Musings on a 'real' PS3 supercap mod. Update - successful bench test.

thor73 posted:

Supercapacitors for decoders are pretty common in the DCC world, to improve reliability over turnouts, intermittent track, and so on. They typically power everything in the engine - motor, processor, lights, and sound. For example, I have an ESU Loksound and 5F cap installed in my Lionel docksider. It will run for four or five seconds after power interruption, and goes over everything on my layout without a hitch, including long curved 096 switches and my #8 double crossover.

Wow!  You mean if you cut track power, the motor will continue at the commanded speed for 4-5 seconds?  I don't get it.  Is this a low-voltage (2.5V, 5V, whatever) supercap that is boosted to provide motor voltage?  

I wonder how large a supercap would have to be to haul my VL-BB and thirty PFE reefers for a second or two?  I wonder how many boxcars I'd have to dedicate to supercap storage.

Separately, there have been OGR threads bemoaning the apparent change of behavior between PS2 and PS3 when track power is suddenly interrupted/lost.  Apparently when power is suddenly lost, PS3 engines (not sure if a few, some, all) might come to an immediate and jerking halt whereas PS2 engines coast/glide to a smooth stop.

If the issue is one of demoting the effect of intermittent or momentary interruptions to track power due to dirty track, going over a turnout frog, crossover track section, whatever then here's another idea.  When you do the math (which gets annoying tedious), you might find that the best bang for the buck is to figure out a way to increase the energy stored in the flywheel.  Most engines have enough clearance on the brass flywheel to add, say, a half-dollar coin (or whatever heavy metal disc).  The amount of rotational energy (momentum) of even a small disc is remarkably huge - again when you do the math.  That is, using a supercap is of course electrical energy storage...which then gets inefficiently converted to mechanical energy via any voltage boosting circuitry, and the relatively large energy loss when converting electrical energy to mechanical energy using a DC-can motor as virtually all O-gauge engines use.  Much better to store energy is "native" form which would be in a flywheel...assuming space is available of course!

Actually, another method that would probably be more effective than the supercap is a NiMh battery array.  Since you'd be charging almost all the time, you'd actually have much more energy density to move the locomotive with the same size battery as the supercap.

stan2004 posted:

thor73 posted:

Supercapacitors for decoders are pretty common in the DCC world, to improve reliability over turnouts, intermittent track, and so on. They typically power everything in the engine - motor, processor, lights, and sound. For example, I have an ESU Loksound and 5F cap installed in my Lionel docksider. It will run for four or five seconds after power interruption, and goes over everything on my layout without a hitch, including long curved 096 switches and my #8 double crossover.

Wow!  You mean if you cut track power, the motor will continue at the commanded speed for 4-5 seconds?  I don't get it.  Is this a low-voltage (2.5V, 5V, whatever) supercap that is boosted to provide motor voltage?  

Yep, it's a 5F, 2.7V capacitor, in conjunction with a boost converter that regulates the output voltage at around 14V.

It's an off-the-shelf module from ESU. It comes with a 1F cap, but works just as well with the 5F cap that I installed on it. During operation, the cap charges to about 2.5V, and discharges to about 1.7V. There is also a setting in the decoder to allow you to specify a max keep alive time, in the event that you want it to stop within a specified time period.

gunrunnerjohn posted:

I wonder how large a supercap would have to be to haul my VL-BB and thirty PFE reefers for a second or two?  I wonder how many boxcars I'd have to dedicate to supercap storage.

Admittedly, it's a small engine, but it illustrates the principle pretty well. 

gunrunnerjohn posted:

Actually, another method that would probably be more effective than the supercap is a NiMh battery array.  Since you'd be charging almost all the time, you'd actually have much more energy density to move the locomotive with the same size battery as the supercap.

Agreed, and perhaps simpler as well, as you wouldn't need the boost converter. Any specific ones in mind?

If the numbers (size/fit, cost, engine keep-alive time, etc.) work for your particular application then so be it.  I recall the "tennis ball can" thread where the proposed cap to keep the engine operating during power loss was the size of a can of tennis balls!  

If component technology has improved to the point where it's practical, then by all means.  Though I think GRJ's idea of a rechargeable battery (vs. a capacitor) might have better cost-size-performance tradeoffs for the stated application.

The bigger issue is how to interface the battery into the electronics and to build a charging circuit that won't interfere with normal operation.  I just figured that the battery would have a better shot of providing the power density for a large locomotive to run a few seconds than any supercap we can fit in.  I think Stan's picture tells it all.

stan2004 posted:

If the numbers (size/fit, cost, engine keep-alive time, etc.) work for your particular application then so be it.  I recall the "tennis ball can" thread where the proposed cap to keep the engine operating during power loss was the size of a can of tennis balls!  

 

If component technology has improved to the point where it's practical, then by all means.  Though I think GRJ's idea of a rechargeable battery (vs. a capacitor) might have better cost-size-performance tradeoffs for the stated application.

Haha, I hadn't seen that before! 

The reality is not quite that bad though. Here's a 10F 2.7V cap that I have laying around:

Assuming an 85% conversion efficiency and operation between 1.5V and 2.5V, that's enough energy to run a motor at 14V and 0.5A for 2.5 seconds.

gunrunnerjohn posted:

The bigger issue is how to interface the battery into the electronics and to build a charging circuit that won't interfere with normal operation.  I just figured that the battery would have a better shot of providing the power density for a large locomotive to run a few seconds than any supercap we can fit in.  I think Stan's picture tells it all.

Yeah, by far the harder parts are the 'gotchas' i posted above.

Supercap, NiMH, Lipo - lots of ways to store energy. We can figure that part out.

I'm just guessing here from running different scales, different control boards, and different drives. It just got me thinking.....

I have a feeling you may be describing something that's different for O scale regular people? The keep alive circuits, would be just for the motors here. The keep alive for the control boards are already built in to the major players systems.

  (I think) there's a big problem with the type of drive being vertical motors and gears that don't allow for smooth stops when the power is cut (diesels?). In different scales with horizontal motors, the flywheels do a good job and the keep alive can do more to keep the engine going over dead areas.

 Warning! I have no practical experience here on the electronic part other than the installs I have done on different brands and types of drives.

@stan2004 The PS3 non-coasting issue really bugs me and I don't believe it was ever resolved.  MTH is pretty quiet about it;  i'm not sure it CAN be resolved.  I can't find the link now, but a long time ago I came across a video in which a fellow with serious machine skills created custom "slugged" flywheels for his Atlas SW switcher by inserting tungsten fly fishing weights in a circular pattern near the outside edge of the flywheel.  (Picture bullets in the chamber of a revolver and you'll get the idea of what he did.)  Having the additional mass near the EDGE of the flywheel makes all the difference.  However, I can only imagine the precision it would take to do this and keep everything balanced.  I've replaced stock flywheels with larger ones, and getting them on straight is a chore.  Sometimes it's necessary to improve the balance by holding a file to the flywheel in situ, while it's in operation.  Crude and approximate at best! 

@Engineer-Joe The Lionel Legacy locos with BACK-DRIVEABLE gears are a better choice IF they make the model you want, and IF you don't mind Legacy instead of DCS.  It's true that the dual flywheels found on horizontally motored locos have more mass, but especially in the smaller scales, they are probably turning a lot more RPMs.  If the manufacturers would wise up and get away from the toy train gear ratios, I think you would experience more coasting and consistent speeds, even without gimmicks like capactiors and speed control.

Another thing I should post publicly here is I am not actively looking for this. I just read complaints here on the forum.

I run mainly MTH engines and control boards in my other brands. I don't have big issues with any dead spots or power kills. My frogs are dead on my Atlas 2 rail switches and my MTH engines pass over them with ease. It's more about proper power pick-up than coasting for me.

 I am interested in the keep alive circuits that run DCC boards thru the dead spots. I always try to keep up with the latest stuff out there. I am currently happy with my MTH stuff. I just always compare what's new.

 Sometimes I think that 3 rail is over rated when I read of problems about coasting. I run with just 2 rails and I don't have these issues. I think that sometimes the center rollers (and center rail ) maybe causing extra grief for what they're worth? What else is it?

Ted S posted:

@stan2004 I came across a video in which a fellow with serious machine skills created custom "slugged" flywheels for his Atlas SW switcher by inserting tungsten fly fishing weights in a circular pattern near the outside edge of the flywheel.  (Picture bullets in the chamber of a revolver and you'll get the idea of what he did.)  Having the additional mass near the EDGE of the flywheel makes all the difference.  However, I can only imagine the precision it would take to do this and keep everything balanced.  I've replaced stock flywheels with larger ones, and getting them on straight is a chore.  Sometimes it's necessary to improve the balance by holding a file to the flywheel in situ, while it's in operation.  Crude and approximate at best! 

Never heard that one!  But as tungsten is over twice as heavy as brass, I can see why he did it.   I was just looking on eBay and was amused to see hundreds of listings for pure tungsten rod - starting at just a few bucks.   Never worked with the stuff myself.

in these videos you'll see the stock MTH engine go over dead frogs on Atlas #7 2 rail switches for the mainline crossover. You'll also see that he had stopped for adding smoke oil and the cars are still coming out of the passing siding with Atlas #5 switch that also has a dead frog. One car blinks and I'll have to see if a truck is not doing it's share on power?

Resurrecting this thread a bit, I've arrived now at a homebrew keepalive solution that I'm happy with for my DCC engines. It's pretty simple to build, and provides 2-3 seconds runtime in a compact package.

Details here.

Now, to figure out a way to tie it into a PS3 engine...

Update 2/3/20:

Over the weekend I started the upgrade on my MTH Great Northern S-2 from PS3 to a Loksound L v5. This is a beautiful loco, and has been on my list for this for a while. As a result, I now have a set of PS3 electronics to play with, so naturally I was curious to follow up on this idea.

I was careful when removing the original electronics to retain everything in a usable configuration for bench testing. I made a modified harness to connect the tender board and the boiler board without needing the wireless drawbar, etc. For simplicity, I also eliminated the 3/2 switch and associated wiring, so the board is just powered directly via it's PCB inputs. I left the motor and tach wiring intact on the chassis, so that I could confirm the actual motor operation as part of the test.

Initial poking around and powered tests with a scope and benchtop power supply proved promising, so I built up a modified version of my O scale keepalive and connected it to the main PS3 board. Power up, let things charge a bit, and...

It works. After letting the cap charge for a few minutes, I pulled the plug for several seconds and then reconnected with no impact on the loco operation. It turns out the PS3 electronics maintains normal operation for almost ten seconds before stopping the loco, given sufficient charge. The actual runtime pulling a train will be less, of course, as the motor is drawing very little current in this test configuration. But with a big enough capacitor, you could achieve run times up to that limit if you wanted. The keepalive is connected to the PS3 board with just two wires, so installation is pretty easy and non-invasive. I'll write up details on the keepalive and hookup to the decoder when I get a chance.

Here's a video of the test setup. The scope is showing the DCC input. The yellow voltmeter shows the capacitor voltage, which you can see charge and discharge when it the track is plugged and unplugged. The orange voltmeter shows the PS3 decoder bus voltage, which runs about 12.3V while powered, and about 11V while operating off the keepalive. The voltage considerations for the boost regulator described in my keepalive post apply here as well, which is why the keepalive voltage is slightly lower than the powered voltage.

Some caveats worth noting:

• I run DCC, so that's what I tested with. I expect DCS would behave similarly, as the protocols are similar, but that is just a guess.
• I tested with conventional A/C as well, and it did not work. The motor stopped when I removed power, despite the keepalive maintaining the decoder bus voltage at 11V. I suspect the processor interprets the loss of track power as a stop/reverse command and acts accordingly.
• Bench test only of course, and only on one PS3 board.

I wonder what it would do with AC if you were in DCS command mode?  A brief removal of track power is not considered a reverse command in that case.

gunrunnerjohn posted:

I wonder what it would do with AC if you were in DCS command mode?  A brief removal of track power is not considered a reverse command in that case.

Interesting idea. I jumpered the DCC/DCS pins and tested it, but it behaved the same, that is, drive stopped immediately after removing power.

Is there anything beyond that needed to put in DCS mode?

I just glanced thru your post of keep alive and couldn't find exactly what was your modified version used for ps3.

I'd be interested if it could help my engines thru my largest Atlas switches with dead frogs. They are starting to give me issues. I think some stock engines are losing their pick ups on certain points.

I like the idea too and am interested in the results and instructions on how an end user, with just enough knowledge to be dangerous, could install.

GRJ, your battery idea does sound simple, but what would happen when you hit the e-stop button?  I have visions of the trains continuing on their path of doom.

Ron

thor73 posted:

gunrunnerjohn posted:

I wonder what it would do with AC if you were in DCS command mode?  A brief removal of track power is not considered a reverse command in that case.

Interesting idea. I jumpered the DCC/DCS pins and tested it, but it behaved the same, that is, drive stopped immediately after removing power.

Is there anything beyond that needed to put in DCS mode?

Obviously you have to have a DCS command system with the signal on the track to be in DCS mode.

Ron045 posted:

GRJ, your battery idea does sound simple, but what would happen when you hit the e-stop button?  I have visions of the trains continuing on their path of doom.

Well, they're not going to continue very far, so I don't' think it's a major concern.

I have been building LUCAS FLUX CAPS for may PS 1 & 2 engines for some years now.

MONK

Engineer-Joe posted:

I just glanced thru your post of keep alive and couldn't find exactly what was your modified version used for ps3.

I'd be interested if it could help my engines thru my largest Atlas switches with dead frogs. They are starting to give me issues. I think some stock engines are losing their pick ups on certain points.

Hey Joe, nah you didn't miss anything. Just got busy at work here and got sidetracked.

It's pretty simple, as it turns out. I added a step down converter to my O sized keepalive, since there was no readily accessible 5V supply on the PS3 board, and then just tied the keepalive into the output side of the giant bridge rectifier. So it makes for an easy, minimally invasive two wire install.

I'll post an update in that thread when I get a chance.

I'll be interested to see how this actually works.

Stop me, as soon as I go off track here.

1)Our main goal is keeping power to the DCS board, over pesky frogs, and unnoticed dirty track sections. 

2) The center rail is the weak point.  There are lots of wheel contacts, but few center rail pickups. Also, the center rail has inherent gaps at switches and crossovers. 

Wouldn't the easiest solution be center rail pickups on the front car and a single thin wire to the loco? Almost everything is already there, if you’re running passenger cars. You just have to pick a favorite front car or 3, for modification.

I strung my polar express cars together this way. Not to solve this problem, but because it allowed me to get rid of most of the electrical pickups on the passenger cars. I also added weight to the loco. I can now pull the 7-9 cars I like to pull. 

The center pickups aren’t hard to attach to a number of trucks. I find the wiper style outer rail pickups to be a bit more difficult as you might have to tap the truck. 

That thin wire might not be the prettiest, but I can’t think of anything that would be less invasive, or cheap. I  think the battery or Cap setup would have to go in another car anyway.    Wouldn’t both wreak havoc on DCS signal, if they were in the signal path? Also with a battery or Cap you would need fully powered trucks 

While I've put rollers on plenty of small steamers, and even some larger ones, I don't think this is super realistic to tie the locomotive to a set of specific trucks, at least for me.

As for the cap causing issue with DCS, the answer is no.  Right after that bridge rectifier, there is a giant electrolytic capacitor anyway, the DCS signal is tapped off before that point.

I don't understand the "fully powered trucks" comment.  We're talking about a locomotive here and tieing into the electronics.  So, we already have drive motors.

gunrunnerjohn posted:

While I've put rollers on plenty of small steamers, and even some larger ones, I don't think this is super realistic to tie the locomotive to a set of specific trucks, at least for me.

As for the cap causing issue with DCS, the answer is no.  Right after that bridge rectifier, there is a giant electrolytic capacitor anyway, the DCS signal is tapped off before that point.

I don't understand the "fully powered trucks" comment.  We're talking about a locomotive here and tieing into the electronics.  So, we already have drive motors.

Fully powered was a poor word choice.  I meant you could simply add a center pickup to the car(s)you modify, instead of also needing trucks that picked up the outer rail as well. In any case, wouldn’t a sufficiently large battery, or cap force you to modify specific cars? Few locos would have the required room. Especially if we are talking about smaller locos (Less pickups) that are more prone to losing voltage. Once in a separate car, you’ll need a tether wire.  anyway

If this is happening to folks on every PS 3 they own, it’s time to go Legacy 😜. If it’s happening to one or two in someone’s collection, I’d make the argument that picking a few power cars for those units isn’t to constricting. Pick a popular rolling stock base and it wouldn’t be too hard to shell swap.  

Supercaps are pretty small, I can fit a lot of them in, even in a small locomotive.   I would VASTLY prefer a solution that's self-contained in the locomotive than one that requires me to tether to specific cars.

And you think Legacy doesn't have the issue?  You don't get around much.

I run my MTH locomotives in DCC as well and have already had to rewire my layout because of "dead" spots.  Having a motor keep alive would be great for them.  Can't wait to see more info on this.

gunrunnerjohn posted:

Supercaps are pretty small, I can fit a lot of them in, even in a small locomotive.   I would VASTLY prefer a solution that's self-contained in the locomotive than one that requires me to tether to specific cars.

Yes you can. I lack your extensive skill set, but can tap into the hot wire, so if my soon to be painted Navy livery, switcher has switch issues, I’ll try my idea first. 

Okiedoke, details are in an update of my O-sized Keepalive post. Enjoy. 

Did you update the first post or add detail later?  If you updated the first post, it would be smart to perhaps put a caption at the top of the post...

**** Updated 02/17/2020 ****

Thor I'm really impressed at all of the thought you put into this.  In order to demonstrate the benefits and show people why this is an important breakthrough, I suggest a comparative series of videos:

Stock MTH Protosounds 1 large steam loco, running in conventional.  45 scale mph track speed with a typical train on tangent track.  How far does it coast into a "dead block?"  (Obviously locos with pick-up roller(s) on the tender will proceed further into the block.  You'll have to measure how far the rear-most roller is from the front of the loco to calcuate the actual coasting distance.)  My guess: minimal coasting, abrupt stop.  Fail.

Stock PS2 steam loco, running in conventional.  45 scale mph track speed with a typical train on tangent track.  How far does it coast into a "dead block?"  My guess: about a foot of coasting.  Good.  Since there were no mechanical changes, this showcases what thoughtfully-designed electronics can do.

Stock PS3 steam loco, same conditions as above.  Coasting distance?  My guess: minimal coasting, abrupt stop.  Fail.  A step backwards from PS2 in this important regard.

PS3 steam loco with keep-alive added, same conditions as above.  Coasting distance?

Note: All MTH steam locos have self-locking gears, so momentum must be maintained through capacitors and/or flywheel action.  In my experience, low RPMs limit energy stored in the flywheel, so the electronics have to do most of the work.

Legacy steam loco without back-drivable gears (examples: Atlantics, 0-8-0 switchers, B6 switchers.)  Same conditions as above.    My guess: 6"-12" of coasting.  Acceptable.  The conservative gearing of these locos and resulting higher RPM, stored energy in the flywheel, and perhaps less magnetic drag from the small motor keep things within acceptable bounds.

Legacy steam loco WITH back-drivable gears (examples: any "premium" large loco; Ten-wheeler; H10 Consolidation.)  Same conditions as above.  My guess: perhaps 18" of coasting.  Very good.

I think you stated above that you don't own a TIU (i.e., DCS command base), and I'm going to guess that you don't own a Legacy comand base either.  But it would be instructive to repeat these tests in the respective command environments.  In this case, instead of cutting track power, the test might consist of pushing the DIR control on the hand-held remote and making a note of the loco's behavior.

Did I guess correctly?  C'mon folks, post up some videos and show why Thor's circuit is a long-awaited boon to PS3 owners everywhere!

Problem with pushing the direction button on the remote is that it's a function built into the electronics, so they will behave as they are programmed to do, which on MTH PS-3 running in DCC is to slow down and then speed up in the other direction.  Legacy will almost come to a complete stop with very minimal coasting.  So to run your test you truly need to cut track power.