Adrian,

Were the measurements in the Original Post done with the club’s 5 TIUs in Super Mode? Do you have any measurements with 1 TIU?

Thanks

Super Mode should have absolutely nothing to do with the track signal coming out of the TIU.

OOPS!

My previous board layout had an error!  I slipped a digit and specified the footprint of the TVS as a DO-214AC, however it actually has a footprint of a DO-214AB.  This doesn't sound like much, but it's actually HUGE!  That being the case, my previous layout didn't really work as the TVS is much larger than I allowed for. 

How much larger you might ask...

The answer, as you can see above, is GOBS! 

So, back to the drawing board, I think I have a reasonable solution.  Since the desire was to keep the board the size of the ACT244 to avoid any issues with clearances, I decided to put the large TVS on the bottom and the four diodes on the top.

This will make the stack slightly taller, but that doesn't seem to present an issue from what I see.  It still remains in the footprint of the ACT244.  I will have to update my project files on OSH Park to reflect the new layout.

Sorry for any inconvenience, I'm about to receive some boards that I'll be tossing and starting over, as I suspect a couple other people are.

This is a link to the new layout: MTH TIU Signal Protection PCB

Since this is for evaluation/proof-of-concept, couldn't you just use the original board and install the larger package at a diagonal (one terminal soldered to existing pad, the other terminal in mid-air with a short wire down to the other pad)?  

You probably could Stan.  That would prove the concept, not a bad idea.  Just bugs me that I mis-typed the size of the TVS.

With Adrian's extensive help (Thanks again, Adrian), we've landed on the beefier TVS as shown in his original post of the back of the TIU with the four yellow sketched diodes. His daughter card design is probably the "silver bullet" fix but isn't exactly practical in a production sense. That's why we are going with the four larger TVS's attached to the signal transformer on the back of the TIU PCB. 

All new Rev L TIU's produced going forward will contain these TVS's. In the field we have a Service Bulletin that was released today, in fact, that outlines removing the old mod (wimpy diodes) with the larger ones. While Adrian noted that there were still occasional failures, it's a balance of fixing it without re-designing and laying out a new PCB vs. applying a solution that gets most TIU's serviceable long-term. 

From MTH's Service perspective, we will be adding this mod to all TIU's that arrive in for service with blown xmit chips or any condition that would indicate such. Additionally, this mod can be applied to any Rev TIU you have in your inventory. The vast majority of TIU's we've seen come through Service as well reports from the field have been on Rev L TIU's. Only the Rev L TIU's have the old mod applied. 

For those of you who are ASC's you will have this service bulletin. I've also shared it with Barry Broskowitz so he can post it on this forum to let those that have the ability apply it to their TIU's. MTH isn't carrying the component at this time so you can get it from Digi-Key per the bulletin.

NOTE - If you are NOT an ASC and you attempt this mod and damage a TIU that is under warranty, it will void the warranty. 

Thanks,

MTH 

MTH posted:

With Adrian's extensive help (Thanks again, Adrian), we've landed on the beefier TVS as shown in his original post of the back of the TIU with the four yellow sketched diodes. His daughter card design is probably the "silver bullet" fix but isn't exactly practical in a production sense. That's why we are going with the four larger TVS's attached to the signal transformer on the back of the TIU PCB. 

All new Rev L TIU's produced going forward will contain these TVS's. In the field we have a Service Bulletin that was released today, in fact, that outlines removing the old mod (wimpy diodes) with the larger ones. While Adrian noted that there were still occasional failures, it's a balance of fixing it without re-designing and laying out a new PCB vs. applying a solution that gets most TIU's serviceable long-term. 

From MTH's Service perspective, we will be adding this mod to all TIU's that arrive in for service with blown xmit chips or any condition that would indicate such. Additionally, this mod can be applied to any Rev TIU you have in your inventory. The vast majority of TIU's we've seen come through Service as well reports from the field have been on Rev L TIU's. Only the Rev L TIU's have the old mod applied. 

For those of you who are ASC's you will have this service bulletin. I've also shared it with Barry Broskowitz so he can post it on this forum to let those that have the ability apply it to their TIU's. MTH isn't carrying the component at this time so you can get it from Digi-Key per the bulletin.

NOTE - If you are NOT an ASC and you attempt this mod and damage a TIU that is under warranty, it will void the warranty. 

Thanks,

MTH 

Very glad to help any way I can!  It makes me happy to think of all the layouts that will benefit from this study!

Hearing direct from the mfg (MTH) on this issue via this forum and its talented members...doesn't get any better

gunrunnerjohn posted:

You probably could Stan.  That would prove the concept, not a bad idea.  Just bugs me that I mis-typed the size of the TVS.

So what will be the latest specked out TVS's for our TIU's?

I notice that the MTH service bulletin specifies 500W TVS parts, I'm using a 1500W one on my little transient suppression board.  Adrian had noted that he had failures of the 500W parts he was using in a similar circuit, so my solution may still be a step up in protection.  Time will tell...

MTH posted:

With Adrian's extensive help (Thanks again, Adrian), we've landed on the beefier TVS as shown in his original post of the back of the TIU with the four yellow sketched diodes. His daughter card design is probably the "silver bullet" fix but isn't exactly practical in a production sense. That's why we are going with the four larger TVS's attached to the signal transformer on the back of the TIU PCB. 

If I understand this MTH statement, and combining with earlier discussion, the daughter-card approach using a handful of 1-cent diodes is the superior technical solution.  It eliminates the problem period, end-of-story.  OTOH, the single TVS method might fail over time particularly in rough operating environments...but doing-the-numbers makes this a better business solution.  I get that.

That said, and this is just my opinion, if you are going to make the mod yourself...or have someone do it for you, I'd find someone willing/able to install the daughter card mod (or GRJ's proposed alternative if that proves out).  The difference in component cost is negligible.  I'd think labor/shipping/overhead of making the mod is something you'd only want to do once.

If clubs ran unmodified TIUs in passive mode with a choke, would/could this issue still pop up?

Not to take anything away from this work which seems very promising for clubs!

Engineer-Joe posted:

If clubs ran unmodified TIUs in passive mode with a choke, would/could this issue still pop up?

Not to take anything away from this work which seems very promising for clubs!

Yeah it would. It's not about high current, it's about the di/dt voltage blowing stuff up.

Engineer-Joe posted:

If clubs ran unmodified TIUs in passive mode with a choke, would/could this issue still pop up?

Not to take anything away from this work which seems very promising for clubs!

Apparently Adrian's club did run in passive mode and still managed to cook TIU's.

"From MTH's Service perspective, we will be adding this mod to all TIU's that arrive in for service with blown xmit chips or any condition that would indicate such. Additionally, this mod can be applied to any Rev TIU you have in your inventory."

So I'm clear, MTH, this new device would benefit Rev G, H, & I TIUs by adding protection.  Yet my ancient Rev G's, which have suffered through many derailments, don't seem to have lost signal strength.  What would you do with pre-Rev Ls that work fine and have been rock solid (at least since MTH repaired the Rev G's some 10-15 years ago).

RJR posted:

hat would you do with pre-Rev Ls that work fine and have been rock solid (at least since MTH repaired the Rev G's some 10-15 years ago).

Don't fix stuff that ain't broke!

At least that's my advice.   If you've had the TIU for 10-15 years and it's still going strong, I wouldn't open it up just for this mod. 

Thanks, GRJ.  You do make sense.  But, given history, it should be done on a Rev L.  I have a Rev L to practice on, because it has failed so many times and been in the shop so many times I've given up on it--latest is that it won't control or even cut track voltage, and I won't stuff it with any more $$.

I will do my mod on my TIU's, as soon as I get the proper board layout.  I screwed up the first one.

Those of us who are human, and admit it, are entitled to an annual screw-up to prove our humanity.

Are you saying you intend to do all your TIUs, even pre-RevL, with the "official" MTH fix?

gunrunnerjohn posted:

Engineer-Joe posted:

If clubs ran unmodified TIUs in passive mode with a choke, would/could this issue still pop up?

Not to take anything away from this work which seems very promising for clubs!

Apparently Adrian's club did run in passive mode and still managed to cook TIU's.

WOW.

Apparently Adrian's club did run in passive mode and still managed to cook TIU's.

I find myself scratching my head over this statement. My understanding is that a short circuit won't affect a TIU in Passive Mode. Is this not true for a current spike, as well?

Barry Broskowitz posted:

Apparently Adrian's club did run in passive mode and still managed to cook TIU's.

I find myself scratching my head over this statement. My understanding is that a short circuit won't affect a TIU in Passive Mode. Is this not true for a current spike, as well?

It's not current that does the damage. It's the di/dt voltage. (here's a tutorial). The train is a motor (inductive). When the current jumps from the short, the 1st derivative, scaled by the inductance is developed across the rails.

An ideal short is a scaled step function Ks I(t)

where I(t) = 0 on the interval t<0,   I(t)=1 on the interval (t>0) and Ks is a scale for the current:

So the first derivative is the dirac (delta) function centered at t=0;

V(t) = d KsU(t)/dt

V(t) = Ks dirac(t)

The weight of the delta function voltage is Ks which is the peak voltage the TIU driver experiences. This is what the TVS is there to clamp.

Adrian! posted:

Barry Broskowitz posted:

Apparently Adrian's club did run in passive mode and still managed to cook TIU's.

I find myself scratching my head over this statement. My understanding is that a short circuit won't affect a TIU in Passive Mode. Is this not true for a current spike, as well?

It's not current that does the damage. It's the di/dt voltage. (here's a tutorial). The train is a motor (inductive). When the current jumps from the short, the 1st derivative, scaled by the inductance is developed across the rails.

An ideal short is a scaled step function Ks I(t)

where I(t) = 0 on the interval t<0,   I(t)=1 on the interval (t>0) and Ks is a scale for the current:

So the first derivative is the dirac (delta) function centered at t=0;

V(t) = d KsU(t)/dt

V(t) = Ks dirac(t)

The weight of the delta function voltage is Ks which is the peak voltage the TIU driver experiences. This is what the TVS is there to clamp.

 

I was following this thread until now.

Barry Broskowitz posted:

Apparently Adrian's club did run in passive mode and still managed to cook TIU's.

I find myself scratching my head over this statement. My understanding is that a short circuit won't affect a TIU in Passive Mode. Is this not true for a current spike, as well?

I see Adrian beat me to it, but yes, a voltage spoke will affect any electronics that's exposed to it, it doesn't matter if there is current flowing in that circuit, it just has to be connected.  One statement in his tutorial says it all.

Opening of inductive circuit may cause voltage spikes of thousands of volts!

Think of how the Kettering ignition system system that we have used for a hundred years in cars works, from a lowly 12V battery, we manage to generate thousands of volts to fire the plug.  It's all about the collapsing field in the inductor.

Guys,

Thanks for the information!

Hey Barry sounds like another chapter, if not another book . Adrian even has awoken the sleeping Giant...Lol!

Great bunch of info guys, and I know it is answering questions, that all we've ever done in the past is use the volt/amp meter, and the Signal strength meter, when hunting for signal problems or lack there of.....Thanks to all of you folks on here again.....

The di/dt voltage spike from the motor inductance in O-gauge engines is usually behind the bridge-rectifier and DC bus  capacitance and hence does NOT reach the track.  Instead, it is the inductance of the track itself and the track-to-transformer wiring that "generates" the killer voltage spikes.  

Above video shows my "surge detector" that can be moved around a layout to localize the severity/frequency of voltage spikes.   The 2 red LEDs flash when there is a positive or negative voltage spike on the track.  This widget was discussed in an earlier thread about spikes, TVS placement, etc.   Nothing to do with whether the TIU is in passive or active mode.  Obviously there is no motor inductance in this setup.  Instead the voltage spikes (flashing red pulses) are caused by the rapid current changes from shorting the track which might be from a derailment, going over a switch/turnout, whatever.

Would the problem be less for those of us that operate our layouts using DC power? 

DC power doesn't make you immune to voltage spikes, just one less source of them.

stan2004 posted:

The di/dt voltage spike from the motor inductance in O-gauge engines is usually behind the bridge-rectifier and DC bus  capacitance and hence does NOT reach the track.  Instead, it is the inductance of the track itself and the track-to-transformer wiring that "generates" the killer voltage spikes.  

Above video shows my "surge detector" that can be moved around a layout to localize the severity/frequency of voltage spikes.   The 2 red LEDs flash when there is a positive or negative voltage spike on the track.  This widget was discussed in an earlier thread about spikes, TVS placement, etc.   Nothing to do with whether the TIU is in passive or active mode.  Obviously there is no motor inductance in this setup.  Instead the voltage spikes (flashing red pulses) are caused by the rapid current changes from shorting the track which might be from a derailment, going over a switch/turnout, whatever.

 

Yeah the cabling/track inductance makes more sense than the motor inductance after spending 20 seconds thinking about the circuit conditions during a short.

Also...  You know those chokes all of us are putting in series with the power supplies to keep the RF impedance high? Under a short circuit transient those are going to cause a massive di/dt event at the TIU too...

1. It's inline with the full short circuit current pulse

2. compared to a 1KV transient... either side of the transformer is like a solid ground. The impedance is low looking into it (which is why the choke was there in the first place), and on the KV scale (ground) or (ground + 18V) are the same thing.

3. So the transformer side of the choke is basically pegged to low potential, and the choke has a KV across it, meaning the TIU/track will have a KV across it.

 Maybe I'll be putting a TVS in parallel with those too in the future.

Interesting thought about the chokes, I never gave that a thought!

"You know those chokes all of us are putting in series with the power supplies to keep the RF impedance high?"

Adrian, please tell me more about this. I'm not one of the all.  I wonder if it would eliminate cross-talk (doubletoot) between 2 TIUs?????

stan2004 posted:

The di/dt voltage spike from the motor inductance in O-gauge engines is usually behind the bridge-rectifier and DC bus  capacitance and hence does NOT reach the track.  Instead, it is the inductance of the track itself and the track-to-transformer wiring that "generates" the killer voltage spikes.  

Above video shows my "surge detector" that can be moved around a layout to localize the severity/frequency of voltage spikes.   The 2 red LEDs flash when there is a positive or negative voltage spike on the track.  This widget was discussed in an earlier thread about spikes, TVS placement, etc.   Nothing to do with whether the TIU is in passive or active mode.  Obviously there is no motor inductance in this setup.  Instead the voltage spikes (flashing red pulses) are caused by the rapid current changes from shorting the track which might be from a derailment, going over a switch/turnout, whatever.

 

I don't know why I found this so entertaining? but I did. Although it seems crazy to do this it does represent what can happen during a derailment.

You must be using MTH power supplies though? The (big L's) PH180s breaker trips if the weather changes ........or if you stare at the track too hard...… get it?

I think of the people designing things and always wonder what they think is abusive? I watched our 2 grandsons play tug of war with a brand new MTH RTR steamer and tender that I bought them. They were yelling it's my turn to run it!

 …..and I come here and read of users complaining that the wireless tether breaks too easily.

Adrian! posted:

 

Also...  You know those chokes all of us are putting in series with the power supplies to keep the RF impedance high? Under a short circuit transient those are going to cause a massive di/dt event at the TIU too...

 

Good point.  If you open the TIU you will see such a choke at the Voltage In connectors.  A short circuit current on a TIU output channel flows thru the TIU choke.  So the di/dt phenomenon, for lack of a better term, is the same as the passive-mode TIU when adding the choke between the PSX-AC and the track.  Same choke application - to keep the RF impedance high to protect the DCS-signal integrity.

Engineer-Joe posted:

...

You must be using MTH power supplies though? The (big L's) PH180s breaker trips if the weather changes ...

As I see it, the breaker/fuse method is a completely separate issue.  It doesn't matter how fast or type of power supply interruption method is used.  Here's my thinking.  A short-circuit consists of a rapid increase in current when the short starts - this is one di/dt event.  At some point, the breaker responds to the overcurrent and then interrupts the current causing a rapid decrease in current - this is another di/dt event.

Perhaps it was explained in one of the various threads, but it still baffles me as to what's so special about Adrian's club layout that causes the TIU signal level failures with clock-like repeatability.  Do they really have that many more short-circuit derailments per day that any other layout?  Or maybe no one has undertaken the (considerable) time/effort to gather data.

Another source of di/dt induced spikes is momentary losses of track power going over a section of dirty track of whatever.  That is, several Amps of current flows from track thru the wheels/rollers to the moving engine.  Dirty track, rollers, whatever cause brief losses of contact which causes track current to drop to zero (a di/dt event) and then instantly jump back up to whatever current (another di/dt event).  The size of the current change may be less than that of a short-circuit (10 Amps or whatever) but then these micro-interruptions are surely occurring more frequently than derailments.  So is this a case of death-by-a-thousand-cuts?  

You make good points Stan, dirty track, poor connections between power districts, or even bad track joints could cause spikes every time a train passes.

Theory versus practice, you guys are going to start scaring people.  Adrian did post an event sheet somewhere, maybe a different post.  Yes it was insane, almost  two or more derailments a minute, some time simultaneously.  If we step back, no one else was reporting this kind of problem other then large clubs.  TIUs do have failures so this is occurring, but folks plug stuff in the wrong holes, have major events, or just loose one channel out of 4 even thought they are using 4 channels.  I said this earlier, I have not seen this kind of issue, and there has only been a few, maybe 2 TIU I could not repair, and I have NEVER Replaced a TVS on a Mod Rev L.  NO other TIU have these TVS.  Having said that, maybe one of the two not repaired had a bad TVS.  So the average home is not having this problem.

So to start talking about all these di/dt events from dirty track, dirty wheels, bad pickup rollers is going to drive people scared.  Based on your theory I should be repairing TIU at the same rate as engines, and that is not even close.  Maybe .5% of my work is TIU, if that; and other techs send people to me.  Why don't we see how MTH fix works, plus I am sure some (Clubs and others) will buy Johns fix.   G

George,

Well said!

GGG posted:

Theory versus practice, you guys are going to start scaring people.

I got rid of all my older TIUs and now have only Rev L versions (two primary and one backup).  Following this topic got me wondering if I should sell my Rev Ls and go back to older ones.  Now G has me wondering if I should keep my Rev Ls and hope for the best (he did state an unfixable TIU may of had this problem).

Here is my power setup for each TIU channel:

I'm using more power but have a much faster CB (PSX) and a large choke (34A 22uH).  For you experts that know what you are talking about, is this setup more or less susceptible to the failure you are talking about.  If it is more susceptible I may get rid of the Rev Ls.

GGG posted:

Theory versus practice, you guys are going to start scaring people.  Adrian did post an event sheet somewhere, maybe a different post.  Yes it was insane, almost  two or more derailments a minute, some time simultaneously.  If we step back, no one else was reporting this kind of problem other then large clubs.  TIUs do have failures so this is occurring, but folks plug stuff in the wrong holes, have major events, or just loose one channel out of 4 even thought they are using 4 channels.  I said this earlier, I have not seen this kind of issue, and there has only been a few, maybe 2 TIU I could not repair, and I have NEVER Replaced a TVS on a Mod Rev L.  NO other TIU have these TVS.  Having said that, maybe one of the two not repaired had a bad TVS.  So the average home is not having this problem.

So to start talking about all these di/dt events from dirty track, dirty wheels, bad pickup rollers is going to drive people scared.  Based on your theory I should be repairing TIU at the same rate as engines, and that is not even close.  Maybe .5% of my work is TIU, if that; and other techs send people to me.  Why don't we see how MTH fix works, plus I am sure some (Clubs and others) will buy Johns fix.   G

Okay.... there's a balance to be had here I think. It's totally true that not everyone has a TIU failing, actually it's probably a very small percentage of folks that have the same problem our club did. DCS will work fine if you aren't abusive to it. I'm not going to kick out all our under 12 members...  so we had to do something else to solve it.

That said, if your DCS system is acting unresponsive, and you aren't happy with it, and you did buy Barry's book,  and you already looked into the basic stuff (wiring, antenna, batteries, loose radio board...), a shorted TVS or damaged ACT244 is definitely high on the list of things to look at. If it happens to you the first time, it means you're someone that the modification would benefit going forward.

The catch is you pretty much have to have an oscilloscope to really know if it's a problem or not. You can get signal tests of 10 on the remote anywhere from the normal 14V down to about 3-4V.