Weak DCS signals, Failed TIU Output Drivers, and Design Solutions devloped under collaboration between AGHR and MTH

johnf posted:

I found someone with an oscilloscope who is going to let me test a TIU on it, and I have a couple of basic questions.

1) If my TIU is powered externally but not connected to a layout and no locomotives are present, what do I press on the remote to get the TIU to issue a bunch of signals that I can measure? 

2) Can I just connect the scope directly to the TIU outputs, or do I need the filter that GRJ mentioned in an earlier post.

2 gotta have the filter or you won’t see anything and won’t be able to trigger at a set level with the 18V 60 Hz underneath

1 it sends packets anytime the TIU is reset or of there’s any transients on the track so it's easiest to just toggle power.

Thanks Adrian. I couldn't find the post where you describe how to make the filter.  Is it in this thread or another?

johnf posted:

Thanks Adrian. I couldn't find the post where you describe how to make the filter.  Is it in this thread or another?

Old post

https://ogrforum.com/...ck-signalling-format

Thanks for the link. I think I even have the parts on hand to put one together.

Thanks again for your help Adrian. Here is a graph of the DCS signal that I captured from 1 of the channels on my TIU. The other 3 channels were all about the same. 

johnf posted:

Thanks again for your help Adrian. Here is a graph of the DCS signal that I captured from 1 of the channels on my TIU. The other 3 channels were all about the same. 

Looks a bit ringy, but workable.

Ringy?  As in "One Ringy-Dingy?"

Ringy?  Please explain?  These are suppose to be square waves, but in all of the pictures that we have all shown, none of them are square.  Is the so call "square wave" suppose to look as we have shown them in the photos or is there a lot of distortion being introduced some where in the circuit?

Bob D

rad400 posted:

Ringy?  Please explain?  These are suppose to be square waves, but in all of the pictures that we have all shown, none of them are square.  Is the so call "square wave" suppose to look as we have shown them in the photos or is there a lot of distortion being introduced some where in the circuit?

Bob D

It's a square wave plus an e^(yt) x cos(wt) term from all the wire inductance like the last part of the tutorial talks about. The amplitude is fine and if johnf tightened up the loops in the wire it's be a square.

I don't get all that square a wave when tapping right off the filter with no track or extra wire in place.  I wonder if a terminating resistor would help?

gunrunnerjohn posted:

I don't get all that square a wave when tapping right off the filter with no track or extra wire in place.  I wonder if a terminating resistor would help?

The downside is the terminator itself costs you  excursion voltage. The ringing shouldn’t matter much because the engine itself acts like a terminator and it’s always co-located with the decoder.

If it’s really bad it may be time to tighten up the split power-gnd sections of wiring. The split of + and - is what causes those rings 

Adrian! posted:

rad400 posted:

Ringy?  Please explain?  These are suppose to be square waves, but in all of the pictures that we have all shown, none of them are square.  Is the so call "square wave" suppose to look as we have shown them in the photos or is there a lot of distortion being introduced some where in the circuit?

Bob D

It's a square wave plus an e^(yt) x cos(wt) term from all the wire inductance like the last part of the tutorial talks about. The amplitude is fine and if johnf tightened up the loops in the wire it's be a square.

The part I am not following is the measurements that I have taken, have been straight off the output terminals of the TIU (no track wiring), where the impedance should be very low  (small R) resulting in minimal ringing.  But I still seem to be getting a lot of ringing?  What am I missing?  

Sorry for all the questions.  Thanks

rad400 posted:

Adrian! posted:

rad400 posted:

Ringy?  Please explain?  These are suppose to be square waves, but in all of the pictures that we have all shown, none of them are square.  Is the so call "square wave" suppose to look as we have shown them in the photos or is there a lot of distortion being introduced some where in the circuit?

Bob D

It's a square wave plus an e^(yt) x cos(wt) term from all the wire inductance like the last part of the tutorial talks about. The amplitude is fine and if johnf tightened up the loops in the wire it's be a square.

The part I am not following is the measurements that I have taken, have been straight off the output terminals of the TIU (no track wiring), where the impedance should be very low  (small R) resulting in minimal ringing.  But I still seem to be getting a lot of ringing?  What am I missing?  

Sorry for all the questions.  Thanks

I think the discrepancy is some of us are using high speed scope probes (low L) and some of us aren’t. The TIU itself has a transformer at the output so it may ring if it’s not de-Q-Ed by the scope probe. I think this plot and the other one by Johnf are fine. Those overshoots won’t do much to the decoder since the one and zero levels are very clear 

So you're saying use a crummy probe and it'll look better?

Adrian! posted:

rad400 posted:

Adrian! posted:

rad400 posted:

Ringy?  Please explain?  These are suppose to be square waves, but in all of the pictures that we have all shown, none of them are square.  Is the so call "square wave" suppose to look as we have shown them in the photos or is there a lot of distortion being introduced some where in the circuit?

Bob D

It's a square wave plus an e^(yt) x cos(wt) term from all the wire inductance like the last part of the tutorial talks about. The amplitude is fine and if johnf tightened up the loops in the wire it's be a square.

The part I am not following is the measurements that I have taken, have been straight off the output terminals of the TIU (no track wiring), where the impedance should be very low  (small R) resulting in minimal ringing.  But I still seem to be getting a lot of ringing?  What am I missing?  

Sorry for all the questions.  Thanks

I think the discrepancy is some of us are using high speed scope probes (low L) and some of us aren’t. The TIU itself has a transformer at the output so it may ring if it’s not de-Q-Ed by the scope probe. I think this plot and the other one by Johnf are fine. Those overshoots won’t do much to the decoder since the one and zero levels are very clear 

Adrain thanks.  I am trying to address the DCS issues at the club and wanted to make sure that I am dealing with TIUs which are generating a good signal before address layout issues.

Adrian,

       I commend you for your effort of diagnosing this problem and providing us with a solution. 

John,

      Just wondering if you had any success with the prototype you designed for the TIU?  Ive noticed there has not been a post in this thread in over a month, and just wondering if there are any more updates with your prototype.

Lionel16 posted:

Adrian,

       I commend you for your effort of diagnosing this problem and providing us with a solution. 

John,

      Just wondering if you had any success with the prototype you designed for the TIU?  Ive noticed there has not been a post in this thread in over a month, and just wondering if there are any more updates with your prototype.

 

Well at AGHR we haven't had a single weakened TIU channel in the 9 months since the changes at the top of this post. We used to break them weekly.... Every Saturday I go around with the scope and re-validate all 20 channels (4x5TIUs) one by one, and I haven't seen any changes in signal excursion since we started. I think the design adjustment is solid, so the only thing left is to figure out how to easiest apply it for those that would like to. That's more GRJ's dept, I'm just a circuit designer.

I'll have to get my little board back in the saddle, I have kinda left that one sitting.

Hi Adrian...

I was wondering approximately how many train "mishaps" might have occurred during that period to cause signal degradation. Or would it be too many to count?

Junior posted:

Hi Adrian...

I was wondering approximately how many train "mishaps" might have occurred during that period to cause signal degradation. Or would it be too many to count?

Uh... probably on the order of 10^3 < X < 10^4

LOL hysterically!

For those interested in how well this mod has worked in probably the most chaotic and abusive DCS layout in the world... Here's 7 months of data I just transcribed. We had one channel fail one time in June, but that was due to one of the clamp diodes coming off the board becasue of my poor soldering. We glue them down now.

Anyways, seems pretty reliable compared to the chart way at the top of the post. I'd say with this much testing, it's clear the solution is good.

Adrian, do you think my little piggy-back board will do the trick?  If so, I can post the design and Gerber files for anyone that wants to install these.  I'm using a 1500W TVS isolated with Schottky diodes.  The diodes only have 15pf capacitance, so the total of 30pf shouldn't hack the signal up too bad, looks like about a capacitive reactance of 1.6k at 3.27mhz.

gunrunnerjohn posted:

Adrian, do you think my little piggy-back board will do the trick?  If so, I can post the design and Gerber files for anyone that wants to install these.  I'm using a 1500W TVS isolated with Schottky diodes.  The diodes only have 15pf capacitance, so the total of 30pf shouldn't hack the signal up too bad, looks like about a capacitive reactance of 1.6k at 3.27mhz

Probably will work but I’ve been tricked so many times while solving this. Months of testing is the only way to be sure it’s good! 

My problem is that I don't have the platform for "months of testing" right now, so it's all theoretical for me right now.   I was hoping someone would have the venue that could give it a proper run and see if it's a real solution.  I put them on a TIU by just sticking the bottom to the 244 chip and then running the wires to the pads.  I found the easiest way was to solder the leads to the 244 chip sticking up, then mount the protection board, and finally tack the leads to the protection board pads.

I bought a 2nd hand rev-l tiu, and put it on my BIG layout.  Signal strength is 1-2.   I hope I don't have the issue in this thread! 

Its way too early for me to tell tho.  Gotta run some basic experiments off the layout to see if the signal strength is ok on single piece of track.

swise posted:

I bought a 2nd hand rev-l tiu, and put it on my BIG layout.  Signal strength is 1-2.   I hope I don't have the issue in this thread! 

Its way too early for me to tell tho.  Gotta run some basic experiments off the layout to see if the signal strength is ok on single piece of track.

add the light bulb (with enough draw) or filter and see if it improves, or swap out the whole TIU. It sounds like you already know how to test and what to look at!

Swise, you didn't say whether this replaced a TIU or is a new DCS installation.  If the former, out another TIU in place of the newly-acquired one and see what signal strength is.  If the latter, first thing I'd ask is what gauge wire feeds the tracks and do you have multiple center rail drops per block.  If the answers do not point to an issue, I'd say borrow & try another TIU.

swise posted:

I bought a 2nd hand rev-l tiu, and put it on my BIG layout.  Signal strength is 1-2.   I hope I don't have the issue in this thread! 

Its way too early for me to tell tho.  Gotta run some basic experiments off the layout to see if the signal strength is ok on single piece of track.

Sounds like exactly the job for the $20 TIU tester ....

Or the slightly more expensive digital 'scope!

I have the boards! I will be putting one together and getting the kits ready today. To those that have ordered one I will be sending out an email soon.

Hi RTR12....curious as to what level of assembly these kits entail....mounting them in a case? Significant soldering? Do you need to be an EE to assemble? I think I missed the posting concerning the "TIU Signal Tester" 😕.

My TIU signal on a few pieces of new, cleaned gargraves was 9-10 on all channels.  Whew.    But I do have a USB scope and a laptop, I suppose I could get my geek on and look at signals.

Now, am I correct in my reading of this thread that the PSX-AC boards damage the TIU to eventual failure  by virtual of doing their job (surge protection)?  IE the more derails, fat fingered car handling you have, the more it damages the TIU?    Is it only when the TIU is used passively?  Or both active and passive setups?  I have 4 major power blocks, and PSX-AC boards between my Z400s and the track.  I like that they kill power when there's a short, but restore it when the short is fixed.  Very handy for operating.  But I don't wanna kill my TIU.   And I'd rather not open the dang thing up and do surgery...

Sorry if this is already asked and answered...

The PCBs are pretty easy to assemble, all thru hole parts and nothing too small. Basic soldering is about all you need to know. The TIU tester has it's own thread we should probably move over to.  Here's a link to the other thread where GRJ developed the PCBs and files for getting the PCBs made. The link goes to a picture of the one GRJ built and there is another picture of one that JohnF built on the last page of the thread. Stan also posted some info on how to calibrate the tester when finished. I have not gotten that far on mine, still working on it. The boards I have are exactly like the one's in GRJ's post that I linked above, got them from the same place and all the parts are from Digikey.

Stan came up with a nice way to calibrate this without needing a 'scope, pretty simple, link below

Stan's DCS TIU Signal Tester Calibration Procedure

This looks like it should get you pretty close as far as calibration is concerned.

Hi RTR12....Thanks for the info and the link to the new discussion...just finished reading all the pages...all good stuff!

Are you selling the PCB boards? I'd love to build one of these and very interested in buying one. The device would definitely allow me to easily check my TIUs for failure (I have one Rev. L that's suspect already).

BTW....which version of the board did you get produced....singe LED or dual LED?

rtr12 posted:

The PCBs are pretty easy to assemble, all thru hole parts and nothing too small. Basic soldering is about all you need to know. The TIU tester has it's own thread we should probably move over to.  Here's a link to the other thread where GRJ developed the PCBs and files for getting the PCBs made. The link goes to a picture of the one GRJ built and there is another picture of one that JohnF built on the last page of the thread. Stan also posted some info on how to calibrate the tester when finished. I have not gotten that far on mine, still working on it. The boards I have are exactly like the one's in GRJ's post that I linked above, got them from the same place and all the parts are from Digikey.

 

I would like one of your kits if you have any left.  G

I apologize if I missed this: is MTH planning to make a new version of the TIU that will eliminate the problem?

Jeff2035 posted:

I apologize if I missed this: is MTH planning to make a new version of the TIU that will eliminate the problem?

I seriously doubt it, and we sure haven't heard anything about a new TIU version.  I think their solution was to add the TVS, which Adrian has already shown doesn't really solve the problem.

I also would like to be in line for a kit or even just the PCB if either was available.

Don