Skip to main content

Hi All,

Here's a new project we just finished designing at AGHR I thought I'd share.

I think a lot of people here on the OGR forum already understand that the low impedance of many parallel connected light bulbs from the passenger cars have an adverse effect on the DCS packet excursion voltage. The packet excursion voltage is reduced from full scale to the voltage divided result of the total parallel bulb resistance over the sum of the total parallel bulb resistance plus the Thevenin equivalent output impedance of the DCS driver (TIU or Explorer).

V_train = R_Train/(R_DCS+R_Train) * V_DCS

We've been having grief in our AGHR club where folks come in and pile 5-10 older passenger cars on the layout, and suddenly the DCS signal excursion drops well below the logic levels where the decoders can work reliably. The normal way people recommend to solve this effect is to convert the lighting in passenger cars to LEDs. The Thevenin equivalent of the LED is quite high (Kohm range) and so it doesn't create such an obnoxious voltage divider. The thing is LEDs change the look of a train and a lot of folks don't like that, so we've instead started to put series chokes (100uH ish) in all the passenger cars, so they can still keep the light-bulb look since the 60 Hz goes right through, but while also allowing the passenger car to retain a high reactance so the Thevenin impedance in the MHz frequency range where DCS operates stays high.

As we started adding chokes everywhere it became easy to loose track of which piles of passenger cars have and have not had the choke inserted yet. This is not super easy to measure on the layout since the DCS packets are super short (few ms) and you need an oscilloscope to see whats going on. Also some passenger cars are complicated and have parallel connections so even if you put a choke at one pickup, a low impedance path is still presented at the second.

 

To overcome this we invented the super useful AGHR choke-meter with schematic shown here:

chokemeter

How it works:

Not too complicated... We take in 24V (needed to drive the display) and regulate 5V with a good old LM7805 regulator. Going from 24V to 5V is a lot of drop so that's going to need a heat sink. From there we use a ring oscillator built out of a good old 7404 hex inverter (HCT variant) and just load it with shunt capacitance until it oscillates at the same frequency as the spreading rate of DCS (About 3.25 MHz). If the tuning is a little off it's not a big deal, as long as it's not orders of magnitude away. After the ring oscillator we have a 100 ohm series to protect against shorting the 7404 output, and then connect to the the center rail of a test track. We also use a super simple diode envelope detector (I use a 1n4148 diode but it probably doesn't matter). Then I just have a single OP stage to pull out the offset and give us some gain. I went with an LM358 but probably any old thing could work (LM741, LM318, LM356, ...). Finally I just go out to a volt meter. I found these nice self-powered ones on Amazon (VOLTMETER). The only trick is since they are self-powered you need to tune the offset so they sit above 3V.

Anyways wrap it into a nice board with a few connectors and it's done.

chokemeter_layout

 Does it work? (Of course it does or I wouldn't be posting). Here's an MTH subway car( without chokes) and one of my Weaver cars with the nice LED kit from GRJ that has a choke built in:

 

So the basic idea is we just tell folks to measure their cars and if the excursion is under 10V, they need to put chokes or LEDs before they can run trains. Super simple, but really really good at eliminating the signal loading issues in a club environment.

Yes, basically this is just a ohm meter at 3.2MHz  instead of DC.

Attachments

Images (2)
  • chokemeter
  • chokemeter_layout
Videos (1)
CHOKE_METER
Last edited by Adrian!
Original Post

Replies sorted oldest to newest

I can see this being useful in a club situation to identify potential running issues.  I do have to disagree with the not converting to LED's in all but the most die-hard "I want to keep it 100% original" guys.  You can get the LED's in any color temperature you like, so changing the look is normally restricted to having much better and more even lighting in the cars.

A couple of things overlooked in sticking with incandescent bulbs is the vast power savings with LED lighting, on the order of 90-95%, and the flicker-free running.

For a club situation, the power savings can be a major boost.  When our modular club runs, we regularly have issues with passenger trains causing issues with both power and DCS.  More than once we've had to have someone take their passenger train off if another passenger train was running on the same track, the 180W bricks just couldn't support two passenger consists.  Converting to LED's solves both issues.  Since, at least for my conversions, I have a DCS choke in every car, the DCS issues are solved as well as the excessive power consumption issues.

gunrunnerjohn posted:

I can see this being useful in a club situation to identify potential running issues.  I do have to disagree with the not converting to LED's in all but the most die-hard "I want to keep it 100% original" guys.  You can get the LED's in any color temperature you like, so changing the look is normally restricted to having much better and more even lighting in the cars.

A couple of things overlooked in sticking with incandescent bulbs is the vast power savings with LED lighting, on the order of 90-95%, and the flicker-free running.

For a club situation, the power savings can be a major boost.  When our modular club runs, we regularly have issues with passenger trains causing issues with both power and DCS.  More than once we've had to have someone take their passenger train off if another passenger train was running on the same track, the 180W bricks just couldn't support two passenger consists.  Converting to LED's solves both issues.  Since, at least for my conversions, I have a DCS choke in every car, the DCS issues are solved as well as the excessive power consumption issues.

I agree John. You can now get direct replacement LED bulbs in soft white, which look almost identical to the originals. I have bought a few of these. They make E10 18V or 14V and Ba9s and there are lots of other options as well. The purests can keep an old bulb around to put back in when the train is not running.

George

Last edited by George S

Does this mean that nearly all lighted cars - cabooses, halloween, Christmas, etc. - are going need chokes installed in order not to degrade the DCS signal?

 My club and another in the SF Bay Area have been struggling to implement DCS for several years.  It seems that putting chokes in every lighted car or changing the lights to LEDs is a huge task and expensive undertaking.  For example, I have a lot of K-Line passenger cars with the old light boards installed.  NH Joe 

Basically Joe, every incandescent bulb puts a little dent in the DCS signal.  The more bulbs you have, the more degradation of the DCS signal you'll experience.  Some cars with electronics are even worse as without the proper choke installed, they will really impact the DCS signal.  Atlas cars were somewhat notorious for that issue.

If you don't want to upgrade to LED lighting with proper DCS isolation, then this is a useful tool to identify the cars that will significantly impact the DCS signal.  The addition of a 22uh (or higher value) choke to these problem incandescent lit cars will fix the issue as well.  My thinking is that if you open the car to put the choke in, why not improve the lighting while you're there!

gunrunnerjohn posted:

Basically Joe, every incandescent bulb puts a little dent in the DCS signal.  The more bulbs you have, the more degradation of the DCS signal you'll experience.  Some cars with electronics are even worse as without the proper choke installed, they will really impact the DCS signal.  Atlas cars were somewhat notorious for that issue.

If you don't want to upgrade to LED lighting with proper DCS isolation, then this is a useful tool to identify the cars that will significantly impact the DCS signal.  The addition of a 22uh (or higher value) choke to these problem incandescent lit cars will fix the issue as well.  My thinking is that if you open the car to put the choke in, why not improve the lighting while you're there!

John,

Thanks for the information.  I think that the club may have to host a lighting and choke installation workshop.  The biggest issue is taking cars apart and putting them back together.   Also, soldering is a challenge for me.  Joe

gunrunnerjohn posted:

I can see this being useful in a club situation to identify potential running issues.  I do have to disagree with the not converting to LED's in all but the most die-hard "I want to keep it 100% original" guys.  You can get the LED's in any color temperature you like, so changing the look is normally restricted to having much better and more even lighting in the cars.

A couple of things overlooked in sticking with incandescent bulbs is the vast power savings with LED lighting, on the order of 90-95%, and the flicker-free running.

For a club situation, the power savings can be a major boost.  When our modular club runs, we regularly have issues with passenger trains causing issues with both power and DCS.  More than once we've had to have someone take their passenger train off if another passenger train was running on the same track, the 180W bricks just couldn't support two passenger consists.  Converting to LED's solves both issues.  Since, at least for my conversions, I have a DCS choke in every car, the DCS issues are solved as well as the excessive power consumption issues.

I make all these cases for LEDs weekly too, but sometimes it’s better to just put a choke there than having a 6 month debate. It’s obnoxious levels of power I know! 

Adrian! posted:

I make all these cases for LEDs weekly too, but sometimes it’s better to just put a choke there than having a 6 month debate. It’s obnoxious levels of power I know! 

I can't disagree with that, I get a lot of requests for stuff that I scratch my head about, but I just do them if that's what they want.   I was charging a guy $35/car for LED upgrades of a bunch of Premier cars, if he wanted just chokes, it would still be $20-25, so why not just go for the gusto?

To the point changing out lights, I just converted 4 cars from the ~1999 K-Line  K4670A  New York Central Empire State Express  15" 4-PAC.  These cars have so called “streamlighting”.  Each 15” car has 12 lights in it.  To experiment, I cut six bulbs off the strip and the lighting was still plenty bright for my taste, thus the conversion to led.  Saves a lot of power as you might imagine down from the 48 incandescent bulbs from 4 cars. 

73B76989-7AFF-4D0C-91DC-22E3C309BF1D

Attachments

Images (1)
  • 73B76989-7AFF-4D0C-91DC-22E3C309BF1D

Obviously we're talking about a DCS command-control environment with ~18V on the track.  I'd think anyone operating incandescent cars would have to do something about the excessive brightness since traditional cars weren't designed for command control voltages.  I guess some guys get lucky and find 18V or 24V bulbs and swap them in if practical.  Others use the "trick" of inserting a 5 cent diode so bulbs receive on only half of the AC cycle thus cutting brightness about in half when running on a command control layouts.  Of course inserting a diode is an equivalent effort/modification to inserting the 100uH inductor/choke!

Well, you can use the DCS tester we did a while back for such readings.  It could be expanded to have more signal strength LED's so you could see how much signal was there.  The only downside the DCS tester has it it didn't like chopped waveform on the tracks, that slipped through the DCS filter and skewed the readings.  However, I seem to recall that NJ-HR has a bunch of Z4000 transformers powering the layout, right?

Before the TIU Tester (I think), there was also Adrian's 'AGHR DCS Telemetry Train Rev-7' thread where he made a piece of rolling stock into a very nice piece of DCS testing equipment. Adrian also provided all the info here to make one, as I recall.

This is on my ' want to try someday list ' along with at least a thousand other things...

Last edited by rtr12
rtr12 posted:

Before the TIU Tester (I think), there was also Adrian's 'AGHR DCS Telemetry Train Rev-7' thread where he made a piece of rolling stock into a very nice piece of DCS testing equipment. Adrian also provided all the info here to make one, as I recall.

This is on my ' want to try someday list ' along with at least a thousand other things...

Simple isn't what I'd call that solution!

Simple it is not, but it's pretty cool and would be great fun to watch going around the layout. Especially a very large layout! I have been thinking about this since reading Adrian's thread when he posted it. I lack the know how, but I'm working on it!

It was for Bob D above, and his reference to something on a piece of rolling stock to check around the layout. I bet he has the know how, but his thoughts above would definitely be a lot simpler. 

Adrian, in your video, the excursion voltage dropped quite a bit. From 15v to 4v. Would a second subway car totally knock it out? It would make one think that running a subway set or a string of passenger cars, on a small layout, that you would totally loose command control. But you don't. Also, many layout are built with MTH's RealTrax that have lighted Lock-On's and lighted switch tracks. It would be interesting to see a test connecting a few MTH track together with a few switch track and powered through a Lock-On. Then, take a measurement with the bulbs pulled out and see if the excursion voltage drops when the bulbs are put back in. If it does, it would be nice to see MTH get someone to start manufacturing (that style) replacement bulbs that are LED.

A second car would knock it down some, but remember it's basically a voltage divider, so when you start with a lower voltage and divide it, you don't have the same absolute change.  OTOH, just seeing it go from 15V to 4V with one car suggests that the subway with multiple cars would be a major issue for DCS.

One question that should be asked here, is the voltage drop we're seeing with your measuring board the same as we'd see with the TIU?  I suspect that the TIU may have a lower impedance driving the track than the board with 100 ohms in series with the output.

Adrian, how close does this track to the behavior of the TIU signal?

gunrunnerjohn posted:

A second car would knock it down some, but remember it's basically a voltage divider, so when you start with a lower voltage and divide it, you don't have the same absolute change.  OTOH, just seeing it go from 15V to 4V with one car suggests that the subway with multiple cars would be a major issue for DCS.

One question that should be asked here, is the voltage drop we're seeing with your measuring board the same as we'd see with the TIU?  I suspect that the TIU may have a lower impedance driving the track than the board with 100 ohms in series with the output.

Adrian, how close does this track to the behavior of the TIU signal?

The TIU is 4 channels of 50 ohms (ACT244) ganged together in parallel so like 12 ohms, then a sqrt(2):1 transformer, so like somewhere around 24 ohms total. This one is set to 100 ohms so you don't blow the 7404 which is not a large quad driver like the ACT244 line driver. I thought about replicating the output stage of the TIU, but then you'd need to get the same transformer (which is a pain), and go surface mount for the ACT244 driver  (which is a pain), and it didn't feel necessary just to figure out who's trains are adjusted and which ones still need to be done.  I do have gain/offset trimpots there so you could turn the gain below 1 to try and match the load line of the TIU. That's referred through the diode detector though so you may have different shaped IVs and would only be able to match over a small range.

Despite the difference in gain, it is linear as a divider. A train that reads 5V does present have the impedance of a train that reads 10V. Again just trying to sort out "good" and "bad" quickly. If you want precision, definitely stick with the scope and look at the packets directly. Also note the DCS deocoders fail gradually. When the excursion is like 5V everything works fine, by like 3V you struggle with the long packets that add engines but still have short packets work like whistle and speed. At about 1.5V it's non-functional.

Thanks for the additional info Adrian.  This looks slick for someone that just wants to determine if a car (or locomotive I suspect) will impact the DCS signal.  It would be useful to test locomotives as well as many Legacy and some TMCC locomotives also whack the DCS signal.

Will you be posting Gerber files and a BOM (with parts values/numbers) so that some enterprising soul could maybe assemble a kit of parts for these?

Dave Zucal posted:

Adrian, in your video, the excursion voltage dropped quite a bit. From 15v to 4v. Would a second subway car totally knock it out? It would make one think that running a subway set or a string of passenger cars, on a small layout, that you would totally loose command control. But you don't. Also, many layout are built with MTH's RealTrax that have lighted Lock-On's and lighted switch tracks. It would be interesting to see a test connecting a few MTH track together with a few switch track and powered through a Lock-On. Then, take a measurement with the bulbs pulled out and see if the excursion voltage drops when the bulbs are put back in. If it does, it would be nice to see MTH get someone to start manufacturing (that style) replacement bulbs that are LED.

Generally I would say measuring the real DCS packets on a scope on the layout is definitely the best and gives you a detailed understanding of what's going on... not only amplitude, but overshoots, and general wave shape, all of which can be important. The telemetry train was like the middle tier and just gives amplitude readings of the DCS packet. The DCS packets are super short, like a few ms, and the line can stay idle for minutes if no one is pushing buttons. If you're alone and have to push buttons to generate signals on top of running the measurement it adds more complication to debugging. Sometimes the telemetry train's ADC, or even my scope will miss a packet if it's timed just wrong. That infrequency of DCS signal makes debugging hard, compared to the continuous waveform of  the circuit here.

The idea of this is it continuously has the square wave on the track that's DCS-ish enough to be representative, but is continuously present so you can just tweak and tune with instant gratification on a display and not worry about how to send what command, and figuring out which one is train-bound and which one is TIU-bound. The scale is different, but you can quickly tell which passenger cars are, and are not, the major offenders.

 

One thing you could do is connect this circuit to a layout and see what happens to the voltage swing in order to detect non-helpful shunt impedances. You'd have to disconnect the TIUs first so their output impedance doesn't load the measurements.

 

gunrunnerjohn posted:

Thanks for the additional info Adrian.  This looks slick for someone that just wants to determine if a car (or locomotive I suspect) will impact the DCS signal.  It would be useful to test locomotives as well as many Legacy and some TMCC locomotives also whack the DCS signal.

Will you be posting Gerber files and a BOM (with parts values/numbers) so that some enterprising soul could maybe assemble a kit of parts for these?

I'll do one better than that. Here's the design in diptrace:

Attachments

gunrunnerjohn posted:

Hmm...  None of the small parts have values Adrian.  I can guess at most of them, what are you using?

0.1uF and 10K for the detector stage. 74HCT04 for the generator, LM7805 for the reg. 330 ohms with the LED, 100 ohms for the series Rs, 1N4148 for the diode. The shunt caps on the ring oscillator I just changed over and over till I was near 3.25 MHz. I breadboarded first.

Dave Zucal posted:

... It would make one think that running a subway set or a string of passenger cars, on a small layout, that you would totally loose command control. But you don't. Also, many layout are built with MTH's RealTrax that have lighted Lock-On's and lighted switch tracks...

The operative term is "running."  From what I can tell, this choke meter is meant to be used with a non-running car.  That is, the bulbs are NOT on when using this tool.  I don't know what qualifies as common-knowledge, but incandescent bulbs have vastly different resistance when off and on - perhaps a 10-to-1 ratio from lighted to un-lighted.  Sure, the tool could have been designed to test the DCS signal loading with the car lighted up but the circuit gets a bit more complicated.

Your comment about the lighted lock-ons is also spot on.  Again, if you use this tool on an unpowered layout, the lock-on bulbs would be OFF.  So the bulb filaments would be "cold" and the resistance would be much smaller than when the bulb is lighted.

Apparently, this tool is meant to determine if an incandescent passenger car is unduly loading the DCS signal.  But expanding on your lock-on comment, a lot of guys add TVS protection devices to their DCS layouts.  These "look like" a small capacitor.  The math becomes a bit nerdy, but a typical TVS as recommended in various OGR threads has a capacitance of over 1000pF.  At the tool's test frequency of 3.5 MHz a TVS "looks like" a passenger car in terms of how much in loads down the DCS signal!  So to your point, we know there are guys who scatter multiple TVS devices around the layout (they are "cheap") and the layout continues to function with DCS in the same way that a train with a consist of stock (no added choke) lighted passenger cars functions with DCS.

Likewise, it's my understanding that some fast-acting circuit breakers (like the PSX) require a user-installed choke/inductor so as not to load down the DCS signal.

Last edited by stan2004
Adrian! posted:

...

As we started adding chokes everywhere it became easy to loose track of which piles of passenger cars have and have not had the choke inserted yet. This is not super easy to measure on the layout since the DCS packets are super short (few ms) and you need an oscilloscope to see whats going on. Also some passenger cars are complicated and have parallel connections soeven if you put a choke at one pickup, a low impedance path is still presented at the second.

Curious about the high-lighted comment.   Are you saying the standard practice is to use 2 chokes?  In other words, one per truck inserted in the wire running from the truck roller to a small hole in the car floor?  This goes to the sub-conversation about the hassles of opening up the car to make the modification - and hence why not just do an LED surgery since the patient is already under anesthesia and opened up.  The point being, when you say it's easy to lose track of which passenger cars have the choke inserted, why wouldn't you just look at the trucks to see if the 2 chokes have been inserted?  Not trying to be a wise-guy but am curious; I do realize there are lots of guys who are loathe to open-the-patient to perform wiring/component surgery.  Much easier if the solution can be performed without surgery.

gunrunnerjohn posted:

I put a choke after the rollers, I stick a PTC in the wire between the rollers to protect the wire between them in the case of a derailment.  I see no reason for two chokes, I'll have to understand why you'd need two.

Then you don't have to open the car.  Just cut the "hot" wire on each truck and splice in the 5-cent choke.  I'm just commenting on Adrian's original post.  I would think a typical operator would be able to open up the car if for no other reason than to change burned out bulbs.  OTOH I can imagine some "complex" dis-assembly for cars with detailed interiors where you have to remove/access multiple parts to access and splice into the common "hot" wire after they join together.

Add Reply

Post
The DCS Forum is sponsored by

OGR Publishing, Inc., 1310 Eastside Centre Ct, Suite 6, Mountain Home, AR 72653
800-980-OGRR (6477)
www.ogaugerr.com

×
×
×
×
Link copied to your clipboard.
×
×