Fx Bricks (Michael Gale) announces Fx Track system

[zephyr1934]

@michaelgale You are doing amazing work and contributing so much to the Lego train community.

Just splitting hairs, but why five bogie plates per pack? I would think an even number would make more sense.

On a more practical note, the test bench is impressive. I think at one point in the post you wrote that you have multiple power leads to each the track. Are you also testing it with a single lead? Or probably a better question, do you plan to provide guidance to users, e.g., "we recommend using a additional power connector after every N track segments." The power drop across track segments is a result of the original Lego design and your double straights already cut it in half. But for folks transitioning from plastic to metal rails I bet you would remove some unpleasant surprises if you make sure a customer knows the importance of having multiple leads before finalizing an order so that they get everything they need in a single shipment.

Are simple dumb switches equivalent to the poll reverser somewhere in the "future wish list"? I could see it being beneficial to be able to isolate a siding or block. Definitely not the top priority but a useful element in the completely built out system.

Oh, and are those remote uncouplers in one photo? This all looks amazing, you are bringing the 7777 to life.

Are you testing the new hardware on the original Lego track too? In particular R40 curves and the tight switches. What about testing the new controller with original Lego motors? Are the new controllers true DC or are they PWM?

As you get closer to the release of the new motors, I would love to see a pulling power comparison of your new motors against the original rubber-tired Lego version. Where the train, track etc. are held constant but the motors are swapped.

 

[LM12V]

Incredible! The exchanges didn't last a day, they're already sold out. I hope we don't have to wait another year to see others that last a day.

In any case, I hope to see the new products put on sale all together, because for shipping costs it is better to place a single order than multiple orders.

 

I ask @michaelgale:

 

1) - wasn't it better, for the wheels, to make a box of 4 or 12 or think of more modes? If you think of a train at least you need 10-12 wheels. I'll take about 80.

2) - Do I see this strange segmented track in the photos? Is it perhaps the prototype of the traffic light track?

3) - Shouldn't the power supply also include 12v? I'm aiming for your brand new engines and it would be annoying when they come out to have to change the power supply.

4) – I see the 4x6 plate but not the 5x6 one you announced. There are many 3rd party manufacturers making 4x6 and I doubt the fx pricing will be competitive. Wasn't it better to focus on the 5x6 that would be truly original?

5) – The wheel cover of the lego engine is really ugly and not very realistic, is a restyling planned in view of the release of the FX engine?

6) – I'll ask you again about the conversion of old 12v trains: when the intelligent engine comes with everything integrated (with the first one it's not worth it), why not make a red and black version with connecting rods for the old freight trains of the 12v era? (A 12v-9v adapter cable would also be useful, although I had them made. Thanks to the FX wheels you can run the old 12v motors on the 9v track, but you cannot use the professionalism and torque power of the new FX motors)

 

 

[Modeltrainman]

Can't wait for the track feeder, power stuff, metal wheels!

[BatteryPoweredBricks]

On 2/14/2026 at 2:45 AM, zephyr1934 said:

Oh, and are those remote uncouplers in one photo? This all looks amazing, you are bringing the 7777 to life.

I made a video nerding out over the blog post and one of our discord members pointed out it looks like an isolation segment.  There is a gap between the metal rails and the black component could be a relay for switching the power.  I assumed it was a decoupler as well but I didn't notice the gap.

Also huge thanks to Michael for taking the time to write up the post and sharing pictures of the development process.  I've spent far too much time trying to identify each and every thing shown in the photos 

[GoHabsGo]

Thank you so much for the update. I cannot wait for my order :)

[michaelgale]

On 2/14/2026 at 3:45 PM, zephyr1934 said:

@michaelgale You are doing amazing work and contributing so much to the Lego train community.

Just splitting hairs, but why five bogie plates per pack? I would think an even number would make more sense.

Sure.  But 5x of them fit so nicely and efficiently with our card holder/protector it makes the best bang-for-the-buck in terms of packaging volume.  Pick a number and I guarantee someone will prefer a different quantity!

On a more practical note, the test bench is impressive. I think at one point in the post you wrote that you have multiple power leads to each the track. Are you also testing it with a single lead? Or probably a better question, do you plan to provide guidance to users, e.g., "we recommend using a additional power connector after every N track segments." The power drop across track segments is a result of the original Lego design and your double straights already cut it in half. But for folks transitioning from plastic to metal rails I bet you would remove some unpleasant surprises if you make sure a customer knows the importance of having multiple leads before finalizing an order so that they get everything they need in a single shipment.

There is guidance in user manuals that come with both the 9100 Track Feeder and the 9005 Speed Regulator regarding placement of power feeders.  The overall principle is simply:

1. Have more than one

2. Space them as equally as you can

Are simple dumb switches equivalent to the poll reverser somewhere in the "future wish list"? I could see it being beneficial to be able to isolate a siding or block. Definitely not the top priority but a useful element in the completely built out system.

The simple switches do in fact control track isolation segments so that you can have selective isolation of sidings and blocks. 

Oh, and are those remote uncouplers in one photo? This all looks amazing, you are bringing the 7777 to life.

They are track isolation accessories; although remote uncouplers would be cool--but not sure how to do that just yet!  Any relation to the 77xx LEGO 12V accessory system is purely intentional!  (we took inspiration from it for obvious reasons)

Are you testing the new hardware on the original Lego track too? In particular R40 curves and the tight switches. What about testing the new controller with original Lego motors? Are the new controllers true DC or are they PWM?

We are absolutely testing all of our products with LEGO brand products.  The R56 test loop has a passing loop made from LEGO 9V track elements and our control console has a LEGO brand 9V speed regulator with power feeder.

As you get closer to the release of the new motors, I would love to see a pulling power comparison of your new motors against the original rubber-tired Lego version. Where the train, track etc. are held constant but the motors are swapped.

This is data we're keen to gather as well!  We should be clear that our motors will have different dynamic performance than the LEGO rubber-tire motor.  Our motor bogie will have performance similar to model train motors, i.e. high torque and traction which scales with weight over the drive wheels.  

 

 

[michaelgale]

On 2/14/2026 at 6:25 PM, LM12V said:

Incredible! The exchanges didn't last a day, they're already sold out. I hope we don't have to wait another year to see others that last a day.

In any case, I hope to see the new products put on sale all together, because for shipping costs it is better to place a single order than multiple orders.

We're working on the issues of selling out of key products by taking bigger production commitments with our suppliers.  However, we're also recognizing that there is saturation point for production capacity (at least for the level of quality and competence that we require in our injection molding and overall quality control).   In any case, we're receiving a steady stream of new P40 switches and are releasing them into our shop as we get them.  Stay tuned and sign up for notifications for new stock.

On 2/14/2026 at 6:25 PM, LM12V said:

 

I ask @michaelgale:

 

1) - wasn't it better, for the wheels, to make a box of 4 or 12 or think of more modes? If you think of a train at least you need 10-12 wheels. I'll take about 80.

If you are referring to the 9700 metal wheels, then I think high quantities will typically not be required by the average user.  The metal wheels are only required for the locomotive to power a motor.  Other rolling stock such as wagons do not require metal wheels (unless you want to connect a powered accessory such as lighting).  In general, we don't recommend fitting metal wheels to passive rolling stock since these wheels have higher "drag" or rolling resistance due to the power pickup tabs which contact the back surface of the wheel.

2) - Do I see this strange segmented track in the photos? Is it perhaps the prototype of the traffic light track?

Yes, it is similar!  The segmented track is a remotely controlled isolation track which can be used to selectively isolate a siding or track block.  The signal light is a separate and optional accessory and not combined with the isolation track.

3) - Shouldn't the power supply also include 12v? I'm aiming for your brand new engines and it would be annoying when they come out to have to change the power supply.

For our initial Speed Controller product, we decided that 9V was the best configuration since it is more familiar to existing 9V users and because it guarantees better compatibility with existing LEGO 9V accessories.  However, our motors and the speed controller itself is rated to operate up to 12V.  We have thought about releasing a "Pro" version of the speed controller with a 12V mode.  

On 2/14/2026 at 6:25 PM, LM12V said:

4) – I see the 4x6 plate but not the 5x6 one you announced. There are many 3rd party manufacturers making 4x6 and I doubt the fx pricing will be competitive. Wasn't it better to focus on the 5x6 that would be truly original?

Again, we have decided it is best to remain familiar and compatible with the existing 9V / Train system first.  Alternative variants of our product are always something we can consider for the future.

5) – The wheel cover of the lego engine is really ugly and not very realistic, is a restyling planned in view of the release of the FX engine?

We do not have plans to re-implement the cosmetic motor bogie sideframes at this time.  Every new mold costs money and we have to be careful to prioritize our product development investment on essential elements.

6) – I'll ask you again about the conversion of old 12v trains: when the intelligent engine comes with everything integrated (with the first one it's not worth it), why not make a red and black version with connecting rods for the old freight trains of the 12v era? (A 12v-9v adapter cable would also be useful, although I had them made. Thanks to the FX wheels you can run the old 12v motors on the 9v track, but you cannot use the professionalism and torque power of the new FX motors)

This is a nice idea, but would have a very limited customer base.  Our priority is to release a minimum system of products to bring all the technical benefits of traditional model trains (metal track, more variety of track elements, high performance motors, etc.) to the LEGO train world.  

 

 

[michaelgale]

On 2/16/2026 at 10:06 AM, BatteryPoweredBricks said:

I made a video nerding out over the blog post and one of our discord members pointed out it looks like an isolation segment.  There is a gap between the metal rails and the black component could be a relay for switching the power.  I assumed it was a decoupler as well but I didn't notice the gap.

Also huge thanks to Michael for taking the time to write up the post and sharing pictures of the development process.  I've spent far too much time trying to identify each and every thing shown in the photos 

Great video!  Thanks for sharing and we appreciate your analysis and perspective.  

To clarify, the gapped track segment in the photo is a prototype isolation track.  This allows remote control of isolation segments in a layout for either sidings or block configuration.  A remotely operated uncoupler would definitely be a cool product--it is on our roadmap.  However, it is lower priority compared to remote isolation, remotely operated switches, signals, etc. We would also need to figure out how it would work--e.g. would it be purely mechanical, electromechanical, or both?  Can it work with every variant of magnetic coupler (including some impressive 3rd party products with rare-earth magnets)?

[LM12V]

Thank you very much Michael for the reply. We therefore await the new products. Overall I trust you a lot, you are doing something wonderful and extraordinary.

I leave these considerations on Lego and FX wheels to the readers of the forum. I would also like to know other people's opinions.

I'm actually waiting for these wheels to light up the wagons (and convert all the old 12v to Fx). And in fact you can use just one for wagon by combining it with the original Lego ones, but in that case the color is not guaranteed to be homogeneous. In any case, I'm in no hurry. I just hope that special lights will be made for passenger trains in the future. I completely understand the idea of focusing on the essential things at the moment.

Instinctively though I would say that the FX metal wheels will be used much more for lighting than for running engines. In fact, those who have power up or power function engines have ultimately chosen batteries. Of course he can decide to switch to electric, but at that point perhaps he is better off buying an old 9v engine or waiting for the new 12v FX engine. Buying FX bricks tracks and then running the PF and PU motors on them thanks to the FX wheels, I don't know what use it will have in the long run. It can only be something interlocutory and ultra-minority.

The situation is different for old 12v trains. If there is currently no plan, due to cost issues, for an engine capable of replacing the 12v ones with connecting rods, the only option is to connect the FX bricks wheels to the old 12v engine. I already have the tender ready for the old 7727. And this makes sense, because then with a decoder you will be able to exploit a lot of the FX potential, even with this hybrid solution (in fact, I think that Micheal said that the FX wheels are also ready for the DCC). The real problem in this case is that the 12v engines from the 1980s are increasingly older and sooner or later will still need a replacement.

From a friction point of view, Lego wheels, also taking into account non-original and third-party ones, I believe can be classified as follows, from worst to best:

 

original large and small wheels with cross hole and technical plastic axle

old original 12v wheels with plastic axle (type 7740)

New original wheels without axle (Orient Express type)

non-original wheels with Bricktracks metal axle

original 2878 wheels with metal axle (Maersk train type)

Non-original Bricktracks wheels with metal axle and ball bearings.

 

Now Micheal says that the Fx wheels have some friction due to the tab to get the electricity. However, I don't think the friction is that great or it wouldn't make sense to do it that way. It would be interesting to know, before purchasing, at what level on the scale I made, the FX wheels fit. I believe they are between the 2878 of the 9v era, and those with ball bearings.

In the future, taking into account that Lego wheels are terrible at friction (2878 with separate metal axle), the problem of the FX tab could be overcome with identical wheels without a socket. It shouldn't cost any new molds. Because wheels with bearings, however good, are not professional and are often mounted poorly, not too symmetrically. Finally, in the future, it would also be nice to have professional and large fx wheels for Orient Express and Emerald Night. These could also be done without the need to draw electricity, or in both ways.

 

 

 

 

 

 

 

[Narissis]

I can only speak for myself, but I'm eagerly awaiting the power pickup axles for two projects:

One, a steam engine powered by a dune buggy motor. That motor puts so much load on the battery that a AAA PF battery box can't even run it at all. And it kills my 650mAh battery (marketed fallaciously as 1100mAh; thanks Green Gecko Workshop -_-) in only a few minutes. So that one's going on wall power ASAP.

Two, the Orient Express, which I could certainly have motorized with batteries by now if I wanted to, but given the weight of the train I think it would also have battery endurance issues so I've been holding out to use rail pickup instead. Still haven't decided what motorization approach I want to take, but I prefer for my steam engines to actually deliver their power through the large 'drive' wheels for, I suppose, a certain sense of realism. So leaning toward maybe doing two face-to-face motors connected to a common gear. We'll see!

[Stereo]

It depends on what things cost, but I expect my minimum order specific to trains is the power pickup axles and track connection wire, so I can combine it with PF remote control.  Most likely IR receiver onboard the train, for DCC-like behaviour, but if that turns out to be unreliable (no idea how much the receiver tolerates gaps in power for maintaining speed settings), wall power into the receiver into the tracks is a backup plan.  Aside from being what's available first, it just makes more sense budget-wise to continue to use the working PF motors I have.  Though I suppose for this specific task there is one thing - the PF 9V adapters don't connect to the "constant" side of the PF plug, which is what the IR receiver input needs.  Obviously the workaround is just wire my own adapters the other way.  So I might also order some short FX wires to get spare 9V plugs.

 

[Modeltrainman]

Wait, were the power feeders available? I don't see them even sold out.

[Narissis]

16 hours ago, Modeltrainman said:

Wait, were the power feeders available? I don't see them even sold out.

Not quite yet but they should be dropping soon.

[Modeltrainman]

 

1 hour ago, Narissis said:

Not quite yet but they should be dropping soon.

Thank you. Are you thinking week, month, months?

[Narissis]

Only Michael could answer that definitively but given that the retail packaging is underway, I would hope within a month or two. Maybe less if they're already in primary shipping.

[zephyr1934]

Hi Michael,

Did I thank you for all you are doing for the hobby? I think we all appreciate your perfectionist work. But in that regard, I'm not sure if all of my thinking in my last note was clear.

 

On 2/21/2026 at 3:14 AM, michaelgale said:

There is guidance in user manuals that come with both the 9100 Track Feeder and the 9005 Speed Regulator regarding placement of power feeders.  The overall principle is simply:

1. Have more than one

2. Space them as equally as you can

Someone who does not have experience with 9v might not even realize the importance of having multiple drops. And even if they do know it is important to have more than one power connection, having that guidance at the time of placing an order will help them get everything they need in one shot. You might want to provide a recommended number of track joints, and maybe quantify that with, "for heavy trains have a connector after every 20 joints, for light trains you can go to 30 or higher". I would think a customer would be much happier if they were able to purchase all the items needed in one order, not having to pay extra shipping or wait a few more weeks to get additional parts they did not know they needed.

 

On 2/21/2026 at 3:14 AM, michaelgale said:

Are you testing the new hardware on the original Lego track too? In particular R40 curves and the tight switches. What about testing the new controller with original Lego motors? Are the new controllers true DC or are they PWM?

We are absolutely testing all of our products with LEGO brand products.  The R56 test loop has a passing loop made from LEGO 9V track elements and our control console has a LEGO brand 9V speed regulator with power feeder.

I had a bad experience with Lego brand 9v motors and PWM, that might have been purely coincidence, but it might not have been. It is possible that the Lego motors would benefit from a capacitor to level out the voltage. If I'm off base, no problem, but it would be piece of mind to see, "tested for NNN hours with Lego brand 9v motor." Or if there are any issues, perhaps another product to keep it plug and play for numskulls like myself that would be a capacitor unit that one can add on at the output of your controller to do the smoothing on the "land side".

Meanwhile, it isn't so much Lego brand track but more specifically, doing loops on R40 (presumably a part that you will never have a need to produce) and possibly the tight Lego switches.

These tests just seem like a small extension of what you are already doing.

You are putting TLC to shame with your quality control

 

 

[Toastie]

13 hours ago, zephyr1934 said:

I had a bad experience with Lego brand 9v motors and PWM

Was that any custom PWM device or using TLG's RCX, PF, or PUp? They never failed on me (so far, 10+ motors).

Did you remove the thermistor in the 9V motor, or was it still in place?

Best
Thorsten

[SD100]

I'm going to guess that the PWM issues with the Lego Motor was a Thermistor problem.

On 2/24/2026 at 2:21 AM, zephyr1934 said:

I had a bad experience with Lego brand 9v motors and PWM, that might have been purely coincidence, but it might not have been. It is possible that the Lego motors would benefit from a capacitor to level out the voltage. If I'm off base, no problem, but it would be piece of mind to see, "tested for NNN hours with Lego brand 9v motor." Or if there are any issues, perhaps another product to keep it plug and play for numskulls like myself that would be a capacitor unit that one can add on at the output of your controller to do the smoothing on the "land side".

I'm going to guess that the problems with LEGO Motors and PWM came down to a problem with the Thermistor... It seems to be an incredibly finicky component. Also when it comes to power drops, we generally need more where we DON'T have FX track than when we do. A section of FX even distant from the power feeds seems to relay power better than standard track. We currently run 2 drops on about 70 feet of track, and that seems to work fine.

SD

[zephyr1934]

On 2/24/2026 at 4:14 PM, Toastie said:

Was that any custom PWM device or using TLG's RCX, PF, or PUp? They never failed on me (so far, 10+ motors).

Did you remove the thermistor in the 9V motor, or was it still in place?

I was using a RCX to do a back and forth on an open ended track. It was a stock motor so the thermistor was still in place

 

7 hours ago, SD100 said:

I'm going to guess that the problems with LEGO Motors and PWM came down to a problem with the Thermistor... It seems to be an incredibly finicky component. Also when it comes to power drops, we generally need more where we DON'T have FX track than when we do. A section of FX even distant from the power feeds seems to relay power better than standard track. We currently run 2 drops on about 70 feet of track, and that seems to work fine.

Okay, if the termistors are known to be flaky with PWM and if Fx is using PWM they probably should warn customers about that with their controllers. If I didn't know and I burned out a vintage 9v motor because it has a weak link I'd be upset. With all of the effort to make the new Fx backward compatible it is inviting folks to use the old motors. If the Lego motors are deficient it seems perfectly reasonable for Fx to say that the new controllers are only compatible with the new motors, but it is better to have a warning of that on the sales page than to apologize to customers after the fact. For some users it is no problem to crack open a 9v motor shell and fix the problem, for others, it might be out of their wheelhouse. So the warning could even include a link to a more length explination.

 

7 hours ago, SD100 said:

when it comes to power drops, we generally need more where we DON'T have FX track than when we do. A section of FX even distant from the power feeds seems to relay power better than standard track. We currently run 2 drops on about 70 feet of track, and that seems to work fine.

Excellent to hear. I bet the double length straights help a lot (half as many joints). My home loop is about 12ft x 12ft with two power drops on opposite sides and I still have slowing problems due to power drop. Only about half of my straight track are Fx though and I have a small s-curve. So a 15 ft train can be in 4 curves at once, creating a lot of drag (true for PF trains too). So I have roughly 24 ft between drops and that's probably at my limit because of my curves while you have 35+ ft. Lighter trains with few curves could probably go farther than that. 

I'm just playing the devil's advocate, someone could purchase one controller with one lead and then discover they are under-powered. Last time I ordered from Fx the shipping to the US was around $20. Knowing that I might need the extra power connector at the time of placing my first order I might be inclined to add an extra power drop and connector wires rather than risk needing another shipment with expensive postage. I'm just trying to anticipate realistic negative scenarios

 

 

[Narissis]

FWIW, I don't think thermistor issues present any risk of actually 'burning out' the motors. Just of causing subpar performance.

In the long run, the preference would of course be to run FX motor bogies with the PWM FX power supply; I'm sure those will be highly compatible.

It does make me wonder if this is a small oversight in Michael's testing, though, since I know he prefers to bypass the thermistor in all his 9V motors as a matter of course. But he's also meticulous enough that I feel like he'd test a few non-bypassed ones as well, to ensure compatibility.

Maybe it's an issue that primarily affects worn motors, so fresher ones don't exhibit problems?

I know a lot of my 9V motors struggle even on the standard first-party speed regulator.

[Toastie]

30 minutes ago, Narissis said:

FWIW, I don't think thermistor issues present any risk of actually 'burning out' the motors.

Maybe some people are worried that without the thermistor, a motor may burn out upon operating too long under too much stress. This is what they are there for in the first place.

I have taken out all the thermistors in my 9V motors, as they always caused trouble. All my non-PUp locomotives have 9V motors, two are RCX controlled, most other with PF and some with PUp gear. All these controllers use PWM for setting the "speed". On my RCX trains I have even programmed a PID control loop for controlling speed - the stuff that is now built into PUp hubs, should you use the appropriate PUp motors that have rotation encoders on board. The PUp train motor has of course not ^^. Upon RCX PID control and using 9V motors along with the blue Mindstorms rotation sensors, the RCX PWM outputs go "crazy" when either ramping up speed or trying to keep speed constant. Nothing happened so far to the motors. Here is a 15 years old post along with a video showing that (yes, video sucks, but you can hear the pitch of the changing PWM output clearly):

https://www.eurobricks.com/forum/forums/topic/45440-lego-train-control-using-rcx10-pbricks/

I don't do any shows though - things may change under full stress ...

All the best,
Thorsten

 

[michaelgale]

Interesting discussion guys!

A few points:

I. Feeding Power to Layouts

Your train layout is an electrical circuit. Unlike most electrical circuits, the size and shape of your layout circuit changes dynamically as your train moves around the layout. As the train moves, its distance to the electrical energy source changes and therefore the electrical resistance between the source and train also changes.  

Basically, your layout circuit has one job:  deliver as much electrical energy from the source to the motor. Unfortunately, your train motor never receives 100% of the available energy from the source since some of it is lost due to the combined electrical resistance of the wires and rails to the train motor. 

Under light or modest "loading" of the train, these variances in electrical resistance are not very noticeable. However, when your train is under higher load conditions (e.g. pulling many wagons, climbing an incline, moving around a sharp curve, etc.) it draws more current to maintain torque. This increased current flow causes a proportionally bigger voltage drop between the source and the motor and therefore less energy is available for the motor. We see the result as the train slowing down.

Therefore, the most important way to maximize your layout performance (i.e. deliver as much energy from the source to the motor) is to reduce the layout circuit resistance. This can be achieved two ways:

1) Adding more power feed locations

The most basic layout circuit often consists of a speed controller (the energy source) connected by a wire to a loop of track. If the train is located at the power feed point, it has maximum available performance since the circuit path length is at its minimum and thus electrical resistance is at its minimum.  In the extreme case of the train being at the opposite side of the track loop, the path length is at its maximum with corresponding maximum resistance. A heavy train will run faster near the power feed and slow down when its far away. 

To improve our track circuit, we can introduce another power feed point to our loop. Ideally, it should be placed on the opposite side, equidistant to the original feed point. It is important to note that this extra feed point should come from the SAME source, not another speed controller source. Connecting two different speed controllers to the same track circuit results is poor unpredictable performance and will likely damage one or both of the controllers.

With two feed points, the electrical resistance between the source and the motor is the combined "parallel" resistance of two equal half-length wire and track paths. This parallel resistance is almost always much lower than the single path. Now your heavy train will seem to run with much less variation in speed around the loop.

You can continue to add more power feed locations to your layout to smooth out performance even more, but it is a case of diminishing returns. The improvement from 1 to 2 feeds is most noticeable, from 2 to 3 less noticeable, and so on.

2) Balancing circuit path lengths

The benefit of adding more power feed locations to your layout track circuit can vary greatly depending on how you do it. To achieve maximum benefit, follow two key principles:

i) Position track feeders as equidistant from each other as much as possible. It doesn't have to be perfect, but attempt to keep the spacing as equal as possible. 

ii) Each track feeder should have its own wire from the source to the track, i.e. not shared or split from another feeder's wire. If the track feeders share wiring for some length, then impact of the wire's resistance will be noticeable since it now has to carry twice the current. It is also desirable (but not strictly necessary) for each feeder wire length to be the same. This isn't always practical in every layout, but do what you can to balance the wire path lengths.

There have been requests for guidance on "how many feeders per rail joint etc.".  However, I think the number of rail joints is not always the key parameter. You need to think of the track circuit as a whole, starting from the source, including the wires, and the potential paths along the track. The best advice is simply have TWO (or more) feeds per layout circuit spaced as evenly as possible. I would also advise placing feeders near "high load demand" locations on your layout. In particular, at sharp curves where frictional drag losses are highest. 

II. Thermistors

The thermistor fitted to almost all LEGO® brand train motors has one main job. It is a passive resistive device designed to protect the motor windings from burning out under stall conditions (easily achieved by holding the wheels stopped when powered).  When a motor is stalled, the full current from the source flows through the relatively low-resistance motor windings causing huge power dissipation and eventual burnout.

A thermistor is a passive non-linear resistive device whose electrical resistance varies proportionally with temperature. At lower "normal" temperatures, its resistance is nearly zero and current flows freely through the device. As temperature increases, specifically the increase of temperature due the "self-heating" effect of increasing current, its corresponding resistance also increases (non-linearly). At some point, the resistance increases so dramatically that it effectively acts like an open circuit, cutting off the load (or the motor in this case). It acts like a circuit breaker protecting a load from excessive current. Like a circuit breaker, it also resettable--in this case it is "self-resetting" since its resistance will return to normal when its internal temperature falls back to its normal operating range. Unlike a circuit breaker, it is a "soft" reset.  That is, it does not cutoff instantly.  It's cutoff behaviour follows a curve and is somewhat gradual. 

Thermistors come in a wide range of types, voltage/current ratings, trip/recovery times, temperature derating curves, etc. Choosing one for your particular application can be quite daunting and does require some experimentation (ask me how I know! We surveyed around 20x different devices for our Fx Bricks motor!).

I am known to remove the LEGO® 9V train motor thermistor (at my own risk!) simply because LEGO chose a thermistor with on overly safe trip point / derating curve. The thermistor starts to cutoff under modest load conditions, resulting in a very noticeable speed reduction and in some cases coming to a stop. The motor is capable of safely handling more power dissipation than the thermistor allows. I am willing to take the risk of having no stall motor protection in exchange for predictable continuous operation.  I am also depending on the short circuit/overload protection capabilities of the speed controllers that I use.  They cutoff the track supply under overload conditions and retry until it resumes to normal.

For the record, I recommend that enclosed motorized products should include thermistor protection. Its your motor's last line of defence when mechanically stalled. It doesn't take long for motor current to rise dramatically causing damaging power dissipation in the motor windings. 

III. Motor PWM

LEGO® 9V train motors are completely safe to operate with PWM based speed controllers. In general, PWM speed control is suitable for almost all DC brushed motor applications. You still need to be aware of the usual considerations of not exceeding the voltage and current rating of the motor, using a PWM frequency range which balances current ripple, efficiency and noise/vibration. 

The circumstances under which a motor is damaged with PWM speed control are very nearly the same with DC based control.  Motor "burnout" can be one of two things: 

1. Motor windings literally burned due to excessive heat / power dissipation from high sustained current. The thermistor in the motor should prevent the catastrophic power dissipation which can destroy the windings.
2. Brush contacts oxidized due to excessive sparking to the point where the oxidized deposits (black soot) form an insulating layer and stop the motor from working.

In most cases, motor burnout is a result of 2.  Often, excessive sparking on the motor brushes causes sooty deposits to accumulate on the brushes making them electrically resistive and/or isolating.  Sparking can be caused by dirty motor brushes causing intermittent/poor contact.  Sparking does occur with both DC and PWM based speed control and therefore your motor brushes are equally vulnerable to oxidization effects with either method.  Sadly, the excessive sparking itself causes more sparking since it adds more contamination to the brushes which then increases more sparking, and so on.

The effects of the sparking can be reduced with fitting a capacitor across the motor terminals to smooth out voltage spikes. Also, it is important for your speed controller (motor driver) to have flyback diodes to clamp very high voltage spikes (caused by back EMF) of the motor during PWM voltage transitions. Most modern modern PWM controllers will incorporate flyback diode protection. The Fx Bricks 9005 speed controller has a sophisticated motor driver controller with diode protection as well as a passive RC filter on the output to smooth voltage spikes on its output. 

If you're careful and brave, you can take apart the motor and clean / restore the brushes.  Clean both the brush contact "shoes" as well as the copper rotor contacts on the motor shaft with 99% isopropyl alcohol (IPA) until all of the sooty deposits are removed.  Inspect the surfaces of the brushes and rotor contacts and look for pitting or other visually apparent discontinuities. If you see pitting, use a very fine-grit (1600-4000 grit) sanding paper to burnish the surface of the brushes/contacts so that they are smooth, then re-clean the surfaces with IPA. 

Given how much I've written above, no doubt we need to put some of this material on our website!  We do provide some physical user documentation with the speed controller and the power feeder products; but it is also useful to have more in depth reference material available on our site.  One more for the to-do list!
 

[GoHabsGo]

Thanks a lot @michaelgale for these explanantions.

I hope it will kill forever the legend "do not use PWM with lego 9V motors"

and I cannot wait for using the 9005 :)

[zephyr1934]

Michael, thank you for the detailed response! I do think having this info in an FAQ or similar will be benefit to your customers. Perhaps even including a recommendation on the speed controller listing, "for larger layouts it is recommended that you use a second power drop"

 

10 hours ago, michaelgale said:

The Fx Bricks 9005 speed controller has a sophisticated motor driver controller with diode protection as well as a passive RC filter on the output to smooth voltage spikes on its output. 

Most excellent! Does that include the overlaod protection mentioned earlier in your post where the controller cuts off track supply when an overload is detected?

Lego might claim only the best is good enough, but you are going better than best

 

 

[Stereo]

I suppose you could probe your layout for good feed locations by running a battery-powered (or push) train with a multimeter connected to power pickup wheels.  You'd just short the feeder at the power supply end, instead of connecting it, then measure resistance.  A dynamometer car that just passively shows the voltage across the rails could be interesting too.  Might take a bit of running to get used to interpreting it.