Weak or NO DCS signal from TIU

the revision with which your talking about was only for the rev L tiu older ones were built slightly different and did not have the same problem with 244 ic's  but rather with the max 4639!

Alan

If you have varying signals on the channels, that's likely the ACT244 chips.  The MAX 4639 would likely be a digital event, either no signal or normal operation.

Does anyone have the correct Digi-Key part number to replace the ACT244 chip. I have one channel functioning with Variable voltage but will not emit a DCS signal.

Drop me a line and I can fix you up, I have tons of them in stock.

It is not as simple as you think.  I have seen all sorts of combinations of defects have an effect on the TIU function.  In the end the only thing that matters is can the TIU transmit and RECEIVE the signal back.  I have seen the TIU transmit the initial watch dog pulse, but not the longer sustained one once the TIU finishes starting up and flash the LED ID #.  I have seen a single transmit chip effect other channels, I have seen partial receive chip function, I have seen receiver chip effect Transmit output, and the combination goes on and on.

The MTH guidance in ASC manual was before an actual device to look at output.  Techs did not have oscilloscope (except John:-)).  So it was a logic chart.  Early models more susceptible to receiver failure.  Rev L more susceptible to Transmitter failures.  BUT I do see both on Rev L and sometimes together.

The simplest is when one channel does not work but all the others do.  The worst is none work and LED barely lights or the current draw is high just TIU.  This means a short that is effecting the Power Supply.  So you can pull all the transmit chips and if still present pull the receiver.  If still shorting change the 5V reg, if still shorted, look for round object with lid and deposit TIU in the object.

@GGG posted:

If still shorting change the 5V reg, if still shorted, look for round object with lid and deposit TIU in the object.

I deposit them in my dead TIU box for spare parts.

@gunrunnerjohn posted:

I deposit them in my dead TIU box for spare parts.

A veritable graveyard 😉

From this post and others, I have picked up many good suggestions for trouble shooting TIU issues.  It has been interesting trying to keep the clubs 14 TIUs (8 active, 6 spare) operational. summarized below are some of the TIU trouble shooting problems and" POTENTIAL" hardware causes I have picked up from the various threads.  Before addressing a potential hardware problem make sure that all TIU software setting are correct.

-Variable channels with low or no track power output, potentially a bad power FET (IRFZ44)

-Track power short on the output of any of the TIU channel, most likely a bad TVS which has shorted out (1.5KE33)

-Low or no DCS signal output on any one channel, potentially a bad transmit chip (ACT244 )

-No DCS signal on any output channel, potentially a bad receiver chip (MAX 4639). The Max4639 receives only track signal from a DCS engine

If anyone wants to add, correct or update the above list, please do.

Thanks to all the contributors,

Bob D

NJ-Hi Railers

Add one more,

Variable channels with constant ON track power output, most likely a bad TVS which has shorted out (1.5KE51). These are in parallel to the MOSFETS so when they short, even if the MOSFET is off, you get the full voltage out bypassing the control by the MOSFETS.

FWIW- Zcontrollers have the same (1.5KE51) in the same arrangement in parallel to the variable control MOSFETS and have the same result- constant output regardless of handle position when that TVS fails shorted. Picture and details here https://ogrforum.com/...4#154167236021571964

My method is to use a side cutter to clip one end of the TVS diode leg and then check with a meter when you have one suspected shorted. This removes it from the circuit but still allows you to test the device with a meter. Again, the problem is, we have more than one device in parallel that might have failed- so disconnecting the cheapest and easiest one to replace (2 leg diode VS 3 leg MOSFET).

-No DCS signal on any output channel, potentially a bad receiver chip (MAX 4639). 

Not true, but possible.  Channel that transmit, but can't receive return signals can be the receive chip.   G

@GGG posted:

-No DCS signal on any output channel, potentially a bad receiver chip (MAX 4639).

Not true, but possible.  Channel that transmit, but can't receive return signals can be the receive chip.   G

Thanks GGG for commenting.  Can you elaborate?

I have a TIU that will transmit the watch dog signal on all channels when I first turned on TIU power.  But when I use the hand held, both tethered & wireless, I don't  get any TUI output signal.  Also, can't change the TIU number.  I changed the MAX 4693 receiver chip which seems to be common to all channels, but that didn't help.  Are there any other  components  that are common to all channels that would prevent an output signal when the hand held is used?  

Thanks,

Bob D

Not being able to change ID address or communicate with TIU has nothing to do with the DCS transmit and receive chips that communicate with the engine.  Sounds like some other fault, and that may not be repairable.

If a TIU can transmit the watchdog.  The transmit chips are probably fine.  If you can not load an engine or receive a DCS signal strength response from an engine already loaded, that could be receiver chip.  For example an engine loaded can be started.  But can't respond when necessary.  Transmit fine, receive not.  Being able to load an engine requires a transmit signal, and the receive signal from the engine.  G

How about a scenario where: the engine will load and start. Engine will run, cannot get engine lights to respond and horn/whistle may be intermittent. Smoke takes multiple remote or app hits to make a change. Test signal only reading 6-7. ACT244 chip? 2 other TiU channels in use read 10s with no issues.

The only time I get a signal reading on the scope of any type, is when I first power up the TIU.  When I use the handheld tethered or wireless, I don't see any output signal response on the scope for any of the channels, not transmitting at all.  That's why I thought the problem might be a faulty MAX4693 receiver chip which was not receiving commands from the handheld, so the TIU would not transmit anything out and not execute internal commands to the TIU.    Is there a way to trace the signal in the TIU or does anyone have a schematic of the TIU.  Any suggestions?

Thanks

Bob D

The MAX4693 has nothing to do with the wireless remote communication, it's receiving the DCS track signal coming back from the engines.

@Vernon Barry posted:

Add one more,

Variable channels with constant ON track power output, most likely a bad TVS which has shorted out (1.5KE51). These are in parallel to the MOSFETS so when they short, even if the MOSFET is off, you get the full voltage out bypassing the control by the MOSFETS.

FWIW- Zcontrollers have the same (1.5KE51) in the same arrangement in parallel to the variable control MOSFETS and have the same result- constant output regardless of handle position when that TVS fails shorted. Picture and details here https://ogrforum.com/...4#154167236021571964

My method is to use a side cutter to clip one end of the TVS diode leg and then check with a meter when you have one suspected shorted. This removes it from the circuit but still allows you to test the device with a meter. Again, the problem is, we have more than one device in parallel that might have failed- so disconnecting the cheapest and easiest one to replace (2 leg diode VS 3 leg MOSFET).

Thanks for the additional information and this could be a way to resolve the problem of replacing bad MOSFETs when the variable channels are used as fixed channels.  If the two variable channels are always used as fixed channels, and a MOSFET goes bad, any issue of just jumping the TVS so now you are just getting full transformer voltage at the output of the TIU channel?  

@rad400 posted:

Thanks for the additional information and this could be a way to resolve the problem of replacing bad MOSFETs when the variable channels are used as fixed channels.  If the two variable channels are always used as fixed channels, and a MOSFET goes bad, any issue of just jumping the TVS so now you are just getting full transformer voltage at the output of the TIU channel?  

I would not recommend that. The TVS was there in an attempt to protect against voltage spikes. It's not there as a primary current path. If you are having events that would kill the channels (derailments, overloads, shorts, inductive kickback spike events) that caused the original failure and then now passing primary current through a path meant to be for protection seems like a really bad idea.

Ideally, just use passive mode and external inductive choke and bypass the entire power section. and any  power going through the TIU.

I have replaced only 1 of those TVS during my career of fixing TIU.  As far as jumpering the VAR Fet for constant Fix, yes it can be done.  Believe a few clubs do that.  What you loose is emergency Kill feature since only Fix channels protected by Relay.  The VAR turn off the FETs, which are now jumpered.

You did read my post about the receiver chip versus your issue of non communication with TIU?   The remote and TIU have a transceiver board for communication between remote and TIU.  Engine is done over the track.  G

@GGG posted:

You did read my post about the receiver chip versus your issue of non communication with TIU?   The remote and TIU have a transceiver board for communication between remote and TIU.  Engine is done over the track.  G

G - Yes, thanks for the clarification.   ACT244 transmit chip & MAX 4639 receive chip handles TIU to track/engine communication.  The plug-in transceiver board handles the air link communication between TIU and handheld.  

Where does a tethered hand held get connected into?

Thanks, Bob D

In terms of strength of signal, how important is it to have all track power lockons in the same orientation? What I mean is: orient the lock-on (Realtrax in my situation) so that say the outside rail is always carrying the signal back to the TIU. And then likewise, any isolated outside rail sections triggering signals etc, be the opposite outside rail than the one hooked to the lock-ons? I know this is NOT the case on my layout, and I’m wondering if this may be contributing to my poor DCS signal??

For Realtrax it's not important at all.  Since each piece has the outside rails jumpered, both sides have the same signal.

Failures here are like

90% ACT244  

5% the output transformer

5% the TVS (if you upsized them and added the beefy clamp circuit)

if you double stack the 244 then those driver failures become a lot less common.

@Adrian! posted:

Failures here are like

90% ACT244  

5% the output transformer

5% the TVS (if you upsized them and added the beefy clamp circuit)

if you double stack the 244 then those driver failures become a lot less common.

Hi Adrian

Can you elaborate on your TVS comment  "if you upsized them and added the beefy clamp circuit"

Thanks, Bob D

@rad400 posted:

Hi Adrian

Can you elaborate on your TVS comment  "if you upsized them and added the beefy clamp circuit"

Thanks, Bob D

Uh, if you go back to the original thread when we figured all this out together we had two fixes.

One was to upsize the TVS from the tweezer sized ones MTH originally had in there to a larger part (You want a larger one that can take a hit.... but not one so large that the added junction capacitance shunts out the signal swing... that's how we arrived at the  SA5.0A part).

Two was to add a "beefy clamp" out made out of rectifier diodes in parallel. The TVS clamp fast, but since they're a reverse junction device with a tiny junction area they don't do well with sustained currents (like more than a few ns long). Regular rectifier diodes (like the 1n4148 we used in that post) are fine with sustained current (they're made for rectifying in power supplies) but being a forward junction device have a super duper long recombining time of like 8-10ns and don't turn on super fast.

So you put the rectifiers in parallel and have them at a slightly lower effective threshold voltage than the TVS. That way if a long event happens you still get the fast clamping of TVS, but 10ish ns in, the current is handed over from the TVS device to the "beefy clamp" so the TVS doesn't fail from sustained current.

Once you do this, the TVS parts fail very rarely. I think we blew 2 total in 4 years on 30 TIU channels total. Our club is basically the railroad equivalent of a demolition derby so the setup is definitely stress tested.

Arriving late to this party, but hopefully the answer to my questions will help others of us less sophisticated circuit tweekers.  Thanks as always for your shared expertise.

1. Can the SA5.0A "primary" TVS alternate be installed across the external terminals of the TIU channel?  If so, the output side (or input)?

2. Can the added "beefy clamp" be the two 1n4148 diodes, in parallel but opposite polarities, likewise be across external channel terminals, output (or input) ?

3. Finally, do these steps all apply to any Rev L TIU, or only certain ones after a certain production date?

@cnwdon posted:

Arriving late to this party, but hopefully the answer to my questions will help others of us less sophisticated circuit tweekers.  Thanks as always for your shared expertise.

1. Can the SA5.0A "primary" TVS alternate be installed across the external terminals of the TIU channel?  If so, the output side (or input)?

2. Can the added "beefy clamp" be the two 1n4148 diodes, in parallel but opposite polarities, likewise be across external channel terminals, output (or input) ?

3. Finally, do these steps all apply to any Rev L TIU, or only certain ones after a certain production date?

Hi there,

1. So the transient events the TVS protects against from all the layout sparks and stuff are like 1ns events, which means they are physically about 2 inches long in air (considering 1/4wavelength), and more like one inch long in a cable where the dielectric is 2,3,4 ish (depending on exactly what wire).

What this means is in a 1ns world (not DC and not 60 Hz, but up at GHz land where this stuff is propagating) the TVS is only clamping/limiting the voltage around it within 1 inch of the device, so if you put it out at the terminals you're not doing an amazing job of protecting the other parts on the PCB (upstream driver and logic). You really want it right at the output of the ACT244 where the track connection (where the transients originate) meet the electronics of the TIU board.

2. The beefy clamp is actually a set of series 1n4148 diodes (Stacked to get the right threshold (n x Vf) with two strings in parallel but opposite directions "anti-parallel". You need to cover both directions since a transient may have a positive or negative polarity relative to the ground in the TIU. TVS are also like this, some are + only and some are - only, and some like the SA5.0A has two inside in anti-parallel to cover both directions.

3. Protection is protection. It's not really about rev. You have the angry spark machine that is a train layout connected to the TIU electronics. The better you stop over-voltage stress from being applied to the electronics the less likely and less frequent the failures will be.  We went from blowing a TIU channel weekly to maybe blowing one a year when we made those changes.

@Adrian! posted:

Uh, if you go back to the original thread when we figured all this out together we had two fixes.

One was to upsize the TVS from the tweezer sized ones MTH originally had in there to a larger part (You want a larger one that can take a hit.... but not one so large that the added junction capacitance shunts out the signal swing... that's how we arrived at the  SA5.0A part).

Two was to add a "beefy clamp" out made out of rectifier diodes in parallel. The TVS clamp fast, but since they're a reverse junction device with a tiny junction area they don't do well with sustained currents (like more than a few ns long). Regular rectifier diodes (like the 1n4148 we used in that post) are fine with sustained current (they're made for rectifying in power supplies) but being a forward junction device have a super duper long recombining time of like 8-10ns and don't turn on super fast.

So you put the rectifiers in parallel and have them at a slightly lower effective threshold voltage than the TVS. That way if a long event happens you still get the fast clamping of TVS, but 10ish ns in, the current is handed over from the TVS device to the "beefy clamp" so the TVS doesn't fail from sustained current.

Once you do this, the TVS parts fail very rarely. I think we blew 2 total in 4 years on 30 TIU channels total. Our club is basically the railroad equivalent of a demolition derby so the setup is definitely stress tested.

I could read up, but I thought I'd just ask. I highly respect your work here.

What are you using for power supply to the rails?

I only run a fraction of the time. I have both Lionel PH180s and a MTH Z4000 connected to my O scale, and Bridgewerks DC connected to my G scale, all with 10 amp auto fuses connected inline. I run large consists all the time.

After maybe 15 + years, I've never had a TIU failure. So I'm wondering what's the difference? Is it mainly the amount of trains running? Something else?

I'm always tempted to make these mods and then I don't.

I appreciate the great work and I expect to see this show up in future MTH hardware releases.

Please don't take this the wrong way Adrian. Just thinking out loud.

BTW, I have blown a lot of ten amp fuses!

I came up with a design with Schottky diodes and a 1500W TVS to attempt to mitigate this issue right at the chip.  It clearly hasn't received wide field testing, I can only say one TIU with these installed hasn't eaten any 'ACT244 chips.   This was discussed in another thread, but didn't receive enough interest to invest in a production run.

The module is glued to the top of the 72ACT244 chps and wired to the four pins directly in line with the module connections.  I just stuck them down with Aleene's Tacky Glue, the glue is just to keep them in place until you solder the wires.  I didn't think to take pictures of the TIU when they were installed...

@Adrian! posted:

Hi there,

1. So the transient events the TVS protects against from all the layout sparks and stuff are like 1ns events, which means they are physically about 2 inches long in air (considering 1/4wavelength), and more like one inch long in a cable where the dielectric is 2,3,4 ish (depending on exactly what wire).

What this means is in a 1ns world (not DC and not 60 Hz, but up at GHz land where this stuff is propagating) the TVS is only clamping/limiting the voltage around it within 1 inch of the device, so if you put it out at the terminals you're not doing an amazing job of protecting the other parts on the PCB (upstream driver and logic). You really want it right at the output of the ACT244 where the track connection (where the transients originate) meet the electronics of the TIU board.

2. The beefy clamp is actually a set of series 1n4148 diodes (Stacked to get the right threshold (n x Vf) with two strings in parallel but opposite directions "anti-parallel". You need to cover both directions since a transient may have a positive or negative polarity relative to the ground in the TIU. TVS are also like this, some are + only and some are - only, and some like the SA5.0A has two inside in anti-parallel to cover both directions.

3. Protection is protection. It's not really about rev. You have the angry spark machine that is a train layout connected to the TIU electronics. The better you stop over-voltage stress from being applied to the electronics the less likely and less frequent the failures will be.  We went from blowing a TIU channel weekly to maybe blowing one a year when we made those changes.

Thank you.  I think I'm getting there, but to be perfectly concrete about how to do this:

1. Follow the leads from the track outputs internally to their outputs from the board, and install both of these "devices" across the hot and neutral outputs at that location or as close as we can get it packaged in there. One could create a package of the primary TVS and the beefy clamp, with a shared terminal point at each end, so as to have just one soldered connection to the board outputs?

2. Are you saying the "beefy clamp" is two pairs of 1n4148 diodes, each pair "nose to nose" with say positive sides facing each other and negatives at each end, these two pairs installed in parallel beside each other?

@cnwdon posted:

Arriving late to this party, but hopefully the answer to my questions will help others of us less sophisticated circuit tweekers.  Thanks as always for your shared expertise.

1. Can the SA5.0A "primary" TVS alternate be installed across the external terminals of the TIU channel?  If so, the output side (or input)?

2. Can the added "beefy clamp" be the two 1n4148 diodes, in parallel but opposite polarities, likewise be across external channel terminals, output (or input) ?

3. Finally, do these steps all apply to any Rev L TIU, or only certain ones after a certain production date?

A better answer to your first question and with a doodle:   TVS Placement

@Adrian! posted:

A better answer to your first question and with a doodle:   TVS Placement

Thank you.  The location principle is now understood.  Sorry, but my brain is in “concrete” rather than abstract mode with the above questions, so I’m still unsure about the practical application and packaging-in-the-TIU nuts and bolts.

1. Follow the leads from the track outputs internally within the TIU to their outputs from the board, and install both of these "devices" across the hot and neutral outputs at that location or as close (well under 1 inch) as we can get it packaged in there. One could create a package of the primary TVS and the beefy clamp, with a shared terminal point at each end, so as to have just one soldered connection to the board outputs?  Use of a low wattage soldering pen, low melting point solder, safest least corrosive flux, heat sinks, etc are familiar skills.

2. Are you saying the "beefy clamp" is two pairs of 1n4148 diodes, each pair "nose to nose" with say positive sides facing each other and negatives at each end, these two pairs installed in parallel beside each other?  I’m trying to translate your description of this in the post above.

Again, thanks

@cnwdon posted:

Thank you.  The location principle is now understood.  Sorry, but my brain is in “concrete” rather than abstract mode with the above questions, so I’m still unsure about the practical application and packaging-in-the-TIU nuts and bolts.

1. Follow the leads from the track outputs internally within the TIU to their outputs from the board, and install both of these "devices" across the hot and neutral outputs at that location or as close (well under 1 inch) as we can get it packaged in there. One could create a package of the primary TVS and the beefy clamp, with a shared terminal point at each end, so as to have just one soldered connection to the board outputs?  Use of a low wattage soldering pen, low melting point solder, safest least corrosive flux, heat sinks, etc are familiar skills.

2. Are you saying the "beefy clamp" is two pairs of 1n4148 diodes, each pair "nose to nose" with say positive sides facing each other and negatives at each end, these two pairs installed in parallel beside each other?  I’m trying to translate your description of this in the post above.

Again, thanks

1. There's a doodle in the original thread about where to solder the TVSs too that provide good protection nice and close to the drivers. The beefy clamp rectifiers can be further away since they only kick in for sustained events, and don't need to respond quickly (by scaling 10ns is 10")...

2. So the TVS voltage we used is 9.2V. Then the Beefy clamp is two parallel branches of 13 series 1n4148s so you get an effective forward drop of 0.7V X 13 = 9.1V. This is what you want, the beefy clamp to have a slightly lower turn on voltage, so that in longer pulses, the sustained current is handed over from the TVS to the rectifiers.

https://ogrforum.com/...8#159941806875170498

If anybody has a photo of how this device(s) is attached, please share if possible. I have built the boards but am unsure of the solder locations inside the TIU

The board is placed directly on top of the 74ACT244 with the large TVS diode down.  As I mentioned, I use a glue to keep it in place so I can wire it.  That allows all the numbered connection points to line up with the appropriate pins on the driver chip, see pin layout below.

Notice how the labeled connections line up with the actual pin numbers on the 74ACT244 driver IC.  I used fine #30 solid wire to connect the transient suppression board to the driver.

@Adrian! posted:

1. There's a doodle in the original thread about where to solder the TVSs too that provide good protection nice and close to the drivers. The beefy clamp rectifiers can be further away since they only kick in for sustained events, and don't need to respond quickly (by scaling 10ns is 10")...

2. So the TVS voltage we used is 9.2V. Then the Beefy clamp is two parallel branches of 13 series 1n4148s so you get an effective forward drop of 0.7V X 13 = 9.1V. This is what you want, the beefy clamp to have a slightly lower turn on voltage, so that in longer pulses, the sustained current is handed over from the TVS to the rectifiers.

I think I've been missing something fundamental.  The track voltage is nominally 19v or a bit less.  The TVS and the beefy clamp can't be across that circuit pair with ~9v thresholds or they would be conducting and breaking down immediately given the peaks of the AC track voltage, wouldn't they? So the "driver" outputs you state must be DCS signal outputs that have nominal voltage across the driver outputs of less than 9 volts, is that what I've been missing?  But still can't understand since the TIU outputs have track voltage of 19v, and we've been talking about those output terminals as a possible (but not effective) place for the TVS and beefy clamp to be connected across.  I know I'm missing something critical but it's not lighting the bulb yet.  Also, I looked in the earlier TVS Placement thread, but can't find a diagram that seems to match what you described re: parallel sets of 13 diodes in series.

I must seem rather dense, or "thick" as the Brits say, and I surely am feeling that way  

@cnwdon posted:

I think I've been missing something fundamental.  The track voltage is nominally 19v or a bit less.  The TVS and the beefy clamp can't be across that circuit pair with ~9v thresholds or they would be conducting and breaking down immediately given the peaks of the AC track voltage, wouldn't they?

You're indeed missing something.  We're protecting the driver chips, but they're 5V parts!  They don't connect directly to the track, there's a host of analog logic and a coupling transformer before the signal gets to the rails.

So GRJ, are there 4 74ACT244 chips in each TIU?

(So I need 4 boards per TIU?)

Correct, there is a 74ACT244 for each channel to generate the DCS output signal.

And my personal first, I just had a TIU repair where the TIU came in totally dead. By dead, I mean the power light would not even come on with AUX power, USB power, and definitely not Fixed 1 power. Turns out, they killed all 4 of the 74ACT244 so badly that they shorted the 5V rail. Only after removing all 4 did I get the logic to power and boot. Removing each one, I kept getting a little less droop on the 5V rail but confirmed all 4 were damaged. This thing had a really bad abusive day.