Skip to main content

Adrian! posted:
gunrunnerjohn posted:

Poorly.  The signal strength in the locomotives simply measures how many good packets vs bad packets it gets for a test stream.  Depending on the circumstances, it could be getting all good with low voltage from the TIU.  Also, as we all know, a good TIU with bad wiring/track layout will yield bad results.

This is a tool to test the TIU independently.

Actually, since we’ve been talking about lengths.... the MTH signal test also uses a very short test packet like the length of a speed command (about 500us), so it’s skewed to be very optimistic and doesn’t inform the situation for longer commands like adding engines or app reads

https://ogrforum.com/...-signal-stregth-test

I think I knew this before I asked?

Again, thank you for the time to explain things.

Last edited by Engineer-Joe

Well Adrian, as if you are bored with nothing better to do, could you look at the signal packet for a large consist of engines and determine just how much longer the packets are for them vs. a single engine?

 There were times I could get great response from running a single engine around the track. Then it would seem like "all hell broke loose" when I built a big train. MTH posted that running a big consist took more info to communicate with the consist. I believe they said each engine has to respond back correctly to satisfy the command?

Don't quote me exactly on this stuff as my memory is not working fully right now...

Maybe using smaller LEDs would help, possibly 3mm? 

Gee, you already have it redesigned, drawn up and ready to go with PCB designed. While I have only loaded Diptrace and found the parts section...and that was using Adrian's schematic (no redesign) and it's not yet entered into Diptrace. But, at least it's a start! 

Yogi Berra out driving with a friend with a specific destination in mind... The friend said I think we are lost, Yogi replied, yea, but we're making good time! (That would be me and Diptrace)

Last edited by rtr12

The board is already small at 1.1 x 2, and I happened to have a 5mm LED footprint handy...

Diptrace is great, but as for any such tool, there is a learning curve.  You're going to find out the hardest part is finding parts you don't have in the library and creating new ones.   I've used it for perhaps a couple dozen boards so far, so I've built a library of many of my commonly used parts.  To get the 3D views, you also have to hack the 3D representations of parts, sometimes I grab a totally different part 3D view to "build" the part I need.

Last edited by gunrunnerjohn

Thanks for the encouragement! I hit a snag already, with the CD74HC123. They didn't have one. I did find the part search though. Next step is, as you say, trying to figure out the library stuff to modify the existing generic 16 dip I found to match what I want. Might as well start learning now I guess. I can see this will be taking me a while. Parts in Diptrace are harder to find than searching through Digikey! 

I was kidding about the LEDs, but I will keep the size you had in mind. One goof I did already was to order a 9v battery holder...sounded good at the time of order. maybe now not so much. Diptrace doesn't seem to have those either... 

If you want to whip one out for everyone else, please feel free. I will probably be testing everyone's patience while learning to create mine. I am only doing the simple one that Adrian first posted. I'm still way too green for any feature creep! I do have what I think is a valid parts list from Digikey which anyone that wants it is welcome to. I can post if anyone is interested.

Last edited by rtr12

You didn't look hard enough in the DIPTRACE libraries.  If you search for 74123, you'll find the part.  The SN74123N is the one to pick.  The HC parts are a just high speed CMOS, but they have the same pinouts as the TTL parts.

Part of the magic of parts is knowing that the different technologies are all the same pinouts as a rule.  When you're looking for something like the 74XXnnn part, for instance, search on the 74nnn part instead and you're more likely to find it.  Sometimes, especially for surface mount, the proper footprint isn't available.  Then you can simply find a part with the proper footprint and graft the two together to create the necessary part.  I don't usually have to create a footprint, but for certain parts I have had to do that.  Once you do a few, it gets pretty easy.

Thanks, that is a very good tip on the 74xxx parts. I knew there were different ones, but didn't know about forgetting the letter designations and just using the numbers. Wasn't sure they were all the same either? That really helps a lot! All I found was a bare DIP 16 socket, didn't have any of the designations on it like Adrian's drawing.

I watched the Diptrace tutorials last night, but I was tired and. Probably need a mulligan on the tutorials. They went through things pretty fast too.  

I am off to hunt the SN74123N! 

gunrunnerjohn posted:

Seems it was a bit more parts than I imagined to do the multiple level indicator.   I did put them all on one side, the board is 1.1" x 2".  Maybe this was too much "feature creep".

Hi John: What software did you use to draw the electrical schematic and what software did you use for the 3-D Drawing?

Thanks: Gary

rtr12 posted:

Thanks, that is a very good tip on the 74xxx parts. I knew there were different ones, but didn't know about forgetting the letter designations and just using the numbers. Wasn't sure they were all the same either? That really helps a lot! All I found was a bare DIP 16 socket, didn't have any of the designations on it like Adrian's drawing.

I watched the Diptrace tutorials last night, but I was tired and. Probably need a mulligan on the tutorials. They went through things pretty fast too.  

I am off to hunt the SN74123N! 

Another clue is sometimes it helps to leave off everything but a few numbers from the part, that may find parts you didn't find other ways.  Sometimes it also returns tons of other stuff, that's part of the fun.   Try 123N for instance, it returns twelve 123 single shot parts of various logic families.

Last edited by gunrunnerjohn

OK, I take pity on you guys, here's the complete PCB design with the Gerber files for upload to your favorite PCB house.  You can drop this ZIP file on OSH Park and they'll build you three of these blank PCB's for around $11 shipped.  Oh, the 2K pot is the Bourns  3306W-1-202, available on Digikey.

TIU Signal Tester v1.0 Gerber.zip

 OOPS!  The CTS pot wasn't available, so I had to re-spin with one that was.

TIU Signal Tester v1.0 3D ViewTIU Signal Tester v1.0 Schematic

Attachments

Last edited by gunrunnerjohn

That's even better than part search tips! Thank You GRJ!!!  (Mark and I must be living good clean lives! )

I actually got a schematic and got the PCB 'sort of' laid out. It still needs work, but I am close I think? Your layout (and schematic) is a bit different than mine, yours is a lot smaller and probably much better laid out. Can I load the Gerbers into Diptrace to see what you did? Maybe I can learn a few things from that as well.

The Pot is different as well, but I can get those in a couple of days so not a problem. I never thought to use the type of pot you used, which looks like a pretty good space saver. Years of experience there I guess!

I am still going to keep fiddling with what I have and may even try ordering from OSHPark as well, just to see how it all works. Maybe I can improve somewhat for next time?

Now to learn how to use OSH Park... Thanks again GRJ!

And Mark, I agree with not doing anything after supper, TOB, Tired Old Brain!  

GRJ, My results for the 74123 IC were also quite different than yours, did you modify one or use something else? When I searched for the SN74123N from the above post I got a symbol that was in 3 different pieces. I think the 123 search did have some single item symbols, but I was thinking some of the pin designations didn't match Adrian's schematic? Again, more experience on your end I'm guessing? Just curious on this one?

Lots more to learn about all this, just when I thought some of it was starting to make sense too... 

First, I attached the TIU Signal Tester Diptrace Files.zip, so you can load these into DipTrace and see exactly what I have, and even modify it if you like.

You searched for SN74HC123N, there isn't that specific part.  However, if you leave off the "SN", you do get a couple hits.

In the case I stated, I searched for 123N, that yields 12 parts.  Note the SN prefix was only on the xxLS123N parts.

As far as the pot, I had a larger one of the same style in there, but when I looked for it, it wasn't stocked anywhere.  Oh, all the resistors are 1/4W, and the caps are all 50V or better.

Adrian didn't have pin numbers on his drawing, but I tumbled to the fact that he had the chip laid out like it physically appears, so pin 1 is the left top, then you count down and around and up the other side.  Given that "clue", I just used the hopefully correct pin numbers to match his design.

There was no specific plan to make this as small as possible, I just arranged the components in a square in a somewhat logical pattern based on the interconnections and then drew the board outline around them.  I didn't see any reason to squeeze it any further, if I really wanted to, it could probably be half an inch shorter.

As far as "customization", I do have my own set of libraries that I add parts to.  I try not to modify the stock DipTrace libraries so that upgrades are painless.  If I hack their libraries, I'd have to go back and do it again for updates.  However, it's easy to create user libraries.

Using OSH Park is dead simple!  You just create an account, and when you login, you're presented with a "start" page.  Drag-n-drop the Gerber ZIP file onto the page and follow your nose. 

For quantities of boards, you can also consider SEEED Studio, you can get ten boards for around the price of three on OSH Park.  SEEED does take longer, I'm still waiting on my order of PCB's for the TMCC Buffer surface mount version, ordered on June 23rd.  However, three of those from OSH Park would have been $43, and they were only $12.69 from SEEED Studio, so I went for the cheap option, I've already invested hundreds into the TMCC Buffer project with no returns yet, I can wait a couple extra weeks.

Attachments

Last edited by gunrunnerjohn

Thanks again GRJ! I really do appreciate all your help! I am starting to get the hang of things somewhat and the DipTrace files will really help here I think. If I live long enough I may even be able to make my own PCB!

I did do the 123N search and looked through all the possibilities there, but I guess I just missed a couple or got confused? Also, the labels didn't seem to match Adrain's labels? I guess that's where your experience comes in and I get mixed up. Probably shouldn't have been looking for an exact match on that part either. 

The PCB layout was really interesting how DipTrace laid it all out and then added all the wiring/traces. It's amazing how they program all this stuff to do that! I then fiddled with re-arranging a few things, but that still needs work on my part as well. I think one could make a PCB from the automated one created, but I wanted to try and optimize it like you do yours here on the forum. Fun to fiddle with that part! I haven't yet made it to the 3D part, but I bet that is pretty cool too.

I was looking at making my own library to make a chip 'exactly' like Adrian's, but I have not got that part mastered just yet. More study and trial and error needed here too! I think having your own library is something that would be a very good idea to have as you have done.

I did get the Gerbers posted and ordered at OshPark and that was pretty easy as you describe. The 'drag and drop' was really nice and it only took a few seconds for them to check the files and put them on order. They also take PayPal which I like and the total for the three PCBs was only $10.75 & free shipping, pretty reasonable I think. More amazing software stuff at work, they make it painless!

I would be happy to send payment for my TMCC buffer if it would help any. I know you must have quite a bit into that by now and that one is a lot of work too, I'm sure, and there are lots of them by now too. 

Anyway thanks again for all your help. I'll try not to be such a PITA as much as possible! 

I wouldn't bother to make a part to match his part, you'll have plenty of chances to practice making parts if you use DipTrace for anything else.

PCB schematic capture and layout has gotten a LOT easier over the years.  I used to do the PCB layout with tape, talk about a PITA, and changes were a nightmare!  Here's a page that describes the process, it was not pretty!  PCB Design, a Graphic Artist's Handicraft by Skilled Designers.  Checking the tape job was a very tedious task for large boards, and if you weren't REALLY careful with the taped mylars, you'd lose a trace or two.  You had to create all the top and bottom layers and have perfect registration between them.  Your best friend was the light table.

After you finally got the taped mylar design verified, you got films made.  You had a whole stack of films, one for each layer, both silkscreens, both solder masks (if different), all the copper layers (depending on the number of board layers, etc.  This whole bundle got shipped to the PCB house to actually create the boards.

...

Not to worry about the TMCC Buffer, as soon as I get the SMT boards in and test that layout, I'll be sending out the group email for everyone, then they'll have to commit with real cash to be in on the order.  This is a BTO process, and it'll be a one-time deal, I'm not planning on making the buffers a long term thing.  I'll be doing the board and assembly order from one of the Chinese houses I use, and I have to have a decent quantity to do a build at a reasonable price.  There are also several $10+ parts in the design that I'll have to source, as well as the plastic enclosure and power module.  I have everything figured out as far as production, and I've even built a simple assembly jig to drop the completed boards from China on to add the thru-hole parts with the proper spacing.

I suspect once the initial demand is satisfied, there'd be very few additional orders.  I never expected this to be a cash cow, I just made a commitment to Dale about doing the build.  Even though he's gone, I felt it was only right that his legacy be carried forward, he did the heavy lifting on the project.  Also, I'm sure that some people that get them will later decide they don't really need them, so there's bound to be a "used market" after a time.

That really looks like a terrible process to have to go through for a PCB! I would have never tried one doing it that way. The modern software and PCB makers have really streamlined the process from doing it the old-fashioned way. I briefly fiddled with DigiKey's KiCad but never got too far with it. I think DipTrace does seem to be easier than that one anyway.

I did a tiny bit of drafting in the late '80s (not electronic related) and it took a lot longer than today using a computer with good software. At the time the company had only one IBM PC/AT with Autocad so there were not enough to go around, It wasn't long until they got more PCs and everyone had their own. That was new and fun back then. I will say that I think the computer designing is much easier to make a mistake in, such as copy/paste and forgetting to change something that was a little different (I still do that!). I think in the manual drafting an error was much easier to spot and you had to draw each item so the copy/paste errors were non-existent. The PC was much faster though and also much easier to make revisions on. Time saved outweighed the occasional extra errors. We had nothing like Diptrace though, that is pretty impressive!

The Digikey first order of parts came today. I had absolutely no pots with the mount on the PCB so I ordered some of those and a few other items today. Should be here by Friday. Oshpark emailed that the PCBs will be on their July 20th PCB layouts (or something like that) and then go off to be made. Won't be too long now!

Ok on the TMCC buffer, I knew from your earlier posts that some of the parts were expensive (and so is moving, first-hand experience there a few years ago). That turned out to be a lot of work and I'm sure Dale would be pleased with it all coming together now and his work being carried on. 

I will probably be one of the people that doesn't actually need one, but thought I better get one while the getting was good. You can't have too many electronic gizmos! I do plan to expand my layout, that was the idea from the beginning, but things have not progressed much after I got the first part up and running. It's still in the plans, maybe I'll make it someday? 

To ALL:

If anyone is interested in ordering PCBs from Oshpark I can try to figure out how to share the project and post a link here for anyone that is interested. It was $10.75 for 3 boards with free shipping. For parts, I ordered extras for a discount price so I don't have an itemized cost breakdown per part to get a total cost per board. All parts I ordered are from DigiKey, they had everything.

Or you can wait and see how mine turn out and go from there. I will post my progress as I get the parts in and get things put together. Also, I think I got all the parts correct, but I usually goof something up somewhere so the material list may need a little adjustment?  If you want to wait I can make any corrections to that as well. It's GRJ's PCB design that I used, so I am sure that will be fine, probably no worries there.

Adrian! posted:

I just use Ansys SiWave and Altium for PCBs. Costy but good.

I'm sure that some of the professional PCB packages are great, but I don't have the budget for several thousand dollars for the schematic capture and PCB layout stuff.  The small boards I'm likely to do just don't justify that kind of expenditure.  For what I do, DipTrace works very well, and is considerably easier to use than Eagle, one of the other major players.

The last that I was exposed to printed circuit board design, it was very labor intensive, but I liked the old days of drawing things by hand.  I'm thinking of the late '70s to 1980.  Of course, It took time and each person's productivity was very low compared to today.  Since then, I have worked with already produced products.  Just as someone coming from a postwar train background and being exposed to command control trains, it seems like my background in PCB is from the stone age!  

Actually, I didn't like the "old days" when it comes to PCB design.   Doing all of that stuff by hand was very tedious, and I could visualize much better uses for my time.  Doing something simple like the little TIU Signal Tester by hand would take probably all day doing all the drafting by hand and then routing and taping the PCB design.  Next, you had to do the films, and finally get the package to the PCB house.  I was able to do that little task in about an hour, much better use of time.

I agree it's a much better use of time.  My preference is because I'm not the kind of engineer who can develop anything.  I think it is better that I have found a nitch as one who puts a circuit together using commercial building blocks, writhing work orders for the technicians, and documentation.  Those are all things most don't want to bother with.  Maybe I should have taken the position as a technical writer way back in '76.  It didn't seem interesting at the time, but finally that seems to be what I am good at.  

Jeepers Creepers, this is why Design Engineers need PRACTICAL Engineers to over see them  When I first brought this up, GRJ said it was too hard and too costly, so get an Oscilloscope.   NOW, Adrian designs what I asked for about $10, and all of a sudden GRJ is asking for upgrades for it!!!!

So let me add some practical advice.  Certainly a go no go is ok and I certainly would pay $40 for one that can easily test a TIU output.  Show a perfect signal around 12V with Green, Show and acceptable around 9V with Blue and a Red for no go.

If there is a crude adjustment POT that can be calibrated in rough voltages to get a green to light up at the set value 3, 6, 9, or 12 that would be great, and no other LED required.  Start at 12V, no go, set to 9 and retest, still no go, set to 6V and if it lights up, you have a rough estimate.  But frankly if less then about 9V from what I have been reading and you should replace the Transmit chip.  So a simple device and I do not see how Service techs would not pay $40 for the test equipment.  We far more for a PS-3 Test set.  BUT we need something that doesn't take up square yards on the work bench, so for me a O Scope is out.  This device is in.   Thank you for this effort.  ADRIAN if you build me one I will pay for it and then pass the charge on to MTH  G

George, that was my idea with the four LED's, but as pointed out, it's probably overkill.  However, you have a design that won't take up much room and will give you a quick indication if the TIU has a full strength output.

Truthfully, if I were going to spend more time on this, I'd probably try to eliminate the pot and have it trigger on something like 11 volts peak signals, any less and you'd get a no-go.

To anyone using GRJ's design as a starting point to learn the tool.  Here's a project.

Only 1/2 of the 123 chip is being used.  This means you can add a 2nd threshold and 2nd LED which might make the tool more useful.  So LED1 would blink when a full level packet (12V or whatever you guys come up with) is detected.  And LED2 would additionally blink if the packet is above, say, 50% of the full level.  The component cost would be pennies - LED plus a few resistors, caps.  Not as handy as having 4 thresholds but small steps.

With only one LED, if it doesn't blink it could alternatively mean that the battery is dead.    

Separately, if modifying the 1 LED design, I suggest terminating (e.g., tying to ground) the input pins on the unused half of the 123 chip.  It is not a good idea leave unused inputs "floating" (not connected to anything).

 

 

Stan, I was thinking that simply adding a resistor divider based on a 5V reference, you could have a second LED with no adjustment pot.  As long as you assume the +5V is within around 5%, it would probably be accurate enough, and it sure makes calibration a moot point.  Of course, adding another 123 would give you the four LED's I started with.  With a PCB, it would be pretty easy to add the extra components, no hand wiring involved.

I think I see feature creep rearing it's head!

hc high low input thresholds

So here we have what would happen in a typical engineering company back in the day.  An engineering design review subjects the schematic to the peanut gallery (other engineers) who weigh in and must "sign off" before PCB layout.

The problem with using the 123 is the thresholds are not calibrated.  This being in contrast to the comparator design showed earlier.  In principle the 123 chip is specified to trigger anywhere between 1.35V and 3.15V … a fairly wide range.  Within one chip, the two halves of the 123 will trigger at the same levels just from how the device is manufactured.  But the thresholds can vary between two chips.  This makes the 4 level detector (using two 123 chips) problematic and would probably not pass a design-review in a corporate environment.  But we're talking hobby DIY so I'll grin and bear it.  

Attachments

Images (1)
  • hc high low input thresholds
stan2004 posted:

hc high low input thresholds

So here we have what would happen in a typical engineering company back in the day.  An engineering design review subjects the schematic to the peanut gallery (other engineers) who weigh in and must "sign off" before PCB layout.

The problem with using the 123 is the thresholds are not calibrated.  This being in contrast to the comparator design showed earlier.  In principle the 123 chip is specified to trigger anywhere between 1.35V and 3.15V … a fairly wide range.  Within one chip, the two halves of the 123 will trigger at the same levels just from how the device is manufactured.  But the thresholds can vary between two chips.  This makes the 4 level detector (using two 123 chips) problematic and would probably not pass a design-review in a corporate environment.  But we're talking hobby DIY so I'll grin and bear it.  

Exactly why I went with a trimpot instead of a fixed R-div

More good ideas! I was hoping Stan would be along with some suggestions.  

I will try doing this, but I do think 'feature creep mode' has been activated. Think I'll wait until this percolates a bit more? 

My draftsman left a resistor off of my schematic...just wait 'til I find him! Good thing I used GRJ's to order the first round of PCBs. 

(Looks like it has been 'creeping' while I was writing!)

Last edited by rtr12

Add Reply

Post
The DCS Forum is sponsored by
×
×
×
×
Link copied to your clipboard.
×
×