Barry, you young kids 
don't know first hand about the early days when semiconductors first came out. Failure to use a heat sink was a sure way to kill a component. Back in the 50's I ran RC boats, and often used diodes and other semiconductors to make equipment do what I wanted it to do.
Robert,
A very good friend of mine told me, just the other day, "Welcome to the 21st century." 
RJR posted:Barry, you young kids
1950's transistors and diodes were point contact devices, and they were VERY sensitive to heat and even shock. The junction transistor and diodes soon replaced them and were way more rugged devices.
Thank you, GRJ, for restoring my faith in my memory
.
I remember the CK722 PNP transistor very well, it was the first one I ever got my hands on! I recall they were around $3 in 1955, and that was a lot of money for me at the time! They were very pretty with the blue case. 
stan2004 posted:RJR posted:
On page 3 of the Service Bulletin, what is the ID number(s) for the ACT244s?
What channels do the numbered ACT244s control?
That is, does U340 control Variable 1, etc?
This should help Susan. You might have to click on the graphic to expand it.
Attachments
Perfect! Thanks
Hi is anyone going to order the pc boards for the 1n4148 diodes for the tiu upgrade like Adrian suggested, I would be interested in purchasing some also!
Alan Mancus
You obviously didn't see my much simpler board to address the issue. The Gerber files and schematic are attached. I do these in .031 thick boards. You simply glue this to the top of the ACT244 with the large TVS down and run the four wires from the indicated connections to the matching pin of the ACT244. The large TVS diodes is across the power/ground, and the four Schottky diodes couple the voltage limiting to the pins while limiting the stray capacitance from the TVS.
Obviously, you orient this board so that the connections on this board line up with the IC pins!
MTH TIU Signal Protection Gerber.zip
Attachments
Was wondering if this Handheld Scope would be of any value to the average user who was interested in testing, or monitoring their DCS Signal strength on an ongoing basis.
It's price and reviews are reasonable, just not sure if it would cut it when measuring, or viewing DCS Signal strength.
Here's the Amazon Link:
https://www.amazon.com/Oscillo...aps%2C142&sr=8-7
Another option, although a little more in price, but seems a good jump in functionality:
https://www.amazon.com/Hantek-...aps%2C142&sr=8-3
The Hanteks are pretty nice scopes for the price. I got a 5102P 100MHz 2 Channel at Banggood a few months ago. It's my first scope and I am still learning so I don't feel comfortable offering advice on it's use just yet. It is a nice scope though. It was $229 with free shipping from their US warehouse, arrived in about a week. They still have it for the same price, but have added $2.99 for shipping from US warehouse, or If you want to wait for the added shipping time, it's $209 + $1.20 shipping from China.
Santa delivered a 5102P to forum member Rod Stewart this Christmas and he just posted some info on testing a TIU here: TIU Output Signal Patterns Using a Hantek Scope w/Screenshots. He's more experienced with scopes than I am and he likes it. He also added some very helpful and detailed notes on the process he used to test the TIU, which is worth a look no matter which scope you end up with.
Also, Gunrunnerjohn got a Mustool (MDS8207 I think it was?) portable handheld scope to do just what you are wanting to do. He posted about it with info around here somewhere. I believe it was less than $100 (maybe even like $80-$90 or so?). Maybe he will be along to comment on that one and how he likes it so far (and also provide the correct version if I got that wrong).
Then there is the TIU Port Tester that indicates Good, Weak or Bad DCS signal for the TIU channel being tested. It was developed right here on the OGR Forum by Adrian!, Gunrunnerjohn, and Stan2004 and is available in kit form. Thread here: Design of a $10-20 DCS-TIU Port Tester Tool? Info about the kit is included in the thread.
Pat C... posted:Was wondering if this Handheld Scope would be of any value to the average user who was interested in testing, or monitoring their DCS Signal strength on an ongoing basis.
It's price and reviews are reasonable, just not sure if it would cut it when measuring, or viewing DCS Signal strength.
Here's the Amazon Link: https://www.amazon.com/Oscillo...aps%2C142&sr=8-7
Another option, although a little more in price, but seems a good jump in functionality: https://www.amazon.com/Hantek-...aps%2C142&sr=8-3
Unless you have a screaming need for the ultimate in portability, I'd pass on the small one and buy the bench model. They're light and very portable if needed. I bought a single channel 40mhz 'scope on AliExpress for $75 shipped a few months ago.
gunrunnerjohn posted:Pat C... posted:Was wondering if this Handheld Scope would be of any value to the average user who was interested in testing, or monitoring their DCS Signal strength on an ongoing basis.
It's price and reviews are reasonable, just not sure if it would cut it when measuring, or viewing DCS Signal strength.
Here's the Amazon Link: https://www.amazon.com/Oscillo...aps%2C142&sr=8-7
Another option, although a little more in price, but seems a good jump in functionality: https://www.amazon.com/Hantek-...aps%2C142&sr=8-3
Unless you have a screaming need for the ultimate in portability, I'd pass on the small one and buy the bench model. They're light and very portable if needed. I bought a single channel 40mhz 'scope on AliExpress for $75 shipped a few months ago.
Gunny,
Can you post the make & model of the scope, I'd like to keep an eye open, in case a similar deal pops up.
RTR12 thought it was a Mustool (MDS8207), but wasn't sure...
Thanx...
Pat C... posted:Can you post the make & model of the scope, I'd like to keep an eye open, in case a similar deal pops up.
Here's the thread I talked about it in: New Scope/Meter Arrives
gunrunnerjohn posted:You obviously didn't see my much simpler board to address the issue. The Gerber files and schematic are attached. I do these in .031 thick boards. You simply glue this to the top of the ACT244 with the large TVS down and run the four wires from the indicated connections to the matching pin of the ACT244. The large TVS diodes is across the power/ground, and the four Schottky diodes couple the voltage limiting to the pins while limiting the stray capacitance from the TVS.
Obviously, you orient this board so that the connections on this board line up with the IC pins!
So would you say that this TVS diode (due to its size) solution can be expected to last as long as Adrian's 1n4148 solution?
I am torn a bit on which solution to suggest for my club considering they have 5 TIUs all channels in use, and I can get 100 1n4148 delivered from amazon in one day for $5. I also have perfboard on hand. I cannot do the same with The SMD components. I have no issue soldering with a hotair rework station either. I am of course more concerned about what would be a one and done solution. My club (TMB) has serious output issues based on my old 1980s analog scope(100 MHz Textronix)+120Hz lowpass filter based analysis. They loose control of things like the horn on various positions of the layout. I am willing to gamble it might be those little 9v tvs diodes.
Both solutions (Adrian's "beefy clamp" board and your smaller board) are attached to the 74ACT244 right? (or can the "beefy clamp" only be hooked up to the underside of the signal transformers at the points shown in "Service Bulletin 5-22-18A" ?) (or it that just two ends of the same trace?)
Either of these solutions supplant the MTH accepted (Service Bulletin 5-22-18A) fix right?
In general is it a sign of bad TVS diodes if each TIU (PASSIVE MODE) output channel is drawing 0.2Amps? One of main rail lines(several hundred feet long) has a 2 Amp draw with an empty track as soon as the TIUs are connected.
Ryaninspiron posted:In general is it a sign of bad TVS diodes if each TIU (PASSIVE MODE) output channel is drawing 0.2Amps? One of main rail lines(several hundred feet long) has a 2 Amp draw with an empty track as soon as the TIUs are connected.
I believe you are actually answering (can answer) your own question. If the draw goes away with the track dis-connected, it has nothing to do with the TVS.
I believe TVS pieces fail either open or a dead short.
Stay posted. Someone will tell me I'm wrong again.
A .2 amp draw could be some shorting at switches or similar. A 2 amp draw seems more severe.
Does this draw go to different TIU channels if you move it?
If you haven't tried just the TIU to the power source, try it. Of course, if you have ignore this. It's always hard to tell at a glance who has done what.
Ryaninspiron posted:gunrunnerjohn posted:You obviously didn't see my much simpler board to address the issue. The Gerber files and schematic are attached. I do these in .031 thick boards. You simply glue this to the top of the ACT244 with the large TVS down and run the four wires from the indicated connections to the matching pin of the ACT244. The large TVS diodes is across the power/ground, and the four Schottky diodes couple the voltage limiting to the pins while limiting the stray capacitance from the TVS.
Obviously, you orient this board so that the connections on this board line up with the IC pins!
So would you say that this TVS diode (due to its size) solution can be expected to last as long as Adrian's 1n4148 solution?
I am torn a bit on which solution to suggest for my club considering they have 5 TIUs all channels in use, and I can get 100 1n4148 delivered from amazon in one day for $5. I also have perfboard on hand. I cannot do the same with The SMD components. I have no issue soldering with a hotair rework station either. I am of course more concerned about what would be a one and done solution. My club (TMB) has serious output issues based on my old 1980s analog scope(100 MHz Textronix)+120Hz lowpass filter based analysis. They loose control of things like the horn on various positions of the layout. I am willing to gamble it might be those little 9v tvs diodes.
Both solutions (Adrian's "beefy clamp" board and your smaller board) are attached to the 74ACT244 right? (or can the "beefy clamp" only be hooked up to the underside of the signal transformers at the points shown in "Service Bulletin 5-22-18A" ?) (or it that just two ends of the same trace?)
Either of these solutions supplant the MTH accepted (Service Bulletin 5-22-18A) fix right?
In general is it a sign of bad TVS diodes if each TIU (PASSIVE MODE) output channel is drawing 0.2Amps? One of main rail lines(several hundred feet long) has a 2 Amp draw with an empty track as soon as the TIUs are connected.
Mine just sits on top of the ACT244, it's fairly easy to install, but it does require some "fine" soldering to the ACT244 leads. I use #30 wirewrap wire to make the connections. I did a "final" tweak to the layout, but I haven't built any of these yet. Electrically, they're the same as the previous version, I just put some holes for the wires as it makes it a lot easier to keep them from coming loose as you solder the other end to the chip lead. Even with #30 wire, enough heat came up the wire to unsolder it, it was a bit of a pita. With the holes, you can tun the wire through and lay it over, even if it momentarily melts the solder, the wire won't just fall out.
I've only put this on a couple of TIU's, and I don't get the abuse that Adrian's club dishes out, but there have been no failures and no channel degradation. Obviously, it would be nice if it had a wider exposure before declaring it a "tested" fix, but I believe it is doing the job.
Below is the BOM, schematic, and Gerber files if you want to use this. The boards are so small that it's fairly cost effective to get them at OSHPark, and I get them in the .031 material with 2oz copper.
Attachments
Hi there, we have 5 TIUs with the TVS and beefy clamp in parallel. In 3 years we haven't had a single channel degrade or die except, one TIU that burnt down when the building contactor failed and put 400V RMS to the single-phase outlet. It does seem pretty rugged.
The TVS turns on very fast (like 100ns-1us) so it protects you from very fast transient events. The only issue is if you have continuous voltage overshoot it will eventually go into non-reversible breakdown. The 1n4148 clamp is very slow to turn on (like 5us range) and won't protect from the like (1-1000ns) events, but can handle over-voltage pretty much indefinitely. So if a continuous overshoot occurs the TVS clamps quickly and then hands the shunting current over to the 1n4148 after a few us so it doesn't break. That's why the 1n4148 stack needs a slightly lower threshold than the TVS... so in the steady state all the current in in the 1n4148, not the TVS.
Internally at the ACT244, I guess I'm not sure I understand how you get continuous voltage overshoot, how does that happen?
What would you do to my simple design to address this?
You can get some pretty sustained events since the current in the motor coils can stay charged longer than the make and break of sparking wheels in a derailment. This is rare but can happen when like the train derails but sparks a lot, rocks a bit, unshorts, shudders, shorts again , and repeat.... You just need some secondary clamp that works in forward bias mode.
Engineer-Joe posted:Ryaninspiron posted:In general is it a sign of bad TVS diodes if each TIU (PASSIVE MODE) output channel is drawing 0.2Amps? One of main rail lines(several hundred feet long) has a 2 Amp draw with an empty track as soon as the TIUs are connected.
I believe you are actually answering (can answer) your own question. If the draw goes away with the track dis-connected, it has nothing to do with the TVS.
I believe TVS pieces fail either open or a dead short.
Stay posted. Someone will tell me I'm wrong again.
A .2 amp draw could be some shorting at switches or similar. A 2 amp draw seems more severe.
Does this draw go to different TIU channels if you move it?
If you haven't tried just the TIU to the power source, try it. Of course, if you have ignore this. It's always hard to tell at a glance who has done what.
At the TMB club layout the transformer is almost always connected to the track. There are relays that connect the several TIU output channels to different points on the track when a central "DCS Power" switch is turned on. This keeps the TIUs isolated when they aren't being used. What I was trying to convey is that the empty track with no TIU draws no current. When the DCS relays connect the TIUs into the track we suddenly draw about 2 amps according to the z4000. If i start unplugging TIUs the current drops by .2 amps per channel. The TIUs are all run in passive mode.
Should a totally passive mode TIU draw any current through just the output jacks on its own?
Thank you for answering me! Sorry to drag you off course. You know what the wiring is however, I didn't have a clue. I should have researched your profile to see you are in that club. I now get a good idea of what you're doing.
So just one more question to help get you a proper answer from the techs here. Are the TIUs hooked up with chokes?
OK 2 questions, how are the TIUs powered? the relay cuts the track connection and/or power to the TIU?
I have to confess that many years ago I tried passive mode. The signal was weaker and that was a big deal to me. So from then on I did not use it anymore. I never heard of anyone attaching chokes to restore the signal strength back then. There were no version L TIUs back then.
I am tempted to try passive mode again outside on my G scale. I never saw any significant power draw way back when I tried it with one TIU channel. The way you're club is wired, my info wouldn't apply to you here.
Hopefully a tech will bail me out here as I never heard of TIUs wired this way with relays for isolation. I'm tempted to guess at the reasons but it wouldn't help you here and now.
Engineer-Joe posted:Thank you for answering me! Sorry to drag you off course. You know what the wiring is however, I didn't have a clue. I should have researched your profile to see you are in that club. I now get a good idea of what you're doing.
So just one more question to help get you a proper answer from the techs here. Are the TIUs hooked up with chokes?
OK 2 questions, how are the TIUs powered? the relay cuts the track connection and/or power to the TIU?
I have to confess that many years ago I tried passive mode. The signal was weaker and that was a big deal to me. So from then on I did not use it anymore. I never heard of anyone attaching chokes to restore the signal strength back then. There were no version L TIUs back then.
I am tempted to try passive mode again outside on my G scale. I never saw any significant power draw way back when I tried it with one TIU channel. The way you're club is wired, my info wouldn't apply to you here.
Hopefully a tech will bail me out here as I never heard of TIUs wired this way with relays for isolation. I'm tempted to guess at the reasons but it wouldn't help you here and now.
No problem, Thank you, no chokes on the TIUs or transformers, not that I have seen at least. From what I have been reading here we should probably start there. Considering the 12 amp breaker on the Z4000 I imagine the 16amp 22uH chokes mentioned earlier in the thread would be a good idea for starters.
The TIUs are powered by smaller power bricks and those get switched on via a master DCS power relay. Then the main and branch lines each have seperate switches and relays that connect the TIU output channels to the tracks when the club has a DCS day. (The club has issues running DCS and Legacy at the same time so only one system is ever on at once now. That should be a thread of it's own.)
I also do not know the exact reason yet. But I imagine one benefit is that they are subject to derail spikes a third of the time. These design choices were made long before I joined and I haven't asked why yet.
gunrunnerjohn posted:Ryaninspiron posted:So would you say that this TVS diode (due to its size) solution can be expected to last as long as Adrian's 1n4148 solution?
I am torn a bit on which solution to suggest for my club considering they have 5 TIUs all channels in use, and I can get 100 1n4148 delivered from amazon in one day for $5.
Both solutions (Adrian's "beefy clamp" board and your smaller board) are attached to the 74ACT244 right?
Either of these solutions supplant the MTH accepted (Service Bulletin 5-22-18A) fix right?
In general is it a sign of bad TVS diodes if each TIU (PASSIVE MODE) output channel is drawing 0.2Amps? One of main rail lines(several hundred feet long) has a 2 Amp draw with an empty track as soon as the TIUs are connected.
Mine just sits on top of the ACT244, it's fairly easy to install, but it does require some "fine" soldering to the ACT244 leads. I use #30 wirewrap wire to make the connections. I did a "final" tweak to the layout, but I haven't built any of these yet. Electrically, they're the same as the previous version, I just put some holes for the wires as it makes it a lot easier to keep them from coming loose as you solder the other end to the chip lead. Even with #30 wire, enough heat came up the wire to unsolder it, it was a bit of a pita. With the holes, you can tun the wire through and lay it over, even if it momentarily melts the solder, the wire won't just fall out.
..........(clipped for space)...............
I've only put this on a couple of TIU's, and I don't get the abuse that Adrian's club dishes out, but there have been no failures and no channel degradation. Obviously, it would be nice if it had a wider exposure before declaring it a "tested" fix, but I believe it is doing the job.
Thanks for the feedback. I have done a good bit of work with wire-wrapping wire in the past and agree that is certainly an easy way to make the hookup. The club I am in certainly runs a lot of trains both DCS and Legacy, I am hoping to get to the bottom of their DCS issues and I have a feeling this could be it.
Adrian! posted:
Adrian! posted:Hi there, we have 5 TIUs with the TVS and beefy clamp in parallel. In 3 years we haven't had a single channel degrade or die except, one TIU that burnt down when the building contactor failed and put 400V RMS to the single-phase outlet. It does seem pretty rugged.
The TVS turns on very fast (like 100ns-1us) so it protects you from very fast transient events. The only issue is if you have continuous voltage overshoot it will eventually go into non-reversible breakdown. The 1n4148 clamp is very slow to turn on (like 5us range) and won't protect from the like (1-1000ns) events, but can handle over-voltage pretty much indefinitely. So if a continuous overshoot occurs the TVS clamps quickly and then hands the shunting current over to the 1n4148 after a few us so it doesn't break. That's why the 1n4148 stack needs a slightly lower threshold than the TVS... so in the steady state all the current in in the 1n4148, not the TVS.
That sounds like as good of a review as anyone could ask for.
That makes perfect sense. So to wrap things up, in your solution you removed the 4 small SMD TVS diodes added the SA5.0CA in the positions drawn in yellow above for the very fast transients and used the "Beefy Clamp" 1n4148 array board hooked up in parallel to take the load off after 5us thus saving the SA5.0CA indefinitely (or 3 years so far) then. Correct?
and sorry if I missed it but did you mention the science behind why adding a 22uH choke in series with the transformer output would help the DCS signal output of a TIU in passive mode?
gunrunnerjohn posted:You obviously didn't see my much simpler board to address the issue. The Gerber files and schematic are attached. I do these in .031 thick boards. You simply glue this to the top of the ACT244 with the large TVS down and run the four wires from the indicated connections to the matching pin of the ACT244. The large TVS diodes is across the power/ground, and the four Schottky diodes couple the voltage limiting to the pins while limiting the stray capacitance from the TVS.
Obviously, you orient this board so that the connections on this board line up with the IC pins!
MTH TIU Signal Protection Gerber.zip
It seems like everybody is overlooking a key benefit of John's solution.
"Schottky diodes couple the voltage limiting to the pins while limiting the stray capacitance from the TVS."
I'm going to have to build some more of these and get a bit more testing on them.
Oman posted:It seems like everybody is overlooking a key benefit of John's solution.
"Schottky diodes couple the voltage limiting to the pins while limiting the stray capacitance from the TVS."
Certainly a good point. The argument being that any capacitance is bad for signal quality and the lifespan of the ACT244 itself right? More capacitance equals more work for the driver to do right?
Ryaninspiron posted:
The TIUs are powered by smaller power bricks and those get switched on via a master DCS power relay. Then the main and branch lines each have seperate switches and relays that connect the TIU output channels to the tracks when the club has a DCS day. (The club has issues running DCS and Legacy at the same time so only one system is ever on at once now. That should be a thread of it's own.)
I also do not know the exact reason yet. But I imagine one benefit is that they are subject to derail spikes a third of the time. These design choices were made long before I joined and I haven't asked why yet.
I have to ask, just how many TIU channels are connected?
2amp total/.2 amp draw per channel = 10 TIU channels?
I'd guess 75' per channel max for 3 rail so maybe = 750' total track?
I am amazed at the posted differences with signal problems. I run an outside layout of G scale around the yard. I have maybe a couple hundred feet of track on each TIU. I only use the 2 fixed channels right now as it's powered by DC. I added a second TIU when I added another mainline.
I also run 2 rail O scale inside and have a good length of track per TIU channel.
I know some clubs have a ton of track. I wonder if signal loss is more a function of power dips? If the layout is stationary I would believe optimum wiring practices would be easier to achieve here.
gunrunnerjohn posted:I'm going to have to build some more of these and get a bit more testing on them.
I would like to know if yours can be build on a perfboard using through hole components. pretty much in the interest of time. We all know exactly which TVS to get for the though hole version "SA5.0CA", but which throughhole schottky diode would you say is the equivelant? maybe the SB180? https://www.digikey.com/produc...0/SB180CT-ND/3907950
Or is the SB180 disqualified because it has a 110pF capacitance as opposed to the SMD one you selected having 15pF of capacitance. also about a hundred times more reverse leak current at 500uA@80v vs the SMD one having 5uA @80v.
Engineer-Joe posted:Ryaninspiron posted:
The TIUs are powered by smaller power bricks and those get switched on via a master DCS power relay. Then the main and branch lines each have seperate switches and relays that connect the TIU output channels to the tracks when the club has a DCS day. (The club has issues running DCS and Legacy at the same time so only one system is ever on at once now. That should be a thread of it's own.)
I also do not know the exact reason yet. But I imagine one benefit is that they are subject to derail spikes a third of the time. These design choices were made long before I joined and I haven't asked why yet.
I have to ask, just how many TIU channels are connected?
2amp total/.2 amp draw per channel = 10 TIU channels?
I'd guess 75' per channel max for 3 rail so maybe = 750' total track?
I am amazed at the posted differences with signal problems. I run an outside layout of G scale around the yard. I have maybe a couple hundred feet of track on each TIU. I only use the 2 fixed channels right now as it's powered by DC. I added a second TIU when I added another mainline.
I also run 2 rail O scale inside and have a good length of track per TIU channel.
I know some clubs have a ton of track. I wonder if signal loss is more a function of power dips? If the layout is stationary I would believe optimum wiring practices would be easier to achieve here.
these readings were only taken with a the meter on the Z4000 so far (so not quite a precision current meter) and the setup (based on my observations so far) is 4 TIUs with 8 channels each for the 2 mainlines (they take up the bulk of space in the 4000sqFt warehouse space) The 5th TIU handles the two branch lines with 2 channels each.
I personally did not know about that 75ft limit per channel you just mentioned. but you are close on the estimated track length. But I also was not in the club during the construction of the current layout.
Interestingly the branch lines like the subway that runs the perimeter of the layout has the worst issues. In testing on Monday you could hit a horn and it would either not work or get stuck on.
To select a thru-hole component, I'd have to do a parametric search, same as you're doing. I don't think the 110pf is major factor, however the less the better. The reverse leakage also shouldn't be a show stopper. My goal was to minimize the impact of having the protection board connected to the chip, so those were all considerations.
Ryaninspiron posted:I personally did not know about that 75ft limit per channel you just mentioned. but you are close on the estimated track length. But I also was not in the club during the construction of the current layout.
Interestingly the branch lines like the subway that runs the perimeter of the layout has the worst issues. In testing on Monday you could hit a horn and it would either not work or get stuck on.
just a guess and not a rule on the 75' !!
G scale can be higher! ( I think I have maybe 150' per channel outside?)
I originally had my whole 2 rail layout on a single TIU channel. It was more than 150' of track. I had my G scale on the same TIU and eventually separated them. So it was suggested why have spare TIU channels not in use? The layout was getting on the fringe of lower signal for any reasons it seemed. So I split it up over more TIU channels. 75' just happened to be the split. I kind of made that my regular here after that.
O scale 3 rail seems to be worse than mine from other posts here. I did set up a 3 rail loop all around my basement. It's on an older TIU with bulbs just on the outputs. The track is MTH Scaletrax with soldered joints. I have around 75' per channel and good signal.
If I plug in my TMCC base and connected to any of my layouts, I have issues there.
That subway line has lit cars that may need chokes? or what else is different? type of track? wiring? length? TMCC equipment?
It would be nice if I lived close to a club like this where I could experiment. I did do enough on my own and was happy to dial it in. I come here to try and share what I've learned when I can.
Ryaninspiron posted:Oman posted:It seems like everybody is overlooking a key benefit of John's solution.
"Schottky diodes couple the voltage limiting to the pins while limiting the stray capacitance from the TVS."
Certainly a good point. The argument being that any capacitance is bad for signal quality and the lifespan of the ACT244 itself right? More capacitance equals more work for the driver to do right?
Watching for a response to “More capacitance = more work for driver?”, wondering if block length along with wire size (large for reduced voltage drop) contribute to channel failures in TIUs, as well as signal reduction (have heard this long ago) as capacitance damps the signal. It would diminish peak signal voltage, as well as blur the edges of square signal waves, right?
Does this lead to logic for breaking long mainlines not just into blocks less than 75-110 ft long (range of what I’ve read) with branched leads from a single TIU output, but also into separate TIU channels, for reasons of capacitance per channel reduction?? For signal improvement, and does that also lead to reduced risk of TIU output channel failures?
cnwdon posted:Ryaninspiron posted:Oman posted:It seems like everybody is overlooking a key benefit of John's solution.
"Schottky diodes couple the voltage limiting to the pins while limiting the stray capacitance from the TVS."
Certainly a good point. The argument being that any capacitance is bad for signal quality and the lifespan of the ACT244 itself right? More capacitance equals more work for the driver to do right?
Watching for a response to “More capacitance = more work for driver?”, wondering if block length along with wire size (large for reduced voltage drop) contribute to channel failures in TIUs, as well as signal reduction (have heard this long ago) as capacitance damps the signal. It would diminish peak signal voltage, as well as blur the edges of square signal waves, right?
Does this lead to logic for breaking long mainlines not just into blocks less than 75-110 ft long (range of what I’ve read) with branched leads from a single TIU output, but also into separate TIU channels, for reasons of capacitance per channel reduction?? For signal improvement, and does that also lead to reduced risk of TIU output channel failures?
Do you want the simple answer or the real one?
Real is always better, thanks, Adrian. I have to consider things like 10 gauge leads to the track buss for a few blocks to serve some important old trains, so capacitance appears to be my middle name. If I need to invest in more TIU channels it would be a small cost in the big picture. Or, what else do I not understand yet? Would appreciate the “real” answer!
Don
My experience is that larger wire gauge enhances signal strength on the track. When I built my layout in the early 90s, before DCS, I used 14-gauge throughout, except I had some 16 toggled blocks at the far end of the layout under the main layout, for layovers. To save effort, I used 18-gauge multi-conductor cable--what the heck, I'll have ny jand on the throttle anyway as trains go into those areas, so I can apply a little more power to overconme voltage drop.
When DCS came along, and I inserted the TIUs between transformer and control panel. The 14-gauge blocks had great signal strength; the 18-gauge blocks were pitiful. Eventually I rewired all 16 18-gauge blocks with 14 gauge wire, and signal strength was no longer a problem.
Incidentally, I do not use twisted pairs, on the theory that wires that close together have increased capacitance. All ground is via a 12-gauge buss loop that goes around the layout.
cnwdon posted:Real is always better, thanks, Adrian. I have to consider things like 10 gauge leads to the track buss for a few blocks to serve some important old trains, so capacitance appears to be my middle name. If I need to invest in more TIU channels it would be a small cost in the big picture. Or, what else do I not understand yet?
Don
Right so how it "really" works is DCS communicates with CDMA codes. This is a spread code of lots of 0s and 1s that are compared to represent a real "0" or "1"
0: 1001001010111100...
1: 011011010100011...
The real codes are like 32 bit long and transmitted at a rate of 3.25MHz. In order for DCS to have successful detection 29/32 of the bits need to be right so you can tolorate some error, but not all errors. They are square waves not sine waves so when you look at their fourier representation they have harmonic content at 9.75 MHz, and 16.25 MHz (square waves only have odd harmonics!). The ACT244 driver that drives your track with this waveform is a CMOS gate inside so it basically looks like a current source (when it's pulling up or down, but not statically sitting at value). Capacitors are I=CdV/dt. I is fixed by the ACT244 so dt (the time of the slope from 0 to 1 and 1 to 0 is merely dt=Cdv/I. The bit time of a 3.25MHz square wave is 307ns and once the slope starts getting close to 20% of this time (like 30 ns) your square starts to look more like atriangle. So in the engine decoder it compares a perfect code to a code with slopes caused by the added cap combined with limited drive current. It then all comes down to the deocder's correlation result.
The two codes are compared by multiplying and integrating them timewise... so if we look at an oversampled representation of one bit of the code
000011110000 compared with 000011110000
First multiply each bit and add up the result 0(0) + 0(0) + 0(0) +0(0) +1(1) +1(1) +1(1) +1(1) +0(0) +0(0) +0(0) +0(0) = 4 so the detection level is "4" in this example
Now consider if you have slow ramping from the drive/cap pair:
000011110000 compared with 0000(0.25)(1)(1)(0.25)0000
0(0) + 0(0) + 0(0) +0(0) +1(0.25) +1(1) +1(1) +1(0.25) +0(0) +0(0) +0(0) +0(0) = 2.5 so the detection level is "2.5" in this example
At some point when the wave gets too sloppy the detection level gets too low and the decoder becomes unsure if the code it's seeing is the 1 or 0 code, and then you loose reliable communication. A similar thing happens if the amplitude of the signal is lower than it's supposed to be. Lets say the waveform has 30% of the designed 10V amplitude.
000011110000 compared with 0000(0.3)(0.3)(0.3)(0.3)0000
0(0) + 0(0) + 0(0) +0(0) +1(0.3) +1(0.3) +1(0.3) +1(0.3) +0(0) +0(0) +0(0) +0(0) = 4 so the detection level is "1.2" in this example
So in practice it's important to both make sure that the amplitude of the DCS waveform is good and that the edges are sharp and do not eat more than 0-5% into each bit time. In our club we look at the waveforms in each section to make sure they are crisp and set the lengths from that. Remember that cap per length of track is *ALOT lower* than cap per length of wire. Loosly capacitance is C=eoer A/D where A is the "effective area" of the two signal conductors and D is the distance between them. A twisted pair of wire is a lot closer together than the center rail and running rails of an o-gauge track so generally the pF per foot is of wire is 10-50X higher than the track itself. Setting the length requires balancing these two contributions. Another way we looked at is stacking ACT244s in parallel to make Idrive bigger. This also works but the actual assembly is a headache.
This is only the half story though. If you pull your twisted pair apart you create inductance (which is proportional to the area between + and -). Too much inductance causes ringing which also has a similar effect on the decoders.
Thanks. I have 12 gauge busses for track power around 130 foot mains divided in 2 blocks each, and feeds to the busses from TIU are 12 or 14 gauge. I do have twisted pairs. I’m about to add 10 gauge feeds for at least one block, possibly two, because voltage drop is just an issue even when I think I had it covered, even with trains only drawing 5 amps or so max. It doesn’t take much V drop to slow a train with Pullmor or older-style motors, learned that years ago. Not all have “cruise control” to make it easy, esp the older style locos. TMCC works great and doesn’t care about capacitance.
I’m interpreting what has been posted on the forum to suggest that dividing the layout with more TIU channels might be a way to deal with the compromise, capacitance from necessary wire size and need for less of it to get good signal. Wrong track? Experiments with an oscilloscope for measurement of results (discussed elsewhere) to assess such questions are something I’d like to do.
