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

Dumb question #1:

What kind of signal is the watchdog?  Could it be generated easily so that a circuit could be designed to just send watchdogs without a DCSRC?

Just thinking.... dangerous, I know....

Ed

So I have a bit of a question here...

We know that MTH chose not to have the TIU (or DCSRC) repeat the watchdog constantly, which would have eliminated the need for this fix in the first place.  Do you guys think it might be that the extra 'traffic' of this signal would reduce the effective bandwidth on the signal for other commands, and if so, will supplying a cycling watchdog, as this project is doing, cause any problems with the over-all signal on the layout?  I'm picturing it being a case where it wouldn't be a big deal normally to have two commands sent out from different sources , but with data constantly being sent from one point, will it affect the reliability of signals from a second source?  

I guess this comes back to the idea that there must have been some reason that the designers chose not to have the TIU provide a watchdog at a regular interval, and the only ones I can come up with are that:  1, the electronics used couldn't keep up with that on top of all the other functions, 2, the watchdog would interfere with the system in some way (using too much bandwidth), or 3, a lack of foresight that users might want to power up an engine after the layout has been powered.  If it is the last, though, I would think it would have been changed in a firmware update by now.  

Anyway, I guess this will mostly go to Stan for now, with the DCSRC cycling, do you experience any problems with other DCS functions?  

JGL

My guess is it's a fairly simple signal, but rolling a circuit to generate a modulated 3.27mhz signal would probably be quite a bit harder than spending $25-30 on the DCS-RC which already knows how to generate the signal.

JGL, in my case, the watchdog signal is just impressed on the yard tracks, so it's a moot point as far as bandwidth.  As soon as you drive out of the yard, you're on a different channel.

eddiem posted:

Dumb question #1:

What kind of signal is the watchdog?  Could it be generated easily so that a circuit could be designed to just send watchdogs without a DCSRC?

Just thinking.... dangerous, I know....

Ed

Was also thinking about this, building a stand alone box that would provide the signal either at a timed interval or when triggered to do so.  I think it could be done, but most people in this hobby don't seem to care much if they need a $50 part where a $2 one would work with a little more effort.  Stan's solution, therefore is probably good enough, any probably won't get any letters form Mike's lawyers over patent infringement for duplicating the DCS signal.  

JGL

JohnGaltLine posted:

 Stan's solution, therefore is probably good enough, any probably won't get any letters form Mike's lawyers over patent infringement for duplicating the DCS signal.

Which I can tell you first hand is a real concern!

As far as "a little more work", it's really a lot more work.  First, you have to research the nature of the signal, then you have to come up with the circuit to generate a compatible signal.  Finally, you have to build it.  Not as trivial as you might imagine for most folks in the hobby.

Oh, I understand that most folks here wouldn't have the skills needed to design the circuit, but some could do so, and simply copying a design is within the ability of many.  I think most of the information is already out there on the signal and it wouldn't be too terribly hard to simply record the data stream that a TIU or DCSRC spits out on power up, and use a micro-controler to spit the same 1's and 0's out on command, however Stan's solution is just a simpler way to go.  (assuming we're talking about the couple folks that do have the skills and equipment needed)

Now, it might be worth it for MTH to sell a ready to go box that provides a watchdog on command, or on a timed repeat.  

JGL

gunrunnerjohn posted:

Stan, makes sense, see the updated schematic.

Schematic and PCB layout for DSC-RC Watchdog Board

GRJ, as we meticulously re-arrange the deck chairs on the Titanic...

As has been demonstrated by earlier "boutique" projects, the conundrum with surface mount is someone needs to supply the printed-wire board AND a kit of parts to make it practical.  In this particular case, since a through-hole version fits/works fine, I wonder if a more appreciated physical product might be the albeit larger through-hole PCB?  I realize we're just chatting here but it almost seems like you're about to release Gerbers* ?!

*an industry expression indicating ready-to-go with a circuit board.

JohnGaltLine posted:

So I have a bit of a question here...

We know that MTH chose not to have the TIU (or DCSRC) repeat the watchdog constantly, which would have eliminated the need for this fix in the first place.  Do you guys think it might be that the extra 'traffic' of this signal would reduce the effective bandwidth on the signal for other commands, and if so, will supplying a cycling watchdog, as this project is doing, cause any problems with the over-all signal on the layout?  I'm picturing it being a case where it wouldn't be a big deal normally to have two commands sent out from different sources , but with data constantly being sent from one point, will it affect the reliability of signals from a second source?  

I guess this comes back to the idea that there must have been some reason that the designers chose not to have the TIU provide a watchdog at a regular interval, and the only ones I can come up with are that:  1, the electronics used couldn't keep up with that on top of all the other functions, 2, the watchdog would interfere with the system in some way (using too much bandwidth), or 3, a lack of foresight that users might want to power up an engine after the layout has been powered.  If it is the last, though, I would think it would have been changed in a firmware update by now.  

Anyway, I guess this will mostly go to Stan for now, with the DCSRC cycling, do you experience any problems with other DCS functions?  

JGL

Your comments/questions are quite astute and on-point.  There was a somewhat "animated" thread which for all I know got deleted where another member and I were at odds about whether a DCSRC and a TIU "talking" at the same time on a command track would/could interfere with each other.  Aside from that free-for-all, I don't recall seeing much discussion about this. 

My approach targets GRJ's specific scenario aptly described by this thread's title.  That is we are talking about a yard.  Or (as I have extrapolated) a roundtable.  In other words a limited environment where the TIU channel is talking to one DCS engine doing "simple" things or what I'll call low-bandwidth DCS communications.  So no loading of a new soundset, chain file, or whatever.  My belief is in a low-bandwidth DCS environment you will not see any material effect from a perpetual watchdog.

A question was posed early on about the relative signal power of the TIU watchdog and the DCSRC watchdog.  I did not measure this as it seems the TIU signal power/timing has changed over time so did not want to confuse matters.  I will suggest, anecdotally anyway, that the DCSRC signal is not as strong as the TIU.  Hence this is why I propose this alternative as useful for a limited environment like a yard or roundtable.

I choose not to ponder what it would take or why (or why not) MTH doesn't change the TIU into a perpetual barking watchdog...or at least make it feature that can be turned on/off on a per-channel basis.  I again find it curious that the watchdog length has increased over time and is now up to 45 seconds which seems like strange value!  But this does indirectly shows that watchdog activity can co-mingle with normal DCS command activity.

Not sure I provided definitive or hard-data to answer your questions...other than to clarify that I am proposing an idea to address a somewhat narrow/limited DCS scenario rather than DCS overall.

"*an industry expression indicating ready-to-go with a circuit board."

First, Stan thanks for the additional pics of the board. Second: Jeez GRJ, here is a perfect opportunity to put together another kit like you did for the EMD's. A PCB and a few parts in a bag  with a simple recipe and voila.

Just a thought anyway

Milwrd

OK, here's the thru-hole version.   All the parts are generic stuff and should be readily available for anyone.  As Stan observes, since this should fit in the case...

If anyone is interested, I can generate the set of Gerber files for either of these layouts.  If there was enough interest for around 15-20 of the blank PCB boards, I could arrange to get them made.  I would guess a single board would be around $3 and shipping a couple bucks.  Multiple boards would obviously reduce shipping per board.

Edit: updated to latest design on board layout and schematic additions.

milwrd posted:

"*an industry expression indicating ready-to-go with a circuit board."

First, Stan thanks for the additional pics of the board. Second: Jeez GRJ, here is a perfect opportunity to put together another kit like you did for the EMD's. A PCB and a few parts in a bag  with a simple recipe and voila.

Just a thought anyway

Milwrd

I did sell all of those kits except one, they were more popular than I imagined.

I'm in for one, maybe two. It's a no-brainer for my turntable/roundhouse area setup.

Milwrd

It would be fairly expensive for one or two, I think we'll have to have a few more interested parties.   However, if you know anyone with an in at a board house, the Gerber files are posted above.

John,

If you get production rolling, I'd be in for a couple of boards.

Clearly you're going for a speed record to get a board into production! 

As you know, you need to give time for the designer to waffle and/or add bells and whistles.

Anyway, IF a board is made, I suggest adding R3.  This would make the green LED blink OFF longer (it is kind of hard to see in the original design and not visible in the original video).  It is visible in following video which has R3 added per modified schematic below.  It does not change the performance but is visually more appealing/comforting to see the blinking green light.  Note that the 10K value is already used so it's not a "new" part.  From assembly rendering it looks like it could be placed alongside existing resistors without adding to board size.

The other issue is there may be interference from the mounting posts on the top of the case. 

It seems to me if the 3-pin J2 installed from the solder-side of the PCB (and soldered on the component-side), the board will hit the left post from the case lid in the above photo.  The red box area is the footprint of how I did it.  I realize the choice of 0.1" header/socket influences orientation.

We'll let this simmer for a while and see what the interest is. In the mean time I do have all of the components lined up in case I have to go that route.

Milwrd

I also would be in for four of the PCBs, kits or however you would want to package them. I have all the parts for the thru hole PCBs, but I have no surface mount parts for those type of boards. I was going to make a couple, but it would be a lot neater to have them on a custom made PCB.  

Good input Stan, I'll look at the board.  I have to say, the extra resistor was VERY hard to see in the schematic I used that you posted.

My mistake on the position of the 3-pin header, if it's on the other corner, the board sits where you have it pictured, which is where I intended it to be.

How's this, it looks like I have it in the right position with pin 1 to the edge of the board.  I also added your R3.  100% thru-hole, should be easy for even a novice to assemble.

Edit: final thru-hole layout with proper component and plug positioning.

So here's thing.  If P2 is populated with a 3-pin male header as rendered in your drawing, I think you are forced to then solder it to the DCSRC board.  If a 3-pin female socket is placed on the DCSRC board, the combination of the header and the components on the daughter board will be too tall.  That is why on my implementation there is a female socket on the daughter board; the 3-pin socket is installed on the component-side and soldered on the solder-side like all the other parts.  Does this make any sense whatsoever?

I figure there is a benefit to being able to remove the daughter card by simply pulling it off (rather than de-soldering it) to revert to the original DCSRC.

Quaere:  When the DCS signal is turned on, it appears that there must be a continuous signal emanting from the TIU, since a loco can detect signal strength even without any command being sent.  Up above in this thread, it is suggested to use a second TIU turned on and off to generate a continuous watchdog.  If you have 2 TIUs (or equivalent) sending simultaneous DCS signals (I didn't say commands) to the same block of track, are there any issues resulting?

stan2004 posted:

So here's thing.  If P2 is populated with a 3-pin male header as rendered in your drawing, I think you are forced to then solder it to the DCSRC board.

You can solder the female socket to the board and put the male on the DSC-RC, it doesn't change the PCB layout.  The header was just to position the holes.

I am curious about one thing.  Why does the male/female pair take up more room if you solder the female to the DSC-RC and the male to the daughter board as opposed to the other way?  That confuses me just a mite.

gunrunnerjohn posted:

 

....Why does the male/female pair take up more room if you solder the female to the DSC-RC and the male to the daughter board as opposed to the other way?

There are other male-female 0.1" connector pairs that sit lower but I was figuring the type shown above are the most readily available and probably the most inexpensive.

OK, I get it now, you're putting the components face-down.  That's easily done.

BTW, I do an interesting thing with those headers if I want less clearance.  I push the pins through the plastic holder about the length of the holder so the pins are shorter on top.  I solder then in, trim them, and then pry off the plastic holder.  I get an additional .1" of clearance that way.

From this view, there looked to be plenty of clearance on the top, the tallest thing here should be the transistor at around 1/4".  I always like the board to be components up if I can, it's a character flaw.   However, it's no problem to swap it so it's components down.

I have been following this topic with great interest. I just love how you guys (grj, Stan, Barry, et al) have bounced off each other to develop a really slick modification to a readily available DCS-RC. Just goes to show that two or more heads are better than one. And that design-by-committee actually CAN work!  I really like the idea of the add-on board being plug n' play.

John; count me in for 2 circuit boards if you bring them to fruition, as well as parts kits if you go that way, as with the EMD kits.

I took my DCS-RC (courtesy Larry3rail) apart and I measure the clearance between the top of PCB to the underside of the ribs at 13 mm. If that is tight for the header and socket you could easily trim the ribs in the upper case a little to gain another 3 mm.

Nice to see how this comes along guys!

Rod

Hmm..  At half an inch, I think that Stan is correct, the components should face down.  I'll flip the connector to the other side.

How's that, looks like  it should fit now.

gunrunnerjohn posted:

OK, I get it now, you're putting the components face-down.  That's easily done.

BTW, I do an interesting thing with those headers if I want less clearance.  I push the pins through the plastic holder about the length of the holder so the pins are shorter on top.  I solder then in, trim them, and then pry off the plastic holder.  I get an additional .1" of clearance that way.

This is a good idea.  To elaborate on what GRJ is saying,

Obviously you need the carrier/body in place when soldering...then later slide off the carrier from the pins as shown in the photo.  Though I have seen some carriers which seem almost bonded to the pins and won't slide off so your mileage may vary.  Anyway, once the carrier is removed the remaining pins might need to be trimmed down by essentially the height of the removed carrier piece.

One tip Stan.  Don't trim them after you install them, you remove the nicely tapered ends.  That's why I specifically mentioned placing it in the holes, pushing down the pins to the correct height, and then soldering it in.  After they're soldered in, you remove the carrier, trim the back, and you have the nicely tapered ends to insert into the mating connector.

All my eBay strips come out of the carrier pretty easily.   A little heat from a heat gun makes it go even easier!

I have an A-off-B toggle switch that toggles power feed to two rotary switches. One rotary handles the turntable spoke tracks and the second handles the adjacent freight yard. I patched the DCS-RC into this feed and gave it a try. Works perfectly. Any selected PS-2 or PS-3 engine stays dark and silent until you select it and hit startup. This is a nice simple scheme as mentioned earlier in this thread, for sending the watchdog to a command engine, each time a new TT or yard track is selected. Well after the main TIU watchdog has come and gone.

I also ran a brief test to get an idea how long the TIU watchdog signal lasts. Looks like its less than 15 seconds. This is an I3a rev TIU.

Rod

Rod, the all important question would be, what software is loaded on that TIU?

If I made this PCB, what would the preferences be as to how it's supplied.  Obviously, those choices are priced low to high.   I don't think I'd want to try to accommodate all three options, so it would probably be either a bare PCB, or a choice of the kit or assembled unit.  The price of the kit or assembled unit would depend somewhat on the quantity as parts are cheaper in quantity.

Bare PCB?

PCB with kit of parts?

Fully assembled and tested unit?

Rod Stewart posted:

I have an A-off-B toggle switch that toggles power feed to two rotary switches. One rotary handles the turntable spoke tracks and the second handles the adjacent freight yard. I patched the DCS-RC into this feed and gave it a try. Works perfectly.

By "this feed" do you mean just the freight yard.  I must be misunderstanding but with an A-off-B toggle it seems you would need one DCSRC for the turntable feed and a second DCSRC for the freight yard feed?

gunrunnerjohn posted:

Rod, the all important question would be, what software is loaded on that TIU?

Version 5.0; freshly loaded about 10 days ago, with the help of several glasses of wine I might add!

Rod

gunrunnerjohn posted:

If I made this PCB, what would the preferences be as to how it's supplied.  Obviously, those choices are priced low to high.   I don't think I'd want to try to accommodate all three options, so it would probably be either a bare PCB, or a choice of the kit or assembled unit.  The price of the kit or assembled unit would depend somewhat on the quantity as parts are cheaper in quantity.

Bare PCB?

PCB with kit of parts?

Fully assembled and tested unit?

 

My preference would be option B (PCB with parts kit); and second would be option A (bare PCB).

I am still in for 2, either way.

Rod

stan2004 posted:

Rod Stewart posted:

I have an A-off-B toggle switch that toggles power feed to two rotary switches. One rotary handles the turntable spoke tracks and the second handles the adjacent freight yard. I patched the DCS-RC into this feed and gave it a try. Works perfectly.

By "this feed" do you mean just the freight yard.  I must be misunderstanding but with an A-off-B toggle it seems you would need one DCSRC for the turntable feed and a second DCSRC for the freight yard feed?

 

Stan; schematic of the main cab attached. The lower right corner shows the yard and TT rotary switches. They are both powered by one switch, so I have connected the DCS-RC to the center lug of the switch. (Sorry my drawing lacks the pizzazz of some of the nice cad drafting that I often see on the forum, but I don't have the skills to do the latter! )

Normal operation is to select the yard or TT track I want, then power it up from either the A or B source. This way I avoid hitting each of the intervening tracks with a power spike as the rotary moves past. Not really critical for the yard since there is rarely an engine parked on any of the sidings.

I realize that feeding both rotaries simultaneously introduces a "split" for the DCS signal, but that has never created an issue for me. I also realize that the additional TVS's would be considered redundant, since each TIU channel already has one. But I don't believe they hurt anything, and hey, why not?

Rod

Rod Stewart posted:

 (Sorry my drawing lacks the pizzazz of some of the nice cad drafting that I often see on the forum, but I don't have the skills to do the latter! )

 It's not that hard Rod, I saw your computer work with your nice motherboard document, I know you can do it.  Look into TinyCAD for schematics, pretty nice little free package.

gunrunnerjohn posted:

If I made this PCB, what would the preferences be as to how it's supplied.  Obviously, those choices are priced low to high.   I don't think I'd want to try to accommodate all three options, so it would probably be either a bare PCB, or a choice of the kit or assembled unit.  The price of the kit or assembled unit would depend somewhat on the quantity as parts are cheaper in quantity.

Bare PCB?

PCB with kit of parts?

Fully assembled and tested unit?

 

To simplify I will go along with Rod, option B (PCB with parts kit); or option A (bare PCB).

I am still in for at least 2 also. (Possibly considering a couple more, but the 2 are certain.)

If we can get about 10 boards and/or kits committed, I'll get them made.  I'd like to at least break even on the PCB builds, the minimum order is $36.  I figured the boards costing $3, and the parts another $3.  If you want them built and tested, add $5.  Does that sound reasonable?  Shipping for a kit would be $3.50 first class package, two or three in the package wouldn't add to the shipping.  If I do kits, I probably wouldn't do bare boards as I'm going to have to buy the parts in some quantity to get decent pricing.

Hey John!

Put me down for 6 built and tested units.  Two for home, two for the NCOG club layout in Mundelian Ill, and two for the Lionel Club of Southeastern Wisconsin.

Keep me posted on the progress.

My other problem from today, that you responded to is solved, by replacing the USB to Serial Adapter cable.  How can these things go bad?   

John,

I'll take 2 at (or a bit above) the stated price. Just let me know exactly how much and when to hit your PayPal (friends & family) account.

Thanks!

Rod Stewart posted:

stan2004 posted:

Rod Stewart posted:

I have an A-off-B toggle switch that toggles power feed to two rotary switches. One rotary handles the turntable spoke tracks and the second handles the adjacent freight yard. I patched the DCS-RC into this feed and gave it a try. Works perfectly.

By "this feed" do you mean just the freight yard.  I must be misunderstanding but with an A-off-B toggle it seems you would need one DCSRC for the turntable feed and a second DCSRC for the freight yard feed?

...Normal operation is to select the yard or TT track I want, then power it up from either the A or B source.

Got it.  Your diagram makes it perfectly clear.  I misunderstood the purpose of the A-off-B toggle; I thought it selects (A) yard or (B) turntable...but it selects the TIU source from (A) Fixed 1 or (B) Fixed 2.  So only 1 DCSRC needed.