DCS-RC Perpetual Barking Watchdog Generator (DCS-RC WD Gerber Files Added to First Post)

Don't know if this helps or not but, all the tracks in our engine terminal have a toggle switch for each individual track. Generally, the toggle switches are always "OFF". When a DCS equipped locomotive is powered up, by flipping the toggle switch to "ON", the model first powers up in "conventional" mode, however by simply selecting THAT model in a DCS hand-held, and then pressing "START UP", the model then is in "DCS MODE", and ready to be operated with the DCS hand-held. Seems to work pretty consistent for us.

I know I can do that, the object of the exercise is for the engines to come up dark.

If you could get a switch of adequate capacity with twice the number of terminals as tracks, you could use every other switch position.  This would give the effect Stan seeks.

John,

gunrunnerjohn posted:

I know I can do that, the object of the exercise is for the engines to come up dark.

OK, then I'm obviously confused,,,,again. I am under the impression that the "Watchdog Timer" signal is ONLY issued/transmitted when the TIU is first powered up. Thus, any and all DCS models sitting around on "live/powered" tracks, receive the "Watchdog Timer" and come up dark. Correct?

Hot- the watch dog goes out each time the voltage is raised above 0 on any channel.

The problem is that any toggled off siding will miss  the watch dog.. not a big deal for most.

As I mentioned I like to have a toggle on the input side  of the tiu  as well as the output just in case  it's needed as in  starting a lash-up  . (MU)

I wonder why the watchdog signal is limited to 15+/- seconds????

Gregg,

the watch dog goes out each time the voltage is raised above 0 on any channel.

Actually, the watch dog is issued any time the voltage at a TIU channel output port changes from 0 to any greater value. That's why the watchdog can be issued in passive TIU mode.

This and a whole lot more is all in “The DCS Companion 3rd Edition!" This book is available from many fine OGR advertisers and forum sponsors, or as an eBook or a printed book at OGR’s web store!

Robert,

I wonder why the watchdog signal is limited to 15+/- seconds????

The current duration of the watchdog signal is on the order of 45 seconds.

OK, so I'll reword the question:  why is the watchdog limited to 45 seconds, rather than being left on?

ANother question, which is relevant to GRJ's scenario, is how long must the power to a channel be off before turning it on will re-trigger a watchdog?

GRJ:  I would not that your solution does not account for the wifi, which, unless the app is left on from the last session, has to read the layout each time a loco is newly-started.

Barry,

i am electronics challenged but i will try and ask this correctly... if gunrunner john or myself uses a separate tiu we have in passive mode and use the DCS remote commander when power is applied to the yard tracks after rest of layout has been powered up that the tiu in passive mode and DCSRC is used then the engines will remain silent or did I miss something?

or do you mean wiring power to each yard track from an existing tiu and then using the DCS remote commander it the remote commander supplies the watchdog signal and engines will remain silent?

now how does this RC thing send watchdog signal? 

gunrunnerjohn,

correct me please electronics challenged I am!  if I understand you are making a PCB with some electronic items that will delay the power to a yard track when the toggle switch is set to on so the tiu and track thinks its being powered up just like mainline tiu's do and all tracks see the watchdog signal?

how close am i to understanding this?

gunrunnerjohn posted:

stan2004 posted:

I'm still fixated on eliminating the need to first go to the home/off position whenever changing active track!

I can see that Stan!

Ok, I admit I'm not close to being an expert on DCS or layout wiring, but would it be possible to put a toggle in place of the capacitor before the selector switch?  It changes the behavior slightly in that you would select the track, flip the toggle on, operate the locomotive(s), then when finished, toggle off, switch to the next track, power on, and so on.   If it works this way, sign me up.

RJR posted:

GRJ:  I would not that your solution does not account for the wifi, which, unless the app is left on from the last session, has to read the layout each time a loco is newly-started.

Correct, no plans for adding WiFi to this mix.

StPaul posted:

i am electronics challenged but i will try and ask this correctly... if gunrunner john or myself uses a separate tiu we have in passive mode and use the DCS remote commander when power is applied to the yard tracks after rest of layout has been powered up that the tiu in passive mode and DCSRC is used then the engines will remain silent or did I miss something?

If you have a separate TIU in passive mode, which is what I'm doing, you don't need the DCSRC, the TIU will do the job just fine.  Barry was suggesting using the mainline DCS signal and just generating the watchdog with the DCSRC, however we're using the separate TIU.

Ironhorseman posted:

Ok, I admit I'm not close to being an expert on DCS or layout wiring, but would it be possible to put a toggle in place of the capacitor before the selector switch?  It changes the behavior slightly in that you would select the track, flip the toggle on, operate the locomotive(s), then when finished, toggle off, switch to the next track, power on, and so on.   If it works this way, sign me up.

You could, of course, use a toggle switch before the selector switch to interrupt the power, obviously that would work.  Stan wants to automate it so you just pick a track and everything happens behind the curtain.

gunrunnerjohn posted:

...Stan wants to automate it so you just pick a track and everything happens behind the curtain.

I'm still on the case.  RJR's comments about watchdog timing got me thinking.  Like he says, why not have the watchdog in the yard on all the time?  So my latest scheme:

Use a DCSRC passive-mode powered by a relay that continuously turns on and off once per second (using a homebrew 555 timer or an eBay cycle-timing module for $1-2).  I just checked and a DCSRC will generate a watchdog after a power interruption of a fraction of a second.  This will put a watchdog signal on the track once per second whenever there is yard power. 

Now you can rotate the selector switch directly to the desired track without first going to the off/home position.  And within one second, the newly powered track will get a watchdog signal and the engine will come up silent in command-mode.

The bonus of this approach is the track bumper LED that indicates to the user which track is active never blinks OFF as it would for any scheme that momentarily cycles track power to a passive-mode TIU.

I enjoyed reading all of this.    Run your trains and have fun,  I do.   That is with a perfect 10 signal.  

I switch on the track(s) in the yard I plan to use before firing it all up.  The locomotives then catch the signal, although locomotives are not supposed to be stored in the freight yard! The Super gets her dander up if we don't "follow the rules" on the CL&W.  Being a non-complicated kind of guy I've chosen the simplest method which proves to work, by trial and error. FINALLY got the layout working the way I want so we stopped upgrading and improving; left DCS at the last release I could get to load properly; same with TMCC.

No budget for going to Legacy, but as we don't have any Legacy equipment ... all continues to work with "regular" TMCC.

Budget for new locomotives was frozen in favor of a locomotive repair fun, which is keeping even for now.

I have "blocks" that are controlled by an Atlas Selectors.  I simply: The power is turned ON everywhere for the TIU at the initial start-up so that the TIU can "recognize" all DCS command engines.  I then "kill" the power to each DCS engine on a particular track(s) via specific Atlas Selector toggle switch.  When I re-power later, the DCS engine automatically starts up after first contact with power.  I shut the engine down via remote ("6") and restart (#3).  I'm at the beginning.  Easy.

I started doing this because I'm tired of occasionally an engine will start itself and crash into the turntable (or worse) and I won't realize it happened until how much later?  One time I found an engine spinning its wheels against the outside of the turntable bridge--NOT GOOD.  Turning the power off to tracks DCS engines occupy after initial start-up PROTECTS from unwanted start-ups and worse--derailments.  Works like a charm!

I've never heard of a TIU recognizing command engines.  It just sends a command as directed by the remote to a loco that has already been added to the remote.

Unless one of my grandkids has pressed READ, all of my DCS locos are on the active list in the remote and, if on a powered track, will immediately respond to the startup command, whether or not they were on a powered track when TIU was turned on.

John,

I shut the engine down via remote ("6") and restart (#3).

You can skip the Shut Down and go right to Startup.

This and a whole lot more is all in “The DCS Companion 3rd Edition!" This book is available from many fine OGR advertisers and forum sponsors, or as an eBook or a printed book at OGR’s web store!

Barry, I realize that you can shutdown the engine after it starts without the watchdog.

John,

You just have to be more specific as to which John you're speaking.

John,

I realize this is over the top but I'll document it here anyway for future reference.  This is a refinement to my initial idea of a cycling relay to perpetually turn on/off a DCSRC once per second.  This would generate a DCS watchdog signal that would always be available to a yard siding or round-table spoke that was turned on with track power from an already active TIU output channel (so no watchdog).

The refinement was to recognize that the DCSRC uses a microprocessor chip inside.  And virtually all microprocessor chips have a signal input that will reset/restart/reboot the program.  So rather than an electro-mechanical relay perpetually clicking, why not generate an electronic signal to reset the microprocessor once per second.   Without getting into arcane details, the relevant signals in the DCSRC are conveniently available on a connector.  So by simply installing 3 pin on this connector, a circuit can be added which resets the processor chip once per second.

The 3 pins are +5V, Ground, and the Reset signal.  The 5V circuit is about $1 in readily available commodity parts and generates a narrow pulse about once per second which then resets the processor chip once per second.  The DCSRC appears to generate the watchdog about 3/4 sec after a reset.  Note that the modification is "additive" in that signal traces or components on the DCSRC itself are not modified, removed, etc.  So simply popping off the circuit board restores the DCSRC to its native function.

And here it is in action.

The green LED on the DCSRC briefly blinks off on each reset but it does not show up in the video.  The video shows track power with a perpetual watchdog (provided by the modified DCSRC) applied to a track with a PS2 engine; you can see the lock-on light turn on and hear the engine's relay click on when power is applied.  In each case the engine starts up silent (in command mode).  Then the perpetual watchdog is remove.  Now when power is applied the engine starts up in conventional since there's no watchdog.  And in this example, the engine happened to be locked in conventional forward so it starts moving illustrating the undesirable scenario described earlier and in other threads.

So now the configuration would be:

So this is just another alternative to the methods already described.

It is surely more complicated to implement though arguably easier for the end-user to operate.  Like comparing manual vs. automatic transmission, the clutch is an added step for the end-user.  The automatic transmission makes it easier for the end-user but is a more complex mechanism.  I compare the clutch to the extra step of toggling an extra on-off switch to cycle power to the TIU/DCSRC or having to first return the rotary switch to the home/off position. 

Well Stan, between you, John, Dale and a hand full of others on this forum you continue to come up with these awesome ideas to bring this hobby into the modern age of electronic gadgetry. You certainly have my attention with this one. So, how does one go about acquiring on of these reset generators?  

Milwrd

That's a pretty cool idea Stan.  Any reason you didn't go for more than one second for the watchdog signal?  It does seem to work fine with one second, just wondered if perhaps two seconds would give a better window for certain success.

Apparently that 6-pin connection on the DCSRC is for the programmer to update the processor code, I wonder why the convenient 3-pin connector with just the right signals?

Stan,

That is very interesting!

Dave

milwrd posted:

So, how does one go about acquiring on of these reset generators? 

Look closely, Stan hand built that one on a piece of perf-board.   Looks like it would be pretty easy to lay out one on a PCB, but I doubt there would be enough interest to warrant the expense of making the boards.

"Look closely, Stan hand built that one on a piece of perf-board.   Looks like it would be pretty easy to lay out one on a PCB, but I doubt there would be enough interest to warrant the expense of making the boards."

Hey Stan, is possible to get a few more pics of that board, especially the bottom? The schematic is plain and simple enough but it would clear up any misconceptions I have. Besides the wife hates the smell when I fry some electronic goodies.

Milwrd

That is very interesting Stan. I just happen to have a couple of those DCSRC devices left over from a couple of MTH sets. I might have to make use of them with your neat project here. Thanks for "staying on the case" and posting the results. 

And... thanks to one of your previous posts I might even have the 74HC14, in addition to all the other parts. That part sounds familiar, no doubt from one of you and GRJ's other electronics threads (which, after reading, I secretly order parts from). 

milwrd posted:

...Hey Stan, is possible to get a few more pics of that board, especially the bottom?

I wanted the board to fit inside the DCSRC (and be able put the case top back on) so started with a minimal sized perf-board of about 3/4" x 3/4".  In retrospect, the board area could be several times larger and still fit; a larger board ought to make it easier to assemble.  I took advantage of so-called pin-sockets for the IC chip though most guys don't have these and they are rather expensive (maybe 25 cents each) if you don't have a stash of them; considering the minimal cost of the IC chip (10 pieces for 99 cents on eBay, free shipping from Asia).   BTW note the 74HC14 chip in the photo is date-coded 8607...so that's from 1986 and of course Motorola (bat-wing M logo) is no longer in the chip business!

Here you go, a board that will easily fit inside the DCS-RC.  The header is on the reverse side and the board projects out similar to Stan's board, just a bit smaller.  Yes, it's all surface mount, but I used larger footprint components to make it fairly easy to solder.  There was no model for the diode in the 3D library, so that didn't render on the 3D view.  The pins on the back side are arranged so the circuit board is over the components on the DCS-RC.  There are no components on the bottom.

Updated schematic and layout for more compact parts.

gunrunnerjohn posted:

That's a pretty cool idea Stan.  Any reason you didn't go for more than one second for the watchdog signal?  It does seem to work fine with one second, just wondered if perhaps two seconds would give a better window for certain success.

Apparently that 6-pin connection on the DCSRC is for the programmer to update the processor code, I wonder why the convenient 3-pin connector with just the right signals?

I might be misunderstanding your first question but to improve the chances of success, I'd think a shorter duration between watchdogs is desirable.  By experimentation I noted the DCSRC generates the watchdog about 3/4 sec after it receives power (or after it resets/restarts).  Then it goes silent not generating any further watchdogs.  So I figured the idea would be to restart the microprocessor after 1 sec (or so) to create the 24/7 "perpetual" watchdog.  Stated differently, I see no way to generate watchdogs more frequently than every 3/4 sec since it appears that's how long the DCSRC needs to generate the one-time watchdog under normal operation.

There are 2 un-populated connectors (J1 and J2) on the DCSRC board.  This gets into the arcane but the board uses a PIC processor chip which as you point is essentially a sure-fire indicator that one of those connectors (J2 as it turned out) would have the standard Microchip programming pin-out of Reset*, +5V, and Ground in that order.  Only the first 3 pins of the J2 connector were needed for this; hence you'll note I used "J2" as the reference designator for the 3-pin connection to the DCSRC board!

I figured it was probably a PIC, I didn't feel like taking mine apart to verify.   My point is the watchdog lasts for many seconds, so leaving it on for a few seconds has a longer duration watchdog available to the track, or to state it differently, the watchdog has a longer duty cycle.

gunrunnerjohn posted:

...My point is the watchdog lasts for many seconds, so leaving it on for a few seconds has a longer duration watchdog available to the track, or to state it differently, the watchdog has a longer duty cycle.

Now I see your point.  But isn't that longer watchdog for the TIU (and not the DCSRC)?  Some of the earlier postings refer to a 45 second TIU watchdog interval which seems amazingly long.  I thought it was only a few seconds.  Anyway, I guess I wasn't looking for it but I didn't see any watchdog activity on the oscilloscope from the DCSRC after 1 second.  But you are correct that if in fact the DCSRC watchdog barks for more than 1 second, it would make sense to extend the interval.  The neat thing is the same circuit would work; simply change C1 from 2.2uF to, say, 4.7uF which would effectively double the interval to ~2 seconds.  Good observation!

gunrunnerjohn posted:

Here you go, a board that will easily fit inside the DCS-RC. 

LOL, I'm thinking Lucy holding the football for Charlie Brown!   I do marvel how quickly you can crank out the PCB designs!

When I started to prototype it I was going to use surface mount parts but figured it would be easier to make changes using through-hole parts.  If SOT-23 is a "larger" footprint component, why not the SOT23-6 version of the HC14?  

I know it's simple to do, just wanted to confirm my understanding of the 1 second.  As you say, the time interval is trivial to tinker with.

stan2004 posted:

gunrunnerjohn posted:

Here you go, a board that will easily fit inside the DCS-RC. 

LOL, I'm thinking Lucy holding the football for Charlie Brown!   I do marvel how quickly you can crank out the PCB designs!

When I started to prototype it I was going to use surface mount parts but figured it would be easier to make changes using through-hole parts.  If SOT-23 is a "larger" footprint component, why not the SOT23-6 version of the HC14?  

 

A good point Stan, I could obviously change to that part.  For multi-pin parts, I try to stay with at least 1.27mm spacing on the leads.  .95mm probably isn't that bad.  It was so small with the full sized part that I didn't bother to look for the alternative.

Note that the three leads of the SOT-23 have wider lead spacing so soldering the individual leads isn't that challenging.  It's when you get fine pitch leads all together that it gets tricky.

Stan, makes sense, see the updated schematic.

Schematic and PCB layout for DSC-RC Watchdog Board