?  A solid state amplifier will be far more efficient, and cheaper to boot.  Did I mention it will be a lot smaller as well?  A really high power tube amp for 60hz is going to require some huge transformers!

Obviously, it's 'cause tubes rock and you just can't get that vintage tone for your layout using SS!   Seriously, I'm not going to build a tube amp to somehow power my train - but I might just build another tube guitar amp.  Winters are long in Chicago.