Designing a Transformer--The Right Way!

Classic Toy Trains Forum
1

As some of you may know, I rather want to manufacture model trains and sell them for very inexpensive pricing, for which I was roundly ridiculed. Having had done one of my typical things (that is, ignored said ridiculing), I am continuing to come up with ideas. One is for transformers–I aim to have a complete product line of O Gauge supplies, after all. So, to start with, here’s why I don’t like the transformers available from other sources.
Lionel: Expensive, and produces a chopped sine wave. Yuck. Prices range from $195.99 for an 80-watter to only $999.99 for a top-o’-da-line 620 watt transformer, plus they also have junky DC wall-warts. Any pure DC source is immediately disqualified, as that does diddly-squat to run vintage trains. Especially if it has no voltage control, as that does doubly nothing.
MTH: Seemingly nothing on their useless website. Plus, chopped sine waves (though better ones) on the stuff that is hanging around used.
MRC: AC? What AC? Moving on…
Menards: DC wall-wart. No! No! No!
Used: Good stuff, but no way can I sell it with new trains.

So, that leaves me with no option but to manufacture the transformers myself. Too, there is another issue: Large transformers suck. They’re neat, but they are too inflexible. A modular system would be much more useful: 100-watt transformers (with no voltage controls, just 18 fixed real AC volts and nothing more), control modules, and the like. Let me outline my plan:
Power Transformers: Convert 120VAC into 18VAC, with a rating of 100 watts. Will feature a circuit breaker, pilot light, and an on/off switch. What else do you need? They are designed to be connected in parallel, so that we only need to offer one transformer. The transformer itself can be purchased from Mouser for less than $20, the circuit breaker, light, switch, case, and other hardware will be inexpensive. Guesstimated Retail Cost: $50.
Control Modules: Fully separate modules designed to be wired in series with a power transformer. There will be multiple control modules available. Diesel Control Module (DCM): Features a notched controller to simulate that of a real diesel locomotive, a direction lever, a horn/whistle button, and a bell button. Capable of handling 10 amps. Steam Control Module (SCM): Same as DCM, but throttle is not notched and has more speed steps (how does 16 sound?). Voltage Control Module (VCM): Just a throttle, which is notched and drops the voltage by 3/4s of a volt every notch. It can be used to limit the output voltage of the locomotive control modules (for example, to just 16.5 volts so that no locomotives go around the curves at Mach Stupid), to provide a conveniently-limited voltage to accessories or the like, or even just to run something like a trolley.
This single transformer and these three control modules are the basis of the system. However, here’s the real question that I struggled with for a very long time. How do I control the voltage? The ideal solution would be a variable transformer, just like Lionel made up until they switched over to abominable systems using various forms of chopped sine waves. Sadly, there don’t appear to be any applicable variable transformers for sale, and, worse, they would be exorbitantly expensive to manufacture from scratch. Excepting outsourcing, which I would like to avoid for a variety of reasons, then, a true variable transformer is not practical. Of course, the simplest method would be a cheap-cheap rheostat, but I do not want to use a rheostat. That would cause a lot of issues. So, what other simple ways are there to have an consistent, controllable voltage drop? Since this is AC, what about diode drop? By using a rotary switch and a bank of diodes, I could keep a true sine wave, have a consistent voltage drop, and it would not be expensive. For the switch, I could either purchase a standard notched one, or I could make a switch myself–which would be much more practical for the SCM, while the DCM could probably use a pre-made switch. The VCM would probably also have to have an in-house switch. There would also be a little slide-switch on the back of both the DCM and the SCM, which would prevent turning off power when the throttle is notched down to zero. The reverse lever still would turn it off, though. In this way those control modules would function like the deadman’s handle on old American Flyer transformers.

What do you all think?

1 Like
2

I have read though this post several times and am not sure what kind of trains are the target market. Is it prewar and postwar conventional trains only? Postwar plus modern conventional only engines? Both of the above plus command control engines? What we want to operate determines the minimum power supply features.
Lionel makes some great power supply choices for modern command control engines. For Lionel only operation (no MTH) the ZW-L works great. Remember it has four built in Powermasters, foldback circuits, breakers, and it outputs 180W to each of four channels. It obviously does not have “wall-warts.” It can be directly controlled by the Legacy system. Total output is limited to 620W. Running in command mode at 18V the output is almost a pure sine wave. There are some higher order harmonics but very low amplitude. I have two of them powering eight power districts on my permanent layout. My Gilbert engines run better with the ZW-L than when using a postwar pure sine wave transformer.
The 180W/channel is output power, keep in mind all prewar and postwar transformer ratings are input power so it is necessary to multiply their ratings by 75%. A 240W postwar single control transformer would have the same output as one channel of a ZW-L or of one PH180 brick. There are some O gauge operators that run four powered units on their trains. The current draw is well in excess of 10A. For that, Lionel makes the PowerMaster 360 that parallels two PH180’s in order to supply 20A to each power district. None of these items can be called inexpensive, but there are reasons they exist. The PH180 works well with MTH engines as well as Lionel. Not everyone needs them. The ZW-L is $850, the PH180 is $206,
I think there is some demand for new pure sine wave transformers for postwar engines. The challenge is there are so many postwar transformers available that work or can be rebuilt it is hard to sell enough new items to be profitable.
One problem to be solved is intermixing postwar and modern conventional engines. The postwar engines need 5V to 7V to start moving, where the modern can motor engines are already racing around the track at 5V. I solve this on my conventional only layouts by using MRC AH101 transformers with the tethered remotes. Among many other features of the AH101, they have an adjustable, settable starting voltage. I set it to 2V, the modern can motor engines work well, the postwar 3-pole open frame universal motor engines ignore it, the reverse units do not even cycle. Unfortunately this transformer is long out of production.
The postwar engines I have run better at low and medium speeds with the chopped sine wave of the AH101 than they do with my 19B postwar transformers.
Be careful with the track power always on concept, the American Flyer dead mans handle analog. My 15B and 19B transformers have that, it works great with Gilbert engines. The modern can motor engines from Lionel and American Models continue running at about a scale 50mph unless the handle is lifted.
I would like to see some new AC transformers/controllers on the market. 100W at 18V is only 5.5A, I think a 10A version would also be required to meet market demand. This leaves open accessory power. I use separate power supplies for that, but many O gauge operators do not. Accessory power would subtract from what is available to supply the track. The tapped bank of anti-parallel diodes is a good way to control speed. The minimum steps on the market today is 32, most of us find that inadequate. 100 speed steps works well (TMCC), the 200 in Legacy is probably overkill but it does provide much finer low speed control for starting and switching. Even 32 speed steps is 64 diodes. Just eight speed steps from zero to full throttle would likely result in jerky operation.
This is perhaps sounding negative when I really do not intend it to be. Its just that power supply design for AC model trains has a lot of challenges because there was no standardization in the design of the conventional type engines.

2 Likes
3

Oh, dear. I thought that I’d specified that. :man_facepalming: Anyways, I’d rather that the whole system be flexible enough to run all of the above.

I agree, it does sound like it would work well for your application in particular. I can’t justify something like that for myself (I balked at paying $70 for a Lionel VW), but I do suppose that it would work for others. By the way, what store is it for sale at that price? I haven’t used that specific transformer myself (or the MRC AH101), but I have used power supplies with remarkably non-sine-like waves and have been so disgusted as to never want to touch such things again. I s’pose that it does revolve around the specific wave outputted.

I agree. To be honest, I’d probably leave that segment of the market alone except I’m also looking to do train sets, and can you just imagine the outcry if I boxed up a vintage transformer with new trains?

I’ve been working on deciding on the exact starting voltage to use, I’ll try 2 volts and see how my own cantankerous locos work.

I’ve noticed that about the 18B that I somehow managed to patch together as well, hence the reason I’ve elected to include a switch to deactivate that feature.

I agree, it is a pretty low amperage relative to a Flyer 18B, a Lionel ZW, or even that enormous Lionel ZW-L. Part of the issue (or advantage) of my system is that it’s a complete shift in design, away from large transformers to a bank of smaller ones. I’d like to stock only the one Power Transformer, meaning that I don’t need a 200-watt behemoth to power train sets but I also don’t want to sell a little 25-watt thingy. In order to market my modular system, I have to convince customers to buy multiple transformers and put them in parallel–or else! On the other hand, if I really wanted to I could maybe cram two transformers into the case in parallel (By golly, it’s darn near the Lionel TW all over again) and sell it as 180 watts–which, by the way, I forgot, the wattage of the transformer itself is more like 90-odd watts, not 100. It’s thus kind of like a Lionel 1033 in that regard.

Oh, good. I was hoping that I didn’t just have another random, bizarre idea that wouldn’t work. Looks like I’m going the right way.

I agree, hence the reason that I would also sell a control module for steam, with more speed steps. That said, if I recall correctly, my Menards FP7 has, I believe, eight speed steps, and it sure don’t feel jerky to me–though it does have momentum, so… I wouldn’t even sell an eight-step version except for the fact that real diesels use that and I’d be able to sell such a module for less money. Of course, if one runs a good deal of both and does not wish to swap the control modules, it might be best just to get steam.

Can’t say that I find that to be a low number. I’m going to take a look inside my conventional transformers and see how many windings are exposed. I have to say that they are plenty smooth, and they do have a finite number of speed steps–though certainly more than some of those prewar Lionel transformers like the ‘K’!

4

It might be worth looking at how Right of Way back in the day made their 400W beast.

They used two big 24V fixed voltage transformers, then put a panel-mount variac on the input side of them. Of course you don’t want to connect a variable autotransformer of any sort to the track directly, but using it to control the line voltage into a fixed output transformer is a good way to get variable voltage.

I agree with @AmFlyerTom though that chopped sine waves aren’t evil. I find most conventional locos give better slow speed operation using them than on a true sine wave. They’re a sort of crude PWM(which is what all the good DC power packs use these days). As much as MTH seems to be against them, they also output them from their Z-controllers and the TIU variable channels-the real issue is that some ProtoSound 1 locomotives won’t run on the “wrong kind” of chopped sine wave.

5

Well, I’ve looked at that, but variacs, even low-amperage ones, are :face_with_symbols_on_mouth: expensive. That also wouldn’t work quite as well for the modular system, but I could probably also just put the variac on the low-voltage side of the main power transformer.

If you get one that’s really wrong, no locomotive’s gonna like it–I found that out the hard way with a really defective “sine” wave put out by a dimmer. I will consider various systems putting out chopped sine waves, and I’ll also look at variable frequency control (which is not unlike PWM, although I don’t like PWM either–long story). Part of the issue, at least to me, is that some of those components are not inexpensive–though certainly cheaper than a variac!

6

I’ve not priced them out, but the ROW transformer used fairly low current panel mount units.

In that case, I think the output transformers were rated at 10A(although the breakers were smaller than that). A 24V transformer supplied with line power has a 5:1 winding ratio, which means that you need a 2A rated variac(or probably a little bigger). Of course that’s still not cheap-a bit of quick searching found 2.5A rated ones for $250 from one supplier(building for myself I’d buy used, but not for a production).

There are a lot of components that really are going to drive the cost up-even good circuit breakers that I’d feel comfortable supplying for modern trains are going to run $10-30 each.

At the end of the day, though, you’re likely going to want to have any transformer you sell tested by UL or one of the other consumer products testing labs(I think Lionel uses ETS for a lot of stuff). Do you have to do this, no, as far as I’m aware but I’d also hate to somehow be linked to a house fire where a transformer I’d supplied was present and there was no independent safety testing on it. Bear in mind too that, as far as I know, testing is for a finished product(so it doesn’t matter what rating any individual components you buy may carry) and also, for a modular design, each component of the module is likely going to be separately tested. If I’m not mistaken, aside from the cost of testing, you’ll need to “sacrifice” a unit for the lab to do destructive testing.

Well, light bulbs tend to not care about the quality of the power(or at least old incandescents-some newer ones do) and most people don’t want to pay a fortune for a dimmer. I’d not use that as my basis for the suitability of a chopped sine wave.

7

The prices I listed are from the Charles Ro website. I buy items from them several times each year.
The Right of Way Industries transformer was one that I liked but never purchased. If made today it would likely sell for more than $500, not the cost Mark is trying to achieve.
I would recommend against variable frequency control for AC powered trains. The motors are designed for a specific frequency, supplying them with a different fundamental frequency (rather than just low amplitude harmonics) causes a decline in performance and generates a lot of winding heat. If the PWM is recified in a decoder and then supplied to a DC motor that is fine.
The issue with command control engines is what kind of track power they work with varies. TMCC and Legacy prefer AC, either sine wave or chopped. DCS seems best with sine wave but the newer PS3’s work with any AC. LionChief etc is another animal entirely. Some come with DC power supplies in boxed sets, but do not use that supply for TMCC engines.
Transformer cores can be connected in parallel with a single controller and a single track output. That is exactly what the Lionel PowerMaster360 does with two PH180’s. Like Lionel does, I recommend limiting it to two transformer cores. If you make each core 5A at 18V then two are 10A at 18V, pretty much the two values the market wants.
Going to track voltages, I think based on my experience, 2V is a good number. To get that we need a string of 10A rated diodes. A reasonable assumption is many layouts will have about a 2V drop between the transformer and the engine internals. That drop is resistive so it varies linearly with current draw. So, assuming we start with 16V and want 2V as the first step, we need a 14V drop. Each diode is about .6V, so we need either 20 or 21 pairs of diodes connected in antiparallel. What I would try is 21, make the speed steps every diode at low voltage then go to 2 or 3 diodes between each step at the higer voltages. This could result in 10 usable speed steps.
As far as packaging all this, that is controlled by safety codes. It will be a plastic case with double isolation. Since you will not be passing the Legacy signal back to the house wiring ground wires with this power supply, a three prong plug should not be used.
By the way, your wattages on the old Lionel transformers are advertised input power, not output power as we are discussing here. Even the Lionel ZW, rated 275W, is only about 210W output power based on the modern transformer ratings. What makes it so great is all that power can be sent to a single handle if needed.

8

I’ve definitely seen variacs for lower prices (albeit very large ones), but none actually in the price range I want. Circuit breakers aren’t too bad–the right kind of automotive-style ones are pretty inexpensive and very reliable.

Yeah, I’ll want to jump through those hoops. At least if I can get the price low enough the “lost cost” in the transformers sent will be relatively acceptable.

That is very true–this particular dimmer was so horrendous that it frelled with LEDs. Of course, that’s what I get for buying the cheapest one on Amazon. I’ll definitely have to investigate some such solutions.

Ah, thanks. They have a noticeable markdown over Lionel themselves, it would appear.

That’s very much my plan. Cheap, cheap, cheap, cheap. If the price for mine came out that high, then I might as well just forget the whole thing and package Lionel CW-80s in my starter sets.

Yeah, I’ve noticed that too. Some of my locomotives are more tolerant thereof than others, too. Just kicking ideas around, I guess.

This is part of why I want to do a pure sine wave–it’s not 100% perfect for everything, but (with the exceptions of the occasional oddball vintage Lionel DC train) almost everything will run on it without being damaged. And my Lionel Pennsylvania Flyer did include such a DC wall-wart.

Oh, good. That’s kind of what I was looking at: sell it as “ya wants some watts? Buy one. ya wants lots o’ watts? Buy two (or more)!” In that case I’ve about judged the specific numbers correctly. Say, didn’t Lionel used to sell a version of the ZW where you plugged in up to four PowerHouses? Aha, here it is.

My estimates were based off of .75 for each anti-parallel diode pair. I will have to double-check the diodes which I have and make sure. My estimates were that, for the DCM, I would need to arrange it like so, based off of .75v drop per diode pair.

Notch/Speed Step Voltage Out (Est.) Voltage Dropped (Est.) # of Diode Pairs
0 0v N/A N/A
1 4.5v 13.5v 18
2 6v 12v 16
3 7.5v 10.5v 14
4 10.5v 7.5v 10
5 13.5v 4.5v 6
6 15v 3v 4
7 16.5v 1.5v 2
8 18v 0v 0

The SCM would be more like this:

Speed Step Voltage Out (Est.) Voltage Dropped (Est.) # of Diode Pairs
0 0v N/A N/A
1 4.5v 13.5v 18
2 5.25v 12.75v 17
3 6v 12v 16
4 6.75v 11.25v 15
5 7.5v 10.5v 14
6 8.25v 9.75v 13
7 9v 9v 13
8 9.75v 8.25v 12
9 10.5v 7.5v 10
10 11.25v 6.75v 9
11 12v 6v 8
12 12.75v 5.25v 7
13 13.5v 4.5v 6
14 14.25v 3.75v 5
15 15v 3v 4
16 15.75v 2.25v 3
17 16.5v 1.5v 2
18 17.25v .75v 1
19 18v 0v 0

I’m ignoring the approx 2v voltage drop here, but factoring it into the starting voltage (thus bumping it up slightly).

That is right in line with what I was thinking.

That is true.

Yeah, I guess that I’ve been sloppy in that regard. The transformer that I’m looking at using is rated on output at 100 watts.

However, there was a frell-up, the transformer cores will cost just under $30 each, not $20, and are rated at 100 watts.

9

A 100w core rating (asuming it is output, not input) is fine since your two standard configurations would be 90W and 180W.
Use the forward voltage drop value for the specific diodes you install. .6V is just a typical value we use.
Automotive circuit breakers are fine for conventional engines. They are not nearly fast enough for engines with command control electronics.

10

Okay, good, that matches up. The transformer has an output rating of 100 VA (i.e. 100 watts) Unfortunately, the majority of the components are at Mouser, while the diodes are at Amazon, which is even worse at providing data. I managed to dredge up a data sheet for what is (I hope) the same diode, but had no luck finding the voltage drop, other than that it is supposedly low. I dunno, maybe I’m just tired at missed it. Here’s the PDF in case anyone here is better at finding such notes than me.

Hmm, mine must be under the Cavalier Effect. I’ll look for some more conventional ones.

11

I’d be reasonably certain that the ‘Maximum Instantaneous Forward Voltage’ is missing the word ‘drop’. Obviously the actual working voltages for the various devices are as specified in the first two rows (Vrrm and Vrms), and they are much higher; I’d think 0.95V at 10A forward current would be the magnitude appropriate for voltage drop. Of course it would be nice to have a curve for voltage drop vs. current…

12

It’s worth looking at Mouser(or Digikey) for the diodes. Assuming you’re using a 10A10, Mouser has plenty of options including different packaging types(I’d be tempted to use a TO-220 package since it’s a lot easier to put a heatsink on it) and the cost goes down buying in quantity

https://www.mouser.com/c/?q=10A10%20diode

1 Like
13

Okay, that’s what I was leaning towards but I was not sure. One moment and I’ll rework the chart to reflect that.

Issue with Mouser is that, while normally they are frequently cheaper, the 10A10 is not one such example. I haven’t found a 10A10 diode that beats Amazon’s $0.1706 each price–the closest equivalent is the 10A10-B, which is $0.98. It doesn’t even begin to approach Amazon’s price until you order like 2500 of them. As far as I’m concerned, the lower the minimum order, the better. Amazon sells 50 of 'em, that’s low enough for me. I see no point in ordering 2500 diodes when I don’t need 2500 diodes.

14

Okay, here’s the new table, based off of the .95v voltage drop.
DCM:

Notch/Speed Step Voltage Out (Est.) Voltage Dropped (Est.) # of Diode Pairs
0 0v N/A N/A
1 4.7v 13.3v 14
2 6.6v 11.4v 12
3 8.5v 9.5v 10
4 10.4v 7.6v 8
5 13.3v 5.7v 6
6 15.2v 3.8v 4
7 16.1v 1.9v 2
8 18v 0v 0

SCM:

Speed Step Voltage Out (Est.) Voltage Dropped (Est.) # of Diode Pairs
0 0v N/A N/A
1 4.7v 13.3v 14
2 5.65v 12.35v 13
3 6.6v 11.4v 12
4 7.55v 10.45v 11
5 8.5v 9.5v 10
6 9.45v 8.55v 9
7 10.4v 7.6v 8
8 11.35v 6.65v 7
9 12.3v 5.7v 6
10 13.25v 4.75v 5
11 14.2v 3.8v 4
12 15.15v 2.85v 3
13 16.1v 1.9v 2
14 17.5v .95v 1
15 18v 0v 0
15

I think forward voltage drop may vary with amperage, which is why I wished for a curve. It is possible that your voltage drop may require adjustment based on drawn current – what if the drop is closer to the 0.6V, for example, at 2A than the nominal 10A?

16

Oh, that’s worrisome. That’s part of what I was trying to avoid by using the diodes in the first place. I will have to go tinker with my box of diodes and see what’s what. If I recall correctly, it seemed constant with varying amperage when I had a rectifier hooked up on my test track, but that whole system’s a big, fat mess at the moment and I will need to take a closer look.

17

Voltage drop usually does not vary much with changes in forward current through the diode. The other thing to keep in mind is that track power rarely is at the peak rating of the power supply. With a 180W, 10A supply, the normal continuous current draw is 2 to 4 amps. My ZW-L’s have reasonably accurate Ammeters built in. The only way I see close to 10A is when there are three engines in the same power district.

18

That’s been my experience too. None of my locomotives (that I’ve measured the current draw of) have ever really sucked up more than 2.5 amps. A 10-amp/180 watt-rated architecture would definitely be very cost-effective and suitable.

19

I know that track layout, length, and size is a big deal when dealing with power. DCC vs DC modern or pre/post war. Can it power accessories or not. how repsonsive is it for different trains in power. If possible I would purchase an expandable power supply that is modular and alings with my DC-EX Sytem to include the accessories. Similar to what my T Guage system is

20

I agree, there’s a lot of different factors. Maybe I should try to also market a version viable for small scales, with lower DC voltages.