I had to start the video twice, I thought I was hearing things! 
I'd probably be more curious if the new TVS was functional. 
If I took an old one out because I thought it was bad, it hits the can. I'll have to hook one to my HP supply and do a little test.
I'm suitably impressed. My 36V TVS limited right at 36 volts at around 50ma. I guess that one is a good one.
rtr12 posted:Thanks Stan! Earlier I tried a sample circuit from a book I have and got 52 volts from the Lionel 18 volt accessory transformer on the bench. It used 100uf caps and mine were only 50 volt so I quit after seeing the higher voltage to be safe. Must have been one of the 'variations' you mentioned above and more than a doubler.
I like your setup much better and will give it a try tomorrow. I think I have all the parts needed for it and I think the 0.1uf caps I have are like 1000 volt or something really high like that. I might even try a small PCB or something? Maybe like a battery tester or something where you just touch the TVS leads to some contacts on the PCB. Thanks again!
Not sure which "topology" you tried but all these doubler/multiplier circuits respond to the peak or maximum voltage. So for an 18V AC pure sine voltage from a train transformer, the peak voltage is square-root of 2 times the RMS voltage or 18 x 1.4 = 25V. So if it's a doubler, then you get 50V. Compared to a 0.1uF cap as I suggest, your 100uF cap "looks like" a resistor that is 1000 times smaller hence smaller voltage loss in the conversion. It gets arcane and boring really quickly!
I know you've been learning about PCB layout and this would/could be a small board that would be maybe $1 all-in (raw board plus components for a buck). I figure you could place some square pads/islands (no soldermask) on the board where you'd touch the 2 terminals of the TVS to the board. In all seriousness, it would be fascinating to hear from a cross-section of OGR readers as to the "status" of their TVS clamps.
gunrunnerjohn posted:Remember, a large 1500W TVS has significant capacitance, so if you use too many, it will start to affect DCS performance.
Interesting comment John! I remember about 2 or 3 years ago when TVS usage was generating a lot of usage comments here. One that was mentioned fairly often was this one: (loosely quoting) "I put one on every track connection (e.g. lockon)". So since I typically have 15 or more because of isolating tracks sections due to DCS performance recommendations, I bought 25 and put one on every lockon.
I also remember that DCS users were saying they even put one on each TIU output post and also for those that use star wiring, they put one on each terminal strip. So I do both of those too.
But your comment now says to me that I shouldn't be doing that. Am I interpretting correctly?
As always, thanks - walt
Well, if you experience any DCS signal issue, removing the extra TVS parts would be my first step to see if that helps. Each TIU channel has a TVS, so there's really no need for another one right there.
Stan, This is the one I tried first.
Yours works much better though, so I went with it. Worked great on the breadboard, just like in your video (and it also works just as well without the Oscar Meyer weiner tune). I like the idea of the pads too, which I just figured out how to do in Diptrace.
Although the Lionel Accessory transformer has a pretty substantial range of dial adjustment, it doesn't have much fine adjustment, it's like from 0 to 13-15 volts right away in the first 1/4 to 1/3 of the dial. You can do a bit better if you fiddle with it, but that's a pain too. Coarse adjustment is basically on or off.
I made a PCB for it also, it was $3.90 for 3 PCBs shipped at OSHPark, I ordered 3. It could probably be reduced in size a bit, maybe leave out the corner mounting holes? It started out at like $4.25 fr 3 and I shrunk it nearly an inch each which only got it down to $3.90. Not sure how much savings further shrinkage would produce? Of course, if GRJ got a hold of it he could probably get it down to $1 or so for 3. 
I haven't tested the PCBs, but will as soon as they arrive. I'll post the files for them if it all works. Or if anyone want them now I can do that too, but I would hate to give anyone something that didn't work properly.
Here's JPEGs of what I have so far:
I couldn't get a large pic of the PCB, so I enlarged it (schematic too, but it came out better). Reason for the poor quality. I'll have to do some more self education work on that part. Bigger drawings work better...
Ask and you shall receive. I couldn't hit the dollar figure, how about three boards for 65 cents? Total time to deliver, about 20 minutes. 
I figured some test leads for the test output would be useful instead of the test pads. After all, you need to connect the voltmeter and the TVS under test.
BTW, I don't see any traces on your board, are you sure you routed it before you shipped it off to be produced?
Hey, that's not fair, you used SMD parts! I am not done stocking up on the through hole parts yet and now I need SMT parts. I do have to agree that one is pretty small and for 65 cents you sure can't go wrong. I was pretty sure you could shrink things down a bit there.
Well, I do have that microscope thingie around here somewhere...and Stan was also suggesting a couple SMT chips on the CA-CC traffic signal PCBs which I still need to order and test out. Maybe it's time to give that microscopic stuff a try.
Hmmm...I thought I routed the traces, but I always usually miss or goof up something somewhere. I better go check while there's time to correct the order... Thanks!
How do you get the green 3D boards, mine always come out brown like the one above. I won't go into the missing patterns I have to battle every time I try that...
I probably would have trouble getting a thru-hole design down to less than $1 for three. OTOH, with an SMT design, I could have gone even smaller, just put a couple of the parts on the other side!
If the preview is what you sent, there's no traces. Hard to believe all the traces are on the bottom, but maybe they're the same color as the solder mask and/or board and they're invisible?
Truthfully, I don't know how you got the brown boards. OK, got it. In the 3D preview, look in the upper right hand corner, you pick the colors of everything there.
You saved the day, traces added and PCBs order has been re-done! Thanks again!
The good thing about ordering from OSHPark on weekends is the order, I think, waits until Monday for processing so you have some time to fix your errors.
Good thing, I thought it looked odd. Besides the 3D rendering, the big clue was this one with the ratlines still in place. That's not what I'd expect to see if the routing was done.
BTW, don't feel bad, I did that once as well. Only I didn't see it until later, and I realized looking at the order when I was wondering when it would ship then I noticed I couldn't see any traces! I went back to the Gerber files and could see I didn't route it. 
I just dropped those boards directly in the trash.
Now that you pointed it out, I see the colors. Guess I have seen that before, but it never occurred to me to see what that did.
Yes, all I saw was ratlines when I opened the file. Not sure how I managed that, but I almost always goof up something. Haven't had a 'perfect' one yet, but got pretty close.
OSHPark didn't give me any errors either? They have on previous PCBs, but I forget what the errors were? Guess they don't check for that? I'll have to be more careful with that from now on. I like doing that too, I like watching Diptrace route all the lines all by itself!
Anyway, thanks to a keen eye checking on things, all is well now!
They obviously don't check for that as I sent one down and they simply produced it with no traces. Never a whimper about missing copper, just give me your money and I'll send you useless boards.
In truth, since the verification checks pick that up, I got what I deserved, I normally check every time.
Weird they don't check for that one? Seems like they should. Guess they leave that one to the folks struggling to design the PCBs.
I need to start using the verification more, I don't always do that. Obviously I didn't do it this time!
I also have a few PCBs that were less than ideal when they arrived. I was able to use them, but some of the components did not fit well with the others, too close. Had to do some whittling on some of the terminal blocks too.
I had a number that had tight or "no way" fit early on, that's just getting used to dealing with the spacing. It gets more tricky when the boards get more dense and they don't have 3D models for you to visualize your design first. I had several throw-aways that parts simply didn't fit. I'd build one and sky-wire it to test function, and then correct the issue. It's also why I like to get only a sample of the boards first before I go to press with any number. I did one board order of 40 boards that had a mistake that I couldn't work around, OOPS!
One trick is to print the preview at 100% size and then lay components on it to see if they fit.
rtr12 posted:….
The good thing about ordering from OSHPark on weekends is the order, I think, waits until Monday for processing so you have some time to fix your errors.
Is it still the weekend? The 2 square pads are obviously for momentarily touching the TVS to the board. But the meter must also measure that voltage between the 2 pads. I had imagined using alligator clip cables cut in half for the input and output. The 4 flying-lead ends gets soldered to the board, 2 inputs, 2 outputs. The 2 inputs clip to the transformer. The 2 outputs clip to the meter (set to DC Voltage measurement).
And, if those mounting holes are large enough, you could run/loop the jumper cable wiring thru them for strain relief.
GRJ, that is a good trick about the print out, I'll have to remember that one for future use. So far I have only gotten small quantities either 3 from OSHPark or 5 from JLCPCB. Only thing I have ordered larger quantities of are the TIU Tester PCBs, which I have probably 18 or so left, with parts to go with them. Orders have dropped off lately.
Stan, I was also thinking about something to read the voltage output. By still the weekend, does that mean I need to change something else? I have some of the inexpensive DC digital meters I was using on a few cheap AC/DC converters for the output reading. I also have a few sacrificial alligator clips I got for trial on the TIU Testers, but never used. So I am all set to try all this out.
I have some cheap left over 2-hole screw terminals or the 2.54mm (Arduino type pins) for the inputs and big pads for the outputs. I think the pads are large enough to accomodate the alligator clip wires and then some, even the clips themselves, they are 2.5mm x 4mm. But, I am open to any and all further suggestions, or corrections if my thinking is way off here. I think I can still cancel and re-order if need be. I don't mind getting a second batch either if necessary. All part of the fun and learning.
And, courtesy of GRJ, we will now have electrical connections for all the components on the PCBs too!
The mounting holes are 3mm inside diameter, actually slightly over that, 3.125mm maybe. That was another mistake I made earlier, holes way too small for anything but the tiniest of screws. Of course if we use the GRJ downsized version the mounting holes go away, but he has the holes for soldering or installing pins or something of that nature. That 65 cents is tempting! That would be 18 PCBs for the price of 3 of mine, if I calculated that correctly.
I guess it depends on what you're trying to do. If this is a one-off for yourself, that's one thing. But speaking on behalf of a hypothetical OGR user, I'd think wires with 4 alligator clips (2 to 18VAC transformer, 2 to meter) would make it easier to use. Then you just press the TVS in question to the square pads on the board and read the meter to see if the TVS is still alive. So by it's still the weekend, this would mean adding a couple of pads on the output to solder the meter wires to.
I'm pretty sure that DCS-TIU port tester had a strain-relief hole for the wires to to clip to the TIU output. I don't know how big those holes were but the idea of course is to wrap a pair of wires thru a single hole.
I don't think you need mounting holes at all, but if you want them, then a #4 screw clearance hole which is about 1/8" should be fine. You'll have to look up your hole diameter choices minding the plated hole dimension is slightly smaller than the drilled hole dimension (before plating). Most PCB layout programs allow you to specify plated or non-plated holes but I'm not familiar with your specific program.
As stated earlier, I'd be fascinated to get some field data on whether or not actual 33V or 36V TVS components are failing "open". If not, then why bother replacing them! But I suspect most guys would want a completed tester with wires and alligator clips. Frankly, I'd be pleasantly surprised if someone else pops up, buys the components, raw board, and assembles one of these. I'd like to proven wrong!
DipTrace allows you to specify mounting holes that are not plated, it's a standard feature.
FWIW, my little board was just a demo to show you what is possible, and that wasn't even the full "shrink" that you can do. Even using fairly large SMT components, you can make boards much smaller as you can put the components on both sides and you don't have the thru-hole clearances on both sides. Though I used 0603 sized components, you can larger sizes for easier handling and soldering. 1210 resistors are a good choice as they are 1/4W, a popular power rating for many projects.
Another bonus is many SMT parts are actually cheaper than the thru-hole counterparts.
rtr12 posted:GRJ, that is a good trick about the print out, I'll have to remember that one for future use. So far I have only gotten small quantities either 3 from OSHPark or 5 from JLCPCB. Only thing I have ordered larger quantities of are the TIU Tester PCBs, which I have probably 18 or so left, with parts to go with them. Orders have dropped off lately.
Rtr, I would be interested in the TIU tester board. Could you share some details please and pricing?
Thanks, Rod
Stan, I'm on the case! I'll make some revisions and see what we can come up with for everything suggested. I'll repost the revised results here in a bit. I like the idea about the alligator clips and I see what you are talking about now. I was thinking along those lines too (where to hook the meter for ease of use), just hadn't put it all together yet. There are holes in the TIU Tester for the wires, they are 2.54mm I think and there are 2 of them. I only used one for both wires on my TIU tester so that might be a good size for the hole.
GRJ, I used the Diptrace selection for 'mounting holes' on the above PCB, at least I think that's the one you are pointing out of here. I do plan on trying the SMT parts and the 1210 size sounds like a good starting point for a greenhorn! I think I can do that, but I'll have to look into using both sides of the board, I don't know how to do that yet. I have fiddled with the 'layers' a little though. I am guessing that is key to placing things on the bottom. I have noticed the low pricing on the SMT components too, some are really low!
Rod, the TIU Testers are $17 shipped for one complete kit. The kit and PCB are all through hole components, not too difficult to assemble. They are of the final design GRJ posted in the TIU Tester thread (in case you haven't seen it). There are also some instructions compiled from some of the posts in the thread. If you would like one, email me and I will send you the complete details on ordering. My email is in my OGR profile.
Tom, using both sides of the board is as simple as just selecting the other side in the upper drop-down. Select the bottom, and you're working on the other side.
Similarly, if you want to move components to the back, just right click on the component and select "Change Side", the component will magically be on the other side. If you have a lot of components to move, hold Ctrl while you select them all and you can move them all to the other side in one operation.
The autorouter will add any necessary via's as needed to connect everything when you route the board.
That sounds much easier than I was expecting it to be. Should be pretty easy following your instructions here.
There are still several menu options in Diptrace I have not yet tried, learning is ongoing...
Stan, it's still the weekend and I got some changes made. Comments, corrections, scrap and start over, comments welcome and also wanted. Actually, OSHPark has a 'Pause Order' feature, which I took advantage of earlier (they must have anticipated my use of their order system ). I think the paused order can also still be cancelled at any time.
It's still the same size. Per GRJ's assistance above, trying the SMT components and making a smaller PCB is next on the list.
We will see how the SMT experiment goes for final size. Then once a prototype is approved and working, I would be in for offering these as kits and possibly even assembling a few if there is any interest.
Yup. That's what I meant. As I see it, if the primary objective is to learn more about PCB layout, I'd cancel the thru-hole board order and focus on the SMD version...which, if I understand GRJ's example, gets it down to less than $1 for 3-boards shipped? Wow.
All I hear are crickets chirping in regards to anyone interested in buying this "tool"! The design is simple enough that it's easier to solder it together with whatever components, connectors, whatever a DIY'er has lying around and off you go. As to assembling/selling this, I think the OGR crowd is better served with more complex designs In other words, something like the DCS port tester which is a unique circuit and tedious and impractical to assemble without a PCB.
It is curious that this TVS "tester" and the DCS port tester are cousins in that they relate to the voltage spikes. This reminds of yet another widget I hand-built also simple enough as to not require a PCB. There have been several threads about it but the video is the most fun.
Again, component cost is 25 cents or so. One red LED flashes when a positive voltage spike is detected on the track, the other red LED flashes when a negative voltage spike is detected. Green LED shows power on the track. On my to do list has been to integrate this function with a 75 cent electronic "tally counter". This would count the number of spikes. The missing link is to economically simulate a button press on the tally counter whenever the voltage spike detector is triggered.
So here's a case where a PCB would be useful not so much because the circuit is complex but because you'd probably want a handful of these scattered around the layout rather than a one-off tool. Then, after an operating session it sure would be interesting to see if the voltage spikes on the layout are more frequent in certain locations, or near/follow the engine if dragged around on a powered truck as shown.
I wasn't happy with the big open space in the upper left corner so I put the strain relief hole there. I can arrange things better than Diptrace, but my arrangement skills still need work. Diptrace is worse, but it's really good at routing the traces, that is if you remember to tell it to do that! The strain relief holes are 2.54mm, That is what the TIU Tester holes were as near as I could read the ruler.
I'm still on the SMT case as well! Here is my first attempt at SMT PCBs. You guys have finally put me over the SMT edge. I am not down to GRJ size yet, but this gets it down to $2.60 for 3 PCBs shipped from OSHPark. It uses the 12 size/number SMT components (gotta start somewhere) so the design can still be refined down some. I am still going to try GRJ's suggestion of putting stuff on the back side too, so that will likely reduce the size some. I am not sure how close I can get to the edges and a few other things like that with spacing. At least this gets rid of the big open space in the upper left corner.
I could have had this done sooner, but I got into a battle with Diptrace over 'pads covering the silk' or something like that. I finally beat Diptrace into submission and won the battle by figuring out how the pad things worked. At least enough to get rid of the errors anyway. I am sure it has brutal revenge planned for me in some other section and is just waiting for me to innocently enter into that area so it can pounce.
I know very little about the SMT components, other than I can barely see some of them. My reason for starting out with the larger size, not sure that is called 12, but parts started with that number and were large enough to see. I have a couple of GRJ designed PCBs here for TIU protection and I can only see one of the five parts that are supposed to be soldered to the PCB. He told me they have them even smaller than that!
Don't despair just yet on the TVS tester interest. Things are picking up rapidly, one person wanting one of them already! I don't plan to sell them, I was just going to share the Diptrace files here for anyone interested. But I would also be willing to offer kits at cost, as I do for the TIU Testers. Maybe even assemble a few for folks that can't (or don't want to) do that part themselves. That would make it easier on some folks, may even promote more interest.
I remember that widget you built, I would love to have one of those too. I remember wanting to make one myself at the time. Not sure now why I didn't. Probably lack of knowledge, too many irons in the fire, etc. That is a really neat little module and adding the tally counter would put it over the top! I'm sure you are much more capable than I am, but I would be happy to try and make a PCB for it in Diptrace, if you ever want one. I would think this item would be of interest to many forum members as well. As you say some very interesting information could be gathered with it by using it as you suggest.
Post the files and I could make that board about half the size while not losing any functionality.
Move the four active part on the opposite side.
Optimize the locations of the four TPx holes with the holes between them.
Squeeze everything down a bunch.
Holy smokes - this thread reminds me of electrical circuits class back in college! Not one of my favorite classes
I ordered a batch of the diodes from Mouser...and figure on replacing them once per year.
One aspect of the protection "location" for the diodes is still unclear to me. A number of you have said that the optimal place for the TVS diodes is in the engine itself. But, to my rudimentary understanding of voltage and amperage, if a voltage/amperage spike occurs, doesn't it have to originate from the transformer (after the derailment or collision of trains on the tracks)? Maybe I am mixing this up with short circuits, i.e., when a train derails or collides with another train? That short circuit causes the voltage/amperage to spike from the transformer, and if has circuit breaker protection the breaker trips and shuts off power (but the damaging "spike" already went out via the terminal posts to the tracks). Thus, a TVS at the terminal posts does the "protection" before the spike reaches the trains on the tracks?
The best place for transient protection is near the actual device or circuit to be protected. The spike can be generated by any inductive device, motors, track switches, trackside accessories, or the transformer. The transformer is not the only device that can generate damaging transients, they can come from many sources. As we have discovered in some testing done by Adrian, even the chokes we use to improve DCS performance can, under the right circumstances, generate damaging transients.
GRJ,
Here are the Diptrace files for the TVS tester. I'll include them all, through hole and SMT versions, in case you want to fiddle with both.
Everyone else is welcome to these as well, I think they should work ok. But, I have not yet gotten any PCBs made to test out, so they are fully untested as they are now.
That is pretty small! I see how you arranged things on the back and I like that better than what I had. I think I'll place an order, can't go wrong for $1.20. And I can still see the SMT parts too!
40 cents for the raw PCB. Maybe 10 cents for the parts. Now we're talking!
The real conundrum then becomes how to handle the USPS question. So if you go with the SMT version, you can get the thickness under 1/4" (if you don't ship the cables with alligator clips) and you can use the USPS large envelope rate of just over $1. Otherwise, you need to use the USPS package rate which starts at about $4 albeit you get tracking
It always bothers me when you pay more for shipping than the contents. Yeah, I need to get a grip... 
rtr12 posted:That is pretty small! I see how you arranged things on the back and I like that better than what I had. I think I'll place an order, can't go wrong for $1.20. And I can still see the SMT parts too!
It could have been even smaller, but I got to thinking how hard it is to handle something that's only half an inch square. I actually did that to give you an idea of what you can do by using both sides of the board. Also, with the SMT parts all lined up like that, hand soldering becomes pretty easy, that was one goal here.
My technique for SMT builds...
In a more complicated board, you'll also want to think about the order you put parts on, usually I start with small parts and work up the scale. You can also prep a number of the positions with the solder dot so you can just place a bunch of parts without going back to the solder step.
In truth, for something like this that is intended to be used the way it will be used, I'd probably make it larger and have mounting holes to mount it on a piece of wood or Styrene to allow it to sit flat on the bench while I used it. It may require three hands to use this tiny board to get the TVS devices.
Stan, I agree on the shipping, it is becoming more than the actual cost of the item being shipped. The postage rates have gone up in the last few months too. I have never tried using the large envelope with less than the 1/4" thickness. I don't think I could get the TIU Testers down to 1/4" thickness or I would try that. I have received one check in a regular envelope that had been half destroyed, required a redo, but so far I have not heard of any of the TIU Tester kits being lost or damaged (keeping fingers crossed).
Been using Paypal shipping for the TIU Tester kits which has also gone up. They are in a padded envelope and are over 1/4" thick, but only weigh a 2-3 ozs. total. I think Paypal is still slightly less than using USPS directly, but you have to buy labels to print and padded envelopes. If you take it to the PO they will do all the labeling for you. I am not sure if they have one that would work here, but they do have some free envelopes as well, I think. Then there is usually a nice long line to wait in that can take a while. Sometimes you just can't win.
GRJ, I am still going to try some of the SMT sized PCBs, just for the experience even if I end up using the larger one. AndI will follow your tips above during the 'experience'. I have been wanting to try some SMT stuff so this is a good and inexpensive starting point. Just wish I would have ordered the SMT parts the other day when I placed a Digikey order...too much thinking all at one time, must have overloaded the mental circuits.
As Stan points out above, the shipping may be cost prohibitive on the TVS Testers? I know some folks won't want to do this, but it would probably be much less if they ordered their own PCBs from OSHPark with free shipping and then got some components from ebay. Digikey has even raised their shipping costs. As you probably know their first class is now $4.99 and if it's over 14 ozs. it's Priority mail or Fed-Ex, $8.99 for either choice. If you order like a dozen 14 or 16 dip ICs the anti-static tube puts you int a larger box and you automatically got the the $8.99 shipping. Still can't win...unless you have devised a way around this, you are pretty good with all this shipping stuff too.
You should be able to ship the TVS tester in a First Class envelope with the SMT parts. What I do is use a piece of corrugated cardboard and cut a hole in the center. Then I put the item in the hole and put clear tape on both sides. Another trick to save a little thickness is to specify the .031" thick boards, there's no reason you need 1/16" thick boards for this kind of project. Obviously, if you solder wires with alligator clips on them to it, they probably will need to be in a padded envelope.
The cardboard is a good idea, and you just use a 'regular' mailing envelope as in for a letter, like I think Stan was talking about. I was wondering about damage with those skimpy envelopes. Thinner PCBs is a good idea too! I'll start saving more cardboard packing materials.
Thanks for the tips, I knew you would have something already figured out on this.
The components and the boards are pretty robust, I have no issue sending them like that. I tape the cardboard insert in the middle of the envelope so the automated postage cancellation or the imprint they sometimes put on the bottom of the envelope doesn't run over the contents. I've sent a bunch of stuff that way without issue.
That sounds encouraging, especially the no problems part. I think I am sold on trying that method, that is, if anyone ever wants another one.
I haven't had any problems with the TIU Tester kits themselves, but I did get a check that had been partially destroyed for one order. It arrived in a USPS envelope that contained the original envelope (torn up and partially missing) with a note about the letter being damaged. I emailed him about it, then went ahead and sent the kit along with the destroyed check. He sent another check that arrived in one piece the second time. All was well.
All things considered, I think the USPS really does a good job for the amount mail and packages they handle every day.
Access to this requires an OGR Forum Supporting Membership
