Stay-alive capacitor hacking

[Brianetta]

By GL I meant "Good Luck (with your decoder mod project" . Just been around the net so long all those as standard as a smiley acronyms are so familiar deployed in common use one considers them universally understood rather than elitist pretension.

I didn't pick up on that (oops). I've been on the net for a long time; in fact, since before smileys became standard. I was a very early adopter of the Betamax® smiley (it's the other way around). I was upset when the rest of the world finally standardised on the normal ones, but I didn't change my ways. (-:

In case it came across differently, no aspersion was intended...

Sorry, that was my fault. I wasn't trying to be defensive, and didn't mean it to come across that way. We've probably both spent a similar amount of time studying our insulated points, realising in the same way that the design is sound and should actually work - as long as all the wheels of a loco come down and hit something. I have one set of two points facing away from each other which is a particular problem; the track top height on one of the rails seems to be a few microns lower than the one it's fishplated to, and the length of the loco so unfortunate, that the loco ends up with two diagonally opposite wheels sat on insulated crossings, and both other wheels on one side (and only one side) on the track. I can rock the chassis over that diagonal axis very, very slightly. Applying torque to the whole loco gets the flanges playing the game, and the train moves off.

The problem would be fixed completely if the wheels could move up and down at all. They can move left and right, and they can rotate on their axle, but there's no vertical compensation at all. They're as rigid as can be. I'm tempted to try a modification of the chassis such that the front wheels (which are the least loaded; it can sit pretty on the back four) can lift or drop by a millimetre or so, which should also fix the problem. That's if my DCC capacitor hack fails, though!

[Bufferstop]

I have one set of two points facing away from each other which is a particular problem; the track top height on one of the rails seems to be a few microns lower than the one it's fishplated to, and the length of the loco so unfortunate, that the loco ends up with two diagonally opposite wheels sat on insulated crossings, and both other wheels on one side (and only one side) on the track. I can rock the chassis over that diagonal axis very, very slightly.

I'd try to fix this problem first. In the long run it will save you having to hack numerous expensive chips.

[Brianetta]

I'd try to fix this problem first. In the long run it will save you having to hack numerous expensive chips.

It's not fixable - I can barely detect the discrepancy with the tip of my fingernail. It's almost certainly within manufacturing tolerance. Any modifications I make to the track will likely end up with dents or kinks in the running surface which will be more exaggerated than the current alignment issue.

Any future locomotives won't be quite so... rubbish, and I definitely won't be hacking numerous expensive chips. Besides, you want to see this succeed, right? Even if I have to post photos of an exploded locomotive, it's all entertainment.

[poliss]

There's an Aussie chappie who has fitted 'stay alive capacitors' to his sound decoders using a 4,700 uF 16 Volt capacitor. See how he did it here. http://www.members.optusnet.com.au/mainnorth/alive.htm

[Bufferstop]

OK! seems you have a value to work with now and if you want to do it for it's own sake I'll not argue agin it. I'd definitely try to introduce some vertical movement into that locos axles though. Just enough for the non driven ones to drop a little. Just enough that if you hold it in the air and put your finger under one of the wheels you can just about feel it lift. Out if interest what's the distance over which both wheels are sitting on plastic.

Best of luck and keep us posted.

[Brianetta]

There's an Aussie chappie who has fitted 'stay alive capacitors' to his sound decoders using a 4,700 uF 16 Volt capacitor. See how he did it here. http://www.members.optusnet.com.au/mainnorth/alive.htm

Yes; he's been one of my biggest sources of inspiration. However, his decoder gave him 12V, not 27V.

Bufferstop, the layout's upstairs. I'll measure it later. Basically, if you put two identical sectional track points back to back (straight to straight, curves in opposite directions) then you have the configuration. My loco's chassis is exactly that long, and will sit on both insulated frogs at once.

There's a video coming up; you should all find it interesting. You get to see whether I'm on the right track here (heh heh) or whether I manage to set fire to my decoder. How? Well, I'll be attaching my Select's aux output to those two wires I added...

[Brianetta]

Video on Youtube now! Click to watch.

Results of Roger's suggestion were not useful. Adding a 10k resistor reduced the voltage to 26V.

Interestingly, adding more locomotives to the track reduced the voltage. When there were two others, it was down to 18V.

[Roger (RJ)]

Brilliant video. You're obviously braver than me to stick 15 volts up the chip.

As the 10k resistor made so little difference (it's only pulling about 2.7 milliamps) I would try a smaller value resistor. Using a smaller resistor will draw more current and should reduce the voltage further. I would aim for about 10ma and this would (in theory) need a 2.7K resistor, which is a standard resistor value. If you don't have a 2.7k resistor, you could put 4 10K resistors in parallel, which would give 2.5K. This would give slightly more than the suggested 10ma.

I don't understand why the voltage drops with more locos on the track but maybe it's something to do with the Select's DCC output, which is known to have a very spikey waveform (one of the main reasons why it's not NMRA approved) and could be the reason why you are seeing 27V

PS
I've got the mixed goods set and one set of wheels on each loco is sprung, front set on the 08 and rear set on jinty. Supposed to help with pickup.

[Brianetta]

Brilliant video. You're obviously braver than me to stick 15 volts up the chip.

Well, nothing ventured, no entertainment to be had. Life's too short to be over-precious about something like a low grade DCC fitted shunter. A little harmless risk really does up the fun levels, and nobody's life ends if my loco decoder fries.

IT'S ALL IN THE NAME OF SCIENCE!

As the 10k resistor made so little difference (it's only pulling about 2.7 milliamps) I would try a smaller value resistor. Using a smaller resistor will draw more current and should reduce the voltage further. I would aim for about 10ma and this would (in theory) need a 2.7K resistor, which is a standard resistor value. If you don't have a 2.7k resistor, you could put 4 10K resistors in parallel, which would give 2.5K. This would give slightly more than the suggested 10ma.

I did try my other resistor value, which was 1K. It dropped a couple more volts.

I don't understand why the voltage drops with more locos on the track but maybe it's something to do with the Select's DCC output, which is known to have a very spikey waveform (one of the main reasons why it's not NMRA approved) and could be the reason why you are seeing 27V

Yeah, I'm beginning to frown in the Select's direction. Still, if I have to over-specify everything for its sake, my locos can run on any controller without going bang!

PS
I've got the mixed goods set and one set of wheels on each loco is sprung, front set on the 08 and rear set on jinty. Supposed to help with pickup.

Mine aren't sprung. You wouldn't have the ones with traction tyres, would you? With only four pickup wheels those would really need a compensated wheel. Mine don't have any movement other than laterally; the axles can slide along their length a little.

[Brianetta]

OK, good news. I've managed to find a workable capacitor solution from Rapid Electronics. I wanted as much capacity as I could get, rated at 35V. The 4700µF capacitor was too big for the inside of the Class 08 body shell. The usable void within the body of the shunter is 30x25x20mm, and that cap was 36x18. Then I spotted the 2200µF capacitor; at 26x16 it would seem that I can get one in there with room to spare! So, that can soak up all 27 volts of electrical goodness through a 100R drinking straw until they're all full up. Then they can drive the locomotive through a diode (shorting out that resistor). Knowing that the decoder normally operates at 27V when alone, and seeing that it can operate at 15V without any problems, it should run smoothly until the capacitors' voltage drops well below about half full. That's encouraging, and I'm looking forward to being able to shunt at prototypical speeds with my proto-laughable shunter on my proto-nonsensical track geometry.

I'm really chuffed, and looking forward to my order being delivered. Yay! (-:

More pics (and if you're good, a video) to follow. May many follow in my footsteps, now I've shown people how non-scary it is.

[Martin71]

Hi

Outstanding post and well done on taking the first steps. I think you have provided an excellent trial and error topic on the "stay alive" caps. I hope you dont mind a little bit of advice on this subject.

Can I suggest you go for the biggest caps possible that you can fit in. 2000uF is a very good start but I would stick with 25v as I have tested the selects output on a oscilloscope and I have seen it spike to 56v. Now in fairness that test was conducted nearly 2 years ago and I believe the select has being updated. As to the dcc waveform...what waveform it was a total mess on the scope.

I use 3 x 470uF in parallel @16v. I can get away with this because I have an NCE with an output of 13.8v or 12.8v on the DC side of the rectifier. I am still cutting it a little fine.
You are on the right track with the 100ohm resister in parallel with the 1N4007 diode on the positive leg. I sometimes use 300ohms on certain locos that have good pick ups so I can stagger the charge over the fleet. see photo installed in a Bachmann 9f Evening Star



For those that are wondering the black box is a composite housing for a speaker and the other wires are for the tender light.

Now I am lucky as you can see form the photo that I have space in the tender for the caps, my biggest worry is soldering directly to a loksound diode rectifier which can be a little tricky.

Their is another huge benefit to the caps and I would be very interested to hear if you actually see it. You mentioned why you wanted the caps and what you were doing with the loco (shunting) so once you have added the 2000uF. Lets see what happens to your shunting speed at step 1 after the install? Can I suggest before install measure out a small section of track, time how long the loco takes to get from point A to B on step 1/128 and then do the same with the cap installed. I am hoping you are in for a surprise.

ON most of my locos the motion is silky smooth, bye bye DC ripples.

I think you also have a great style of writing in the sense that it is not technical. I work in the electrical game and I find your style refreshing and I think most will. Keep it up your nearly there. Once you get better built decoder you may want to look at adding a choke on the neg leg ( only if you have programming problems) and a 3.3k resistor in parallel to drain the decoders memory.

Best of luck.

Martin.

[Brianetta]

Thanks, Martin! I even heard good things about my video in a coffee shop today, which was great.

I'm choosing 2000µF because it fits, physically. 4700µF doesn't fit, and won't unless I want it to stick out of the shunter's case somewhere. I've ordered two of the 2000µF caps, because I suspect I might just be able to get another one into the case above the motor. That'll double the capacity, which is no bad thing.

I'm not keen on sticking to 25V because the rectifier puts out a very steady 27V across its outputs when it's alone on the Select. That is, I'm led to believe, a way to add excitement to any video, but won't do my loco any favours. I'll need a 35V rated capacitor to remain within the bounds of safety. The downside of your 16V capacitors is that your loco couldn't safely visit a layout that's using a Select.

Thanks for the feedback on my style. I'm learning as I go here. Part of my day job as an IT technician is explaining what I'm doing to people who don't have day to day experience in my field. I find that a more conversational style works there, as it appears to work here.

If I get a better decoder, I'm unlikely to be putting it into that shunter. It only has glazed windows because I glazed them, and it has steam engine wheels. No, I'm hacking this one to better it. Trial by fire; if it lives, it'll be brilliant. I have to ask, though... what's a choke to a layman?

I love your pics, and I'm glad I'm not the only one actively mucking about with my trains' brains.

[Bufferstop]

Now i've seen your video (had to wait for the old broadband to climb back to a decent speed) I think I know where the 27v comes from. The track voltage is 15v AC with a control signal superimposed on top. Your meter expects a constant voltage on DC not what's present at the points you are measuring. When a voltage is called 15V AC that is the RMS value not the peak. On AC ranges the meter reads the peaks but is calibrated to show the RMS. If you rectify it and measure the DC the meter will respond to the peaks. So don't loose too much sleep over the 27V

If you've not come across RMS before its a way of getting a meaningful average for a voltage that goes from say +18v to 0 to -18v fifty times a second. A simple average gives you 0. By squaring the values then taking the average of the square root you get a representative value. That's what's shown on the meter scale.

[Brianetta]

Bufferstop, I'm reading the voltage on the output of the rectifier. It's DC there, and the meter is set to the DC 50V range, as photographed. Not only that, but it's a rectified square wave, which means it has an RMS that's (theoretically) identical to its peak. I can, and have, measured the output of other rectifiers, and they show about 14-15V, which is completely expected. This decoder gives 27V, unless other decoders are present on the bus, in which case it drops a little.

Having seen the video, you can know that I'm not measuring the AC by accident, since I used those same points to power the decoder with DC at 15. If I was on the wrong side of the rectifier, the decoder would have remained powered up (at nearly 15V) but the DCC sampler (which samples the unrectified DCC directly) would have seen a constant 15V, switched to DC mode and run off at full speed. It didn't, meaning that the decoder remained powered up but read no voltage at all at the DCC sampler. This kept it in DCC mode, and it continued to operate with the last instruction received.

So, I'm going to have to assume that the 27V is not an illusion caused by a misunderstanding of both of my meters. That said, once I receive my capacitors and attach them to the decoder, I'll measure the voltage across their terminals as they charge up, and will share the result here. Since I intend to be close enough to hold the probes, I am more than happy to have a well over-specified capacitor.

[Bufferstop]

At a guess somewhere beyond the rectifier there's already a capacitor across the supply. If there is spiking as seen by martin71, a capacitor is going to convert it into a higher steady voltage. It'll be Interesting to see what happens to that voltage when you put on the bigger caps. I think a small increase in capacitance will see it rise, but with a big increase it will flatten off and may even drop slightly.

PS Choke = inductor. Old valve man's terminology.