Update and Video: A Built-In Two or Four Port TIU Output Signal Indicator

First off, I'd suggest that the signal threshold will be variable, it will be based on the load on that channel of the TIU.  The simple DCS signal tester works because it tests the channel with no other connection to the channel, so the DCS signal is totally unloaded.  Given the variable loads presented by different layout configurations, I'm guessing you'll have results all over the map.  Truthfully, for one that was going to be active with the layout connected, I'd want more signal strength LED's, not less.

Next, the DCS signal tester only functions properly with a pure sine wave on the input of the transformer.  With electronic transformers, even at full throttle, the indications are frequently erroneous.  Removing some of the filter stages will only make it even more sensitive to the input waveform.  When I try to use the signal tester with my bench Z-1000 at full throttle, I still get erroneous indications, I have to jump in with my old 1033 to get proper readings.

Finally, if I were going to make something like this, I'd probably consider a device that didn't require calibration.  The reason that the HCT123 design requires calibration is the tolerance of the triggering voltage level for the chip.  I'd be looking at a design that didn't need calibration.

Aren't you sorry you asked?

Try using the simple test device with just a lighted bulb from a test track with the test device attached to track.  Totally different output from only the TIU output terminal.

What is wrong with the simple DCS track signal test if you think you have a problem, followed up with a TIU test?  KISS applies in my opinion.  G

I recall a thread a while back about someone observing you can hear DCS activity on an AM radio?  Do they even sell pocket AM radios any more?!

Anyway, the idea might be to place the radio on the track section or loop in question and listen for activity.  I don't know what it would sound like but I'd think you'd hear "bursts" of static?  Obviously there are loose ends to resolve such as interference from other active DCS channels and what not...

Stan, I don't see how that would quantify the signal levels.  The Rev. L is rather prone to having only one or two outputs of the ACT244 fail and just reduce the DCS signal amplitude.  That being the case, it's important to actually check if a full amplitude signal is coming out for your tests.

I use my little tester for every TIU that comes in as a quick test.  If I see anything abnormal, I connect the 'scope and see if the signal is full strength.  I'm not sure how this tester would give me that assurance.

Depends on purpose of the tool.  If it's a "signal indicator" as Rod proposed, then I interpret that as sort of basic dead or alive indicator.  I think in many instances when there is a PS2 or PS3 command problem such as engine-not-found, out-of-range, etc. -  the user just wants to know if the DCS signal itself is present at all at ANY level.  

I agree that it's nice to get some quantitative indicator of signal strength.  However, that's why the original port tester tool "requires" an un-connected TIU output so that it's a known environment or load on the DCS signal generator.  As soon as you attach the DCS output to a track then all bets are off.  Are there engines on the track?  Are there TVS clamps on the track?  Lighted lock-ons?  Are you in passive mode?  What about lighted passenger cars without the DCS choke installed?  Lionel engines modified or not with a choke?  Small oval or gigantic club layout?  Etc. Etc.  I think Adrian in one of his earlier threads pondered, then abandoned, a method of measuring the "RMS" level of the DCS spread spectrum signal.  This is of course very difficult to do in a $10 widget.  So, instead, simple peak detection is used as a proxy of signal level.  As you may recall from the scope photos of the DCS signal, the peak detector is a marginal indication of signal level since you're looking at the ringing-overshoot of a complex digital RF signal.  The ringing is a function of the impedance of the track and its loads which as stated above is all over the map.  Anyway, lots of techno-babble and to what end.

In my opinion, what would be interesting is something like those pocket-size AC detectors that cost only a few bucks used to detect presence of a live circuit in your house.  No electrical contact required.  You place the widget near a wire and it gets brighter as you get closer to a wire with 120V AC on it.

Stan, given the sensor's requirement for a pure AC waveform to properly sense any level of the DCS signal, I'm still wondering how useful this would be.  Of course, this is just one man's opinion, I just don't know if it will actually serve the purpose with any consistency.  With anything but a pure sine wave in, the DCS tester says there's a good signal there all the time, even just connected to the transformer!  Any spikes on the output get through the filter.

Both of the transformers below give me a continuous "good" indication with the DCS tester connected directly and at full throttle.  The sharp slope obviously gets through the filter and triggers the detection.  Almost all the electronically controlled transformers have a similar waveform with the exception of the Z-4000 which synthesizes a sine wave.  That being the case, a ton of layouts would not be able to use this kind of detection.  Since the Z-4000 is buried right now, I haven't actually tried it connected directly to the DCS tester to see if it's "almost sinewave" is good enough not to trigger the detection.  Don't know if those little "squiggles" in the waveform would sneak through the filter.

Z-1000 full throttle waveform

CW-80 full throttle waveform

The Z-4000 was not tested as I don't have it unpacked at the present time.

Z-4000 full throttle waveform

gunrunnerjohn posted:

First off, I'd suggest that the signal threshold will be variable, it will be based on the load on that channel of the TIU.  The simple DCS signal tester works because it tests the channel with no other connection to the channel, so the DCS signal is totally unloaded.  Given the variable loads presented by different layout configurations, I'm guessing you'll have results all over the map.  Truthfully, for one that was going to be active with the layout connected, I'd want more signal strength LED's, not less.

Next, the DCS signal tester only functions properly with a pure sine wave on the input of the transformer.  With electronic transformers, even at full throttle, the indications are frequently erroneous.  Removing some of the filter stages will only make it even more sensitive to the input waveform.  When I try to use the signal tester with my bench Z-1000 at full throttle, I still get erroneous indications, I have to jump in with my old 1033 to get proper readings.

Finally, if I were going to make something like this, I'd probably consider a device that didn't require calibration.  The reason that the HCT123 design requires calibration is the tolerance of the triggering voltage level for the chip.  I'd be looking at a design that didn't need calibration.

Aren't you sorry you asked?

Yep, I sure am! 

Seriously, I appreciate all the comments so far, exactly what I was looking for.  I had not considered any of these issues, so it kind of looks like my little project may have gone up in smoke. I may still order a small batch of boards though, just to experiment with. If nothing else they will go right along with a few others I have that make interesting paperweights! 

Rod

Rod, I have lots of "spare" fiberglass with copper etching, I'm familiar with that kind of project!

Well I went ahead and ordered a few test boards from OSH Park and just got one built and installed. Here is a built up pic before install:

You'll note the blue power indicator led, and the two red signal indicator leds. Note also the two 3362P style trim pots for sensitivity adjustment. The board length was more or less dictated by the distance between the leds because I wanted them centered between the sets of port connectors.

Here is an after installation video clip, showing the startup burst followed by the watchdog signal when the tiu is first powered up:

This test was done today, with about 30 ft of track set up, and a PS-2 engine. I tried it alternating between both fixed channels of the tiu. I have the trim pot on one set at about half, and the other at about ¾ of full gain.

I found the one set for higher gain gave signal blips in all cases, engine near or far. The one set at half gave signal blips most everywhere, except when the engine was within about 6 feet of the track feed. So this seems to indicate that the trim pot needs to be turned up to ¾ of full gain, or higher, to get consistent responses. But that’s OK IMO. 

Most control actions cause a single data blip. Like speed adjustments, whistle, bell, stop, reverse, etc.

The initial startup burst and watchdog signal are good long bursts, as seen above. Another long signal is getting the chronometer or odometer readings from the engine. Quite long and repeated interrogations.

Next test will be at a nearby friend's house on his layout. (I don't have a layout down here in Arizona that I can test on) Stay tuned! 

Rod

I'd be interested in seeing how it works on a full layout.

That will hopefully be next John. 

I am surprised about the lack of indications with the engine closer than 6 feet to the track feed, yet every other part worked fine. (This was the port with the pot set at ½ gain. The one set at ¾ gain did not show this symptom at all.)

But as we know from 17 years with DCS, it is very prone to signal reflection interference. My guess is that the engine causes strong reflections, and within 6 feet of the feed point they are strong enough to at least partially cancel out the tiu output signal, enough that the one shot circuit doesn't trigger. But that’s just a guess. If anyone has a theory about this I would be happy to hear it.

But in the end I am OK with this, if it only means I need to turn the pots up to near max.

I wish I had an "engineered filter" or a suitable bulb I could use to assess if either might improve the overall sensitivity. 

Rod

Rod, you can have a filter pretty quickly, they're dirt simple.

DCS Filter

Thanks for that link John. I knew I remembered reading something about building your own filters from back a few years. I just checked and I have both components. Well, I actually have 220 ohm 1% resistors, which fall inside the margin of error for a 221 ohm 1% one. So I will whip one up and give it a try! 

Rod

UPDATE: Successful layout testing!

Tried the tiu (with signal indicator leds per the video above) on a friend's smallish 036 Fastrack oblong loop, with a passing siding. I used a ps-2 MAC15 and about 6 cars, one being a lighted caboose. All went pretty much as expected. The leds blinked at every handheld input, speed, horn, bell, direction, couplers, etc. I did not notice any dropouts at all. Next test phase is on another friends larger Fastrack layout, hopefully soon. 

Rod

This little project kind of fell by the wayside for a while due to covid and various other distractions. But we are back in Arizona for the winter now and have a mid-size Xmas trainshow layout currently set up and running daily in a local mall. This has made for plenty of opportunity to run the test TIU with both fixed channels up and running with PS-2 engines on both tracks. The layout is about 16' x 12', and there is an 072 outer Fastrack loop, and an 036 inner loop.

The video clip below shows the indicator during a somewhat busier period than usual. To me it works very well and is exactly what I was looking for; a visual indicator that each channel is alive and well and exchanging data with it's engine or engines. The adjustment pots on both channels are set at about 3/4 of max range. At power up the startup burst followed by the watchdog bursts are plainly visible, (though not shown in this short clip.)

I am happy enough with operation that I plan to order up a batch of Asian boards and build up a few. The initial plan is to convert all 4 of my own TIU's and since the one on the home layout uses all 4 channels in fixed mode, it will get the two board treatment.

If anyone else has any interest in this project shoot me an email and I will order a few extra boards. Not sure of the final price but it shouldn't be more than about 2 bucks a board. The components are likely another $5-6 per board, though I don't have an accurate price at this time. If there is interest I can also whip up a BOM. All parts are readily available from Digikey or similar, and its all simple thru-hole assembly. If you have ever built one of the portable TIU tester kits offered by rtr12, the construction is very similar to that project.

Rod

Pretty cool, if my TIU wasn't buried under the layout, that would be a neat option.

@gunrunnerjohn posted:

Pretty cool, if my TIU wasn't buried under the layout, that would be a neat option.

I have the same issue on my home layout John. My workaround for that will be to mount the boards in one or two small plastic project cases and mount them on my main control panel. That will make them plainly visible, and the only wiring needed is light gauge to the red and black port outputs from the TIU, so that can be done easily on the panel too. Operating power for the indicator boards comes from one of the two TIU ports, so there is no additional wiring required for power. Easy-peasy.

Three members have come forward and requested boards already, so I need to up my original order number by another 20 or so!

Rod

One wonders how it'll work with a track and trains providing a load for the DCS signals?  Have you had much experience actually running with it?

I get your concern John. We have had several full days of trainshow running so far, and it seems very responsive to anything we have thrown at it to this point. The sensitivity pots are both set at about 3/4 of full. There seems to be enough signal strength to produce led blips, though there is no way of course to know if we are missing any. And The test Tiu is an older rev I3 version, so a newer rev L version should do even better with its stronger signal. Once I convert my rev L version at home I will know the answer to that I guess.

Back when I first started this topic there was concern expressed that the signal would not be strong enough during typical layout operation. But so far I am not seeing any need for worry about signal strength.

Rod

Interesting, that would be my significant concern as well.

I'm guessing on the variable channels with anything but full voltage that the lights are on all the time.  They are using my tester on the bench.

So you are running the variable channels in conventional variable mode I take it? I have not tried the test tiu like that, but it only has one indicator board mounted, connected to only the two fixed dcs channels of course. For an operating layout this would be the way to run I think, Unless there is a need to run the variable channels also in fixed mode. There is always the portable tiu tester board to test the dcs signal on those variable channels when needed, right?

Rod

I normally run the variable channels in fixed mode as I run 99% command.  However, if I want to run variable, all you have to do is select the channel and change the voltage, you're instantly dropped into variable mode.

I'm just pointing out that the filter we use for the power won't filter out the chopped waveform harmonics so anything but a relatively pure sine wave looks like a DCS signal to the signal strength circuit.

@gunrunnerjohn posted:

I normally run the variable channels in fixed mode as I run 99% command.  However, if I want to run variable, all you have to do is select the channel and change the voltage, you're instantly dropped into variable mode.

I'm just pointing out that the filter we use for the power won't filter out the chopped waveform harmonics so anything but a relatively pure sine wave looks like a DCS signal to the signal strength circuit.

Yep I got it. Unavoidable I guess. I think most command operators switch the variable channels to fixed and leave them there. We're on the same page.

Rod

We have been running DCS quite a bit more over the last several days at our Xmas trainshow. Just want to clarify a couple of things so everyone knows.

First the video clip I posted Dec 18 above was made using the two handles of a Z-4000 transformer to power the fixed TIU channels, and the indicator works perfectly with it. There was concern expressed that it may not work with the Z-4000 as the waveform is a synthesized sine wave, with a few "squiggles" that a pure sine wave does not have. But it is safe to say that the Z-4000 plays very well with the indicator board, even though the board has only two stages of input filtering, compared to 4 stages used in the TIU tester board that rtr12 offers.

Secondly, today I tested the modified TIU with the output from a Z-1000 (which has a chopped sine wave output) and it behaves just as grj has observed above with the portable TIU tester; ie; the indicator leds are lit almost continuously. So it seems safe to say that this indicator board will not work if the TIU is powered by a Z-1000, CW-80, or any other similar transformer-controller combo that uses a chopped sine wave output to the TIU.
But it should work fine with any traditional postwar Lionel (1033, 1044, LW, KW, TW, ZW, etc.), Marx, American Flyer or similar toy train transformer, or any brick power supply (including PoHo) that uses an AC sine wave output.

Rod

@gunrunnerjohn posted:

...

Z-1000 full throttle waveform

.

...

GRJ, it's been a few years but didn't you (or was it me?) document scope waveforms of the TIU Variable channels set to Fixed for command-mode operation?

In your "full throttle" Z-1000 output, one might think (incorrectly) that a Z-controller set to full throttle would look like the original sinewave from the AC brick.

Likewise, I'm trying to recall if the TIU Variable channel set to fixed (i.e., akin to full throttle in a Z-controller) also exhibited some chopping remnants.

Rod, I might have missed it but I re-read this thread and curious if you've attached your signal indicator to the Variable channels set to fixed.  And if so did the Signal Indictor behave identically to when attached to the TIU Fixed channels?

Hi, Rod Steward, I would like the Bom, and order 6 boards looks like it works Well!

Sincerely thanks

Alan Mancus

@stan2004 posted:

GRJ, it's been a few years but didn't you (or was it me?) document scope waveforms of the TIU Variable channels set to Fixed for command-mode operation?

In your "full throttle" Z-1000 output, one might think (incorrectly) that a Z-controller set to full throttle would look like the original sinewave from the AC brick.

Likewise, I'm trying to recall if the TIU Variable channel set to fixed (i.e., akin to full throttle in a Z-controller) also exhibited some chopping remnants.

Rod, I might have missed it but I re-read this thread and curious if you've attached your signal indicator to the Variable channels set to fixed.  And if so did the Signal Indictor behave identically to when attached to the TIU Fixed channels?

I think that may have been you Stan.   However, here's a couple of quick shots.

Notice in fixed mode, the outputs very closely track the input, so with a pure sine wave in, you should get a pure sine wave out.  I think the diodes in the MRC transformer contribute to the flat-topping of the input waveform.  The first shot is with no load.

• Yellow: Transformer output waveform
• Cyan: Variable channel output
• Violet: Fixed Channel output

------------------------------------------------------------------------------------------------------------------------------

• Channel 1 Transformer output full throttle (MRC AH-501)
• Channel 2 TIU Variable Output (fixed mode)
• Channel 3 TIU Fixed Output

Same waveform with no load or a 2A load on variable Channel

• Channel 1 Transformer output full throttle (MRC AH-501)
• Channel 2 TIU Variable Output (variable at 1/2 throttle )
• Channel 3 TIU Fixed Output

No load on the outputs

Note how the switching transients when less than full throttle whack both the fixed and variable channels, very odd.  The spikes even affect the transformer output, which explains why they appear everywhere.

I added a 2A load, and it didn't improve anything, the switching transients look worse if anything!  However, with a load in fixed mode, the outputs look exactly like the first screen with no noise spikes.

Waveform with 2A load on variable Channel

I have a little trouble understanding why there are fairly high switching currents flowing when there's no load on the variable channel!  There must be considerable capacitance across those FET outputs to cause that kind of spike.

All in all, an ugly output, one I didn't expect!

@stan2004 posted:

GRJ, it's been a few years but didn't you (or was it me?) document scope waveforms of the TIU Variable channels set to Fixed for command-mode operation?

In your "full throttle" Z-1000 output, one might think (incorrectly) that a Z-controller set to full throttle would look like the original sinewave from the AC brick.

Likewise, I'm trying to recall if the TIU Variable channel set to fixed (i.e., akin to full throttle in a Z-controller) also exhibited some chopping remnants.

Rod, I might have missed it but I re-read this thread and curious if you've attached your signal indicator to the Variable channels set to fixed.  And if so did the Signal Indictor behave identically to when attached to the TIU Fixed channels?

Stan, you raise an excellent question! I had assumed to this point that they would behave identically when set to fixed. It did not occur to me that they might not. I will try that out with the portable tester asap and see what the verdict is.

Rod

@Rod Stewart posted:

Stan, you raise an excellent question! I had assumed to this point that they would behave identically when set to fixed. It did not occur to me that they might not. I will try that out with the portable tester asap and see what the verdict is.

Rod, see previous post.

Grj, those waveforms you posted above suggest that it's doubtful that the variable tiu channels will work properly (with the indicator) when set to fixed mode, if I am interpretting your results correctly.

And it seems that the Mrc transformer output may be producing spikes on all outputs, which might mess up the indicator board and/or the portable tiu tester on any channel, fixed or variable? This may not be the case with a traditional postwar type transformer, or the Z4K though. I don't have an Mrc transformer so I can't test the indicator board or the portable tester with it powering a tiu.

I need to reread the CD74HCT123 datasheet, but I wonder if it triggers on any vertical rise or fall that the filter passes, or only those that pass through zero?

Rod

It's not the MRC transformer Rod, here's the picture with a Lionel 1033.  It looks like the TIU actually is responsible for the waveform distortion and the spikes.

John, so trace 2 is the variable channel at part throttle if I understand right.

What happens to it in fixed mode I wonder??. Is it still choppy?

(Sure wish I would have brought my Hantek DSO with me to AZ this year!)

Rod

Rod, this is the picture in fixed mode from my first post.  It's an identical picture with the 1033 powering it, no noise spikes.  It's the TIU variable channel chopping the waveform that is providing the spikes.  The 1033 even had some flat-topping of the sine wave, there's something loading that in the TIU as well.

• Yellow: Transformer output waveform
• Cyan: Variable channel output
• Violet: Fixed Channel output

------------------------------------------------------------------------------------------------------------------------------

• Channel 1 Transformer output full throttle (MRC AH-501)
• Channel 2 TIU Variable Output (fixed mode)
• Channel 3 TIU Fixed Output

Same waveform with no load or a 2A load on variable Channel

@Alan Mancus posted:

Hi, Rod Steward, I would like the Bom, and order 6 boards looks like it works Well!

Sincerely thanks

Alan Mancus

Alan, thanks for your interest. I will email you off-forum.

Rod

@Rod Stewart posted:

Grj, those waveforms you posted above suggest that it's doubtful that the variable tiu channels will work properly (with the indicator) when set to fixed mode, if I am interpretting your results correctly.

...

I think you're OK.  My interpretation of GRJ's scope photos is the TIU Variable Channel when set to fixed mode passes the input voltage to the track un-adulterated.

His photos show how a TIU Variable channel, when set to variable mode will affect the TIU input voltage coming from an MRC or 1033 in the examples shown.  In other words, if you're going to operate a variable channel in variable mode, give it a dedicated sinewave input.  Do not share the TIU input voltage with another TIU input channel.

And perhaps obvious, but if using an MTH Z-controller to power the TIU, just use the black brick.  Remove the throttle controller box.  As shown earlier, even at full-throttle position, a Z-controller chops the brick sinewave making it unsuitable as a power source for any TIU channel - fixed or variable.

@Rod Stewart posted:

...

I need to reread the CD74HCT123 datasheet, but I wonder if it triggers on any vertical rise or fall that the filter passes, or only those that pass through zero?

Per your schematic, the HCT123 triggers on a vertical rise that passes thru the filter.  The filter essentially passes "fast" rising edges.  The so-called "high pass" filter effectively blocks low frequency signals and of course DC or "zero" voltage is as low frequency as it goes.  So the HCT123 does not know if the rising edge passed thru zero.  The math is very tedious and I just offer the above technically sloppy reasoning.

So in GRJ's scope shot, if your indicator gadget was hooked up to the two waveforms shown, it would trigger on the fast rising edge indicated in the yellow boxes.  These signals are not passing thru "zero" and in fact are at wildly different  starting voltages.  The filter removes this "low frequency" starting voltage and only presents the fast rising edge to the HCT123.

Shown in pink boxes on the right are fast "falling" edges.  These should not trigger the HCT123 because you used the "B" input pins (as opposed to "A").

@stan2004 posted:

I think you're OK.  My interpretation of GRJ's scope photos is the TIU Variable Channel when set to fixed mode passes the input voltage to the track un-adulterated.

Correct, the TIU at full throttle on the variable channels passes the input waveform quite accurately.  Fixed mode is just setting the channel to full throttle initially.  As soon as you manipulate it with the remote, it drops back to variable mode.  Interestingly enough, if you use variable mode and scroll all the way up to 22V indicated, it cleans up the waveform and just passes the input waveform.

Bottom like, a pure sinewave input on the tiu with all the variable channels at full throttle allows you to use the channel tester on any channel.

Rod, one question, why didn't you use the 4-pole filter for the job?  It's proven to work, so I was curious why you cut it short?

@gunrunnerjohn posted:

...

Rod, one question, why didn't you use the 4-pole filter for the job?  It's proven to work, so I was curious why you cut it short?

@Rod Stewart posted:

...

To save some board space I deleted 2 stages of the original RC filter network, because I noted when testing mine with a scope there was no difference in output signal after 2, 3 or 4 stages. I have tested this with the portable tester and it behaves exactly the same as with all 4 stages.

Maybe the smaller board size makes it easier to cram into the TIU case?

I'm pleasantly surprised it works on a working layout rather than just on an unloaded detached TIU output.  I figure one would become familiar with the kind of flashing patterns for various DCS operations/commands.

I suspect one will see random flashes due to track voltage transients which might prove to be another application of this gadget.  In other words, when there is no DCS activity if there is, say, a dirty section of track I would expect some flickering of the signal indicator.  So over time if the flickering gets worse it can indicate the need for some track maintenance!

@stan2004 posted:

Maybe the smaller board size makes it easier to cram into the TIU case?

Could be, but surface mount components will easily fit all four channels on that single board.

I'm pleasantly surprised it works on a working layout rather than just on an unloaded detached TIU output.  I figure one would become familiar with the kind of flashing patterns for various DCS operations/commands.

It is nice to see that running on a real layout didn't send it off the deep end, a pleasant surprise.

@stan2004 posted:

I suspect one will see random flashes due to track voltage transients which might prove to be another application of this gadget.  In other words, when there is no DCS activity if there is, say, a dirty section of track I would expect some flickering of the signal indicator.  So over time if the flickering gets worse it can indicate the need for some track maintenance!

Track maintenance?  Just run trains.

BTW Stan, love my new Rigol 'scope, what a nice instrument!

I have no need for such a device, but having 4 remotes on my home layout, and having experienced variation is effectiveness among remotes, I's like to get my hands on some means to testing remote output.