I did testing on a few club TIU's for signal strength. The interesting thing seems to be, only the Rev. L units have the bad 244's, even though the earlier versions also use those chips. The obvious question is, what is different about the Rev. L design that causes those chips to fail when in an identical environment, older versions seem to survive just fine!
gunrunnerjohn posted:I did testing on a few club TIU's for signal strength. The interesting thing seems to be, only the Rev. L units have the bad 244's, even though the earlier versions also use those chips. The obvious question is, what is different about the Rev. L design that causes those chips to fail when in an identical environment, older versions seem to survive just fine!
Interesting....
You should do a chirp-up test at the act244 outputs of the rev L and earlier ones to see if the impedance they see is the same or not. I don’t have anything except rev Ls so I can’t do this.
Basically take a sig gen and inject a tone of 1V through a 100 ohm or so series resistor at the act 244 output. You can change the resistor if there’s not enough swing to measure. Then sweep the tone from 1K to like 10 MHz with like 20steps and note the amplitude at each with a scope at the act244 output node This lets you make a bode plot without a vna.
then repeat for both and compare...should be interesting!
Since it's the fails-to-a short voltage clamp that apparently damages the ACT244, isn't the question what changed in the clamping method from earlier TIU revisions? The sweep test might reveal why the magic-light-bulb and/or engineered-filter is no longer needed in later TIU revisions.
GRJ, I just replaced one a few days ago on a REV. I3 on channel FIXED 1.
Interesting, I tested three older TIU's, all had no issues with signal. I've also tested three Rev. L units, and two of the three required at least two ACT244 replacements.
GRJ, maybe the quality or the manufacturer of the 244's has changed. Look at what happened a few years ago with the WinCap capacitors. Why is there such a price difference with the ones listed on DigiKey? Is it their age or quality of the materials used to manufacture them?
Attachments
Images (1)
Hard to say why the prices differ, the ON Semiconductor is the old Motorola plant, but I had no complaints with the quality of Motorola parts in the past.
GRJ, Dave, Adrian, Barry, Stan and anyone else I may have missed.....
I want to mention 2 things here concerning this whole topic.
First.....I really don't know what we DCS operators would do without you guys if we had to solely rely on MTH to identify and fix this problem. I think we would all be in a real lurch (that's not to say that MTH wouldn't do their best to help. I just think they would be overwhelmed). You guys and your generosity of time and talent are overwhelming. A big Thank You to you all.
Second.....I was wondering what the status was of the final fix...or if that's still in flux due to symptoms still being discovered.
And no.....I haven't done my offline testing of my Rev L TIU issues yet. I'll try to get to that today....
Junior posted:GRJ, Dave, Adrian, Barry, Stan and anyone else I may have missed.....
I want to mention 2 things here concerning this whole topic.
First.....I really don't know what we DCS operators would do without you guys if we had to solely rely on MTH to identify and fix this problem. I think we would all be in a real lurch (that's not to say that MTH wouldn't do their best to help. I just think they would be overwhelmed). You guys and your generosity of time and talent are overwhelming. A big Thank You to you all.
Second.....I was wondering what the status was of the final fix...or if that's still in flux due to symptoms still being discovered.
And no.....I haven't done my offline testing of my Rev L TIU issues yet. I'll try to get to that today....
Final fix status:
1. I can offer you something that's never broken ever so far, but is hard to assemble and install. 4 TVS diodes and 4 daughter cards. The top of this thread.
2. MTH can offer you something that they install that only fails very rarely under heavy use (4 TVS diodes). Their service bulletin.
3. GRJ might be able to offer you something that doesn't break ever and is easier to install than the original mod I did, but still requires some soldering to put in.
Adrian, but in fairness, isn't it also true that all 3 fixes have not yet been tested in the crucible of everyday operation over an extended period? I do await your analysis of why the chip fails in Rev L but not the earlier versions.
RJR posted:Adrian, but in fairness, isn't it also true that all 3 fixes have not yet been tested in the crucible of everyday operation over an extended period? I do await your analysis of why the chip fails in Rev L but not the earlier versions.
I only know about the tests at the top of the post (1n4148 boards and TVS on the back). Five of those have been in our club getting hammered on every Saturday since Jan 2018 without any problems. I bring the scope and measure each channel and keep a weekly DCS voltage log (there's a sample in the original post at the top). We saw one channel die over that time, but that was because the TVS fell off the board (bad soldering) and left the channel exposed.
I don't have any earlier versions of TIUs so I need someone else who does to do impedance measurements before I can investigate the differences.
Junior, that is precisely why I recommend against interspersing TIU channels/power districts. I have 76 blocks fed by 6 TIU channels, set up by "loops" or areas, so there is a minimum for points where these short circuits are possible. The blocks in an area served by each channel are contiguous.
With all the power you have, interspersion does nothing beneficial.
As you note correctly, this os off-topic and should be another thread.
Adrian, I didn't realize you've been working the issue this long. Also hard to believe no one in your club has any of the old TIUs.
I'll have to equip a club TIU with my fix, but it'll be a long time in testing as we're a modular club, and the most use the TIU gets is when we do a show.
Analyzing trains crossing blocks really depends on if both blocks come from different channels of the same TIU or the different blocks coming from two different TIUs. You can probably treat it with superposition:
Power:
As long as the 18 VAC is in phase the power source won’t notice much. There will be some small current between power sources because the voltage is mismatched and there’s a potential there but probably nothing serious.
Transients:
in terms of transient events the train already generates KV scale transient events just running down the track as stan2004 has pointedly out. I’d suspect the break in the track between sections may increase the frequency of transients during the train crossing sections but probably not the magnitude since the make-and-break behavior doesn’t care about the physical geometry much.
signal: in the TIU there’s a mux selecting one channel at a time so you’ll never be driving the track simultaneously with two signals even if you connected the outputs together. They are always offset in time. If you have two TIUs the the situation gets more interesting since they can have drive conflicts. However when there’s a short they already deliver into zero ohms on the far side of the output transformer and nothing seems to break so similarly double driving the track/train doesn’t seem to be a problem from a damage perspective. Also the dcs duty cycle is low so statistics are on your side here.
In the worst case it seems the dcs performance may struggle as the train crosses blocks if there’s frequent drive conflicts, but i don’t see an obvious reason for components to be stressed beyond what they already experience within each block
RJR posted:Junior, that is precisely why I recommend against interspersing TIU channels/power districts. I have 76 blocks fed by 6 TIU channels, set up by "loops" or areas, so there is a minimum for points where these short circuits are possible. The blocks in an area served by each channel are contiguous.
With all the power you have, interspersion does nothing beneficial.
As you note correctly, this os off-topic and should be another thread.
Hi RJR....
Done....moved to a separate Post.
Thanks for keeping me honest 🙂
RJR posted:Adrian, I didn't realize you've been working the issue this long. Also hard to believe no one in your club has any of the old TIUs.
The count of home layouts is very very low here
When I return home next week, I will upgrade 2 TIU's to rev "M" and place on the layout. We run daily so the next month should give us good data, if the NJ Hi-Railers test a few, we should have sufficient data to ensure this fix is long term.
As for act244's, I have 200 plus of the ON chips as well as about 100 of the original act244's, I plan on evenly using them in replacing burned out act244's, and will know within 6 months if quality is an issue or not.
Don't forget us, 6 hours or more a day when you include cleaning and special events, 6 days a week.
right now our TIU signal has never been better.
bigdodgetrain posted:Don't forget us, 6 hours or more a day when you include cleaning and special events, 6 days a week.
right now our TIU signal has never been better.
I want to come down and measure sometime in July to make sure its exactly the same. Is the layout doing everything you want it to right now?
Adrian! posted:bigdodgetrain posted:Don't forget us, 6 hours or more a day when you include cleaning and special events, 6 days a week.
right now our TIU signal has never been better.
I want to come down and measure sometime in July to make sure its exactly the same. Is the layout doing everything you want it to right now?
yes as far as I know
we are recommending not to unplug the legacy and see what happens.
so far when we load MTH with Legacy plugged in we get the MTH loaded within 30 second.
we will be in contact with you soon.
Adrian, were the hell have you been all of AGHR life, wow you go man! You have made me and others very happy.
One of the founders of AGHR,
Former COO, president and Self appointed dictator for life.
John Pignatelli
The boss wifes, obedient indentured servant. 
I believe a lot of chips changed with the Rev L that were designed to improve signal strength in support of PS-3. Plus the USB port. MTH has acknowledged that once the Rev L hit the street, there were less failures of the receive chip and more on the Transmit chip 244. Hence the TVS mod.
I wish John had actually test that Rev L TIU operationally on a home layout before changing out the 244 chip. Comparing Adrian's Vpp chart on a home layout vice club layout would be interesting to know. It seems the Rev L even with a damaged 244 still performs satisfactorily on a home layout. But having analytical data here would be nice.
One of the issue with this TVS mod, as I just repaired a Variable channel fet on a mod Rev L tiu; is that you can't see if the TVS are shorted without removing them from the circuit. Not worth the labor unless there is a problem.
So a small test device that can measure output signal that is not an oscilloscope would be helpful. Or some loaded test that simulates a typical large home layout and see how the DCS test works on an engine. If a reduced output on the DCS signal still returns 10s on the layout of the home owner, does it matter what the Vpp is? G
George, FWIW, I was having issues at times adding engines on my simple carpet loops with the TIU that I illustrated in this thread. The one that I just use on the bench occasionally had one bad 244, looked to be one or two gates dead. The one still in the box was pristine, all channels tested full signal. That one was actually purchased used, but I haven't used it other than testing on the bench a few times here.
I had a chance yesterday to test the TIUs at NJ-HI railers and of the 8 TIU's tested, there were only 2 of the 32 channels that had weak DCS signal and they were both on the same TIU. For the time being I will apply the MTH TVS service bulletin and replace all 4 TVSs, and then check the 244 chip if the signal is still weak.
Digikey has several TI 244 chips (as shown above) with similar specs but the price range is between <$1.00 to almost $3. Any recommendations on which one of these chips are good for our application?
Once the GRJ's board is available I am sure I will have a TIU to test it on.
Bob D
rad400 posted:I had a chance yesterday to test the TIUs at NJ-HI railers and of the 8 TIU's tested, there were only 2 of the 32 channels that had weak DCS signal and they were both on the same TIU. For the time being I will apply the MTH TVS service bulletin and replace all 4 TVSs, and then check the 244 chip if the signal is still weak.
Digikey has several TI 244 chips (as shown above) with similar specs but the price range is between <$1.00 to almost $3. Any recommendations on which one of these chips are good for our application?
Once the GRJ's board is available I am sure I will have a TIU to test it on.
Bob D
Tests results below on a new TIU out of the box. All 4 channels were approximately the same results. The PP readings were ~15.4v and the W/O overshoot, the numbers were in the 8V range. Tests was done with no load and no TMCC signal.
Adrain - in one of your posts you indicated the max signal you can get W/O overshoot is in the 14V range. I was only able to get in the 8V range with a brand new TIU. Do I need to refine my scope skills or do I have a weak TIU to start with? Thanks again for all your help with the several threads you are currently posting on this topic of DCS signal issues!
Bob D
Bob D
Attachments
Images (1)
rad400 posted:rad400 posted:I had a chance yesterday to test the TIUs at NJ-HI railers and of the 8 TIU's tested, there were only 2 of the 32 channels that had weak DCS signal and they were both on the same TIU. For the time being I will apply the MTH TVS service bulletin and replace all 4 TVSs, and then check the 244 chip if the signal is still weak.
Digikey has several TI 244 chips (as shown above) with similar specs but the price range is between <$1.00 to almost $3. Any recommendations on which one of these chips are good for our application?
Once the GRJ's board is available I am sure I will have a TIU to test it on.
Bob D
Tests results below on a new TIU out of the box. All 4 channels were approximately the same results. The PP readings were ~15.4v and the W/O overshoot, the numbers were in the 8V range. Tests was done with no load and no TMCC signal.
Adrain - in one of your posts you indicated the max signal you can get W/O overshoot is in the 14V range. I was only able to get in the 8V range with a brand new TIU. Do I need to refine my scope skills or do I have a weak TIU to start with? Thanks again for all your help with the several threads you are currently posting on this topic of DCS signal issues!
Bob D
Bob D
Looks good measurement wise.
Adrian/Stan/GRJ: One of the posts above mentioned the possibility of a signal strength meter as opposed to a scope. Would such a device be feasible and, if so, is it the type of device that one of you could design and make available? I really would have no use for a scope other than my single Rev L, and I already have enough clutter around that I pity my Executor.
The problem with making a signal strength meter would be the limited market, no return on investment. Given you only have one TIU, find a friend with a 'scope. 
Your signals look like Mine Bob, I don't see a problem. I low pass filter might clean up the really high frequency transients, but they're pretty easy to filter out by eye.
gunrunnerjohn posted:Your signals look like Mine Bob, I don't see a problem. I low pass filter might clean up the really high frequency transients, but they're pretty easy to filter out by eye.
John
I was just seeing if I should be getting a value of 14v w/o overshoot, as was inferred in Adrain's post.
When I tested the 8 TIUs, my no load PP voltage ranged from 15 to 17.6 excluding any off the wall peaks that I saw in my earlier post. Typically the reading I had for W/O overshoot was in the was 7-8 v range. Reading Adrain's DCS Symptoms chart, a w/o overshoot voltage of 12+V is a good number to have. If my no load w/o overshoot numbers are only in the 7-8v range, how do I get 12V for a good operating engine? What am I missing??
I think you're calling some of the signal "overshoot". It's not exactly a clean looking signal in the best of times.
Bob, here's a typical signal on my 'scope. Note that the measurement is 19.00 VPP. If you look at the right, you'll see about an 11 volt negative going spike that is driving that measurement. The main body of the signals are around 13-14 VPP.
Here's one expanded, note the signal looks about like yours. I think you are at 2us/div and I'm at 1us/div, so mine is a bit expanded. This trigger didn't get the spike and the measurement was a more realistic 14.8PP
Attachments
John
I am with you. I was just going by Adrain's recommendation of measuring the signal correctly and to take the reading w/o overshoot as shown below. I all my current measurements I have disregard any erroneous spikes as you indicated on the right side of the picture above.
gunrunnerjohn posted:The problem with making a signal strength meter would be the limited market, no return on investment. Given you only have one TIU, find a friend with a 'scope.
Wouldn't every MTH Tech want one? I would. Plus home layouts that are large and have multiple TIUs. G
GGG posted:gunrunnerjohn posted:The problem with making a signal strength meter would be the limited market, no return on investment. Given you only have one TIU, find a friend with a 'scope.
Wouldn't every MTH Tech want one? I would. Plus home layouts that are large and have multiple TIUs. G
Making it is easy.... Making it substantially cheaper than a scope is not. I've come up with a few variants but they are all costy. Comparators (even fast ones like a TVL3500 aren't quick enough to give you a meaningful response on the DCS packet, and envelope detection sucks because the packet duty cycle is so low. The successful stuff I've worked out is all sampling the waveform with a fast 10MS/s ADC, but then you need a clock, and logic (i've been using an Xlinx ARTY FPGA)... so you're already at +$200 when you get it all nice on a PCB. I also did a USB oscilloscope and raspberry pi... still $200 ish. Still thinking about it though...
Adrian! posted:GGG posted:gunrunnerjohn posted:The problem with making a signal strength meter would be the limited market, no return on investment. Given you only have one TIU, find a friend with a 'scope.
Wouldn't every MTH Tech want one? I would. Plus home layouts that are large and have multiple TIUs. G
Making it is easy.... Making it substantially cheaper than a scope is not. I've come up with a few variants but they are all costy. Comparators (even fast ones like a TVL3500 aren't quick enough to give you a meaningful response on the DCS packet, and envelope detection sucks because the packet duty cycle is so low. The successful stuff I've worked out is all sampling the waveform with a fast 10MS/s ADC, but then you need a clock, and logic (i've been using an Xlinx ARTY FPGA)... so you're already at +$200 when you get it all nice on a PCB. I also did a USB oscilloscope and raspberry pi... still $200 ish. Still thinking about it though...
I think I'll just recommend the 'scope, we know that works. 
You can get a 'scope in that price range that will be more than sufficient to do this, not to mention it's worthwhile for a lot of other tasks once you have it.
FWIW, I equipped the first TIU with my protection circuit. Not everything was as easy as it looks. Specifically, the one chip next to the heatsinks makes it virtually impossible to get to the leads on the chip to solder with the protection module mounted. I had originally put the leads on the protection modules and then just tacked them to the chips on the board. I glue the modules on the chips first and let that dry, that keeps them from wandering around while you're trying to solder them.
That technique worked fine for three of the four modules, but the one didn't allow soldering. So, I pulled the module off the chip and first soldered the leads to the chip, and I'm currently waiting for the glue to dry. You can see the leads sticking up on the left of the board at U440, they'll be tacked to the protection board and the "fix" will be installed. Not sure exactly how to "test" this, other than to verify everything works and the signals aren't affected.
Attachments
$200 ish your cost is cheap. I’d throw $350 or so at one. That’d help pinpoint issues and save a ton of time. The ROI is the time. The only thing I never have enough of and can’t replace
I will say that a scope is much more versatile
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.
