???

Looks that yo've made a simple mistake somewhere, doesn't it? It should work, shouldn't it? Why can't circuitry just behave?

A note: AFAIK flickering getw wors when there are more messages flowing on track; nothing to do with the size, booster or whatever, but simply the number of addresses locos.

/Lars

Makes sence.

I've read about the super capacitators, "Golden cap", about 1 F or something. Hav you considered?

/Lars

Check the still unavailable Viessmann 5063 at http://www.viessmann-mod...sprache_nr=2&gr_nr=2

It would be interesting to check the circuit schematics of this thing.

Luís

There are several sources for flickering lights on a digital layout:

1. Flickering due to asymmetry of the digital signal.

This kind of flickering is caused by the fact that different digital messages have different ratio between the positive and negative voltage.This kind of flickering only occurs when half-wave rectifiers are used. Typical example is when you use the loco/car body as return for a lamp controlled via a decoder, or when you use LED lights without a full-wave rectifier. This kind of flicker does not occur in coaches with normal light bulbs without a function decoder. The full wave rectifier + small capacitor solution will cure this kind of flicker .

2. Flickering due to bad contact.

If you can't improve the contact, only a large capacitor or a battery will help. If you use lamp sockets, the problem could be bad contact in the lamp sockets, and then an ever so big capacitor will not help you!

3. Flickering due to weak power feed

With weak power feed, the variation in current draw caused by varying load on the loco will cause a fluctuation in the track voltage. The problem can be either the wiring or the voltage stability of the control unit/booster. Cured by better wiring and/or stronger (possibly regulated) power supply/booster.

4. Flickering due to intermittent shorts

If there is something causing intermittent shorts that are short enough to not trigger the over-current protection, this can also cause flickering. There are many possible sources for this kind of problem.

This requires access to both "red" and "brown" in each car. So with your "design rules" it is not possible to have the half of a full-wave rectifying circuit in each car.


With a single diode, you will get a half-wave rectified signal of 20 V. This can be evened out to a 20 V DC with a large capacitor. To get 40 V you will need a more complicated circuit. It is possible using only diodes and capacitors, but these solutions are usually not capable of suppying a lot of current.

According to so far known information about the mfx feedback, this is not the case. The mfx feedback is superimposed on a constant +20V or -20V signal. See http://www.mue473.de/mfxrahmen.htm. This is different from the DCC feedback principle, where the track voltage is turned off during feedback.


No, with a single diode you will either have +22V or -22V, not both.

Hi all,

Thanks for all the helpful feedback. I think I am one step closer now... Examining the track signals with a oscilloscope reveals that the output of my Delta booster indeed has a significant level of LF fluctuation.

Let's look at the pics.... Please excuse the poor quality. Seeing a picture on the oscilloscope screen is one thing, picturing it is another (if anyone has a nice of these nice, new digital Tektronics storage scopes with USB left he does not use - you have a buyer;-)

The picture below shows the output signal. Two digital signals can be seen, a relatively slow one (the old Motorola protocol, I assume) and a higher frequency one (possibly the MFX RDS signal - I did not check that really). The signals superimpose.

Note the considerable level of fluctuation. The scope uses a 1:10 probe, so one division is 20 Volts. So the fluctuation is close to 5 volts (25%).



Another picture, again showing fluctuations.


The next pic suggest it might be an DC issue - the booster has problems staying at the correct level after a fast pulse train came by.



And, as comparison, the output of a unloaded MS. A quite stable signal, no fluctuations.



However, I do not yet know what causes it. Is it the Delta design (when used as a booster). Though using Delta as a booster was 'officially' advertized by M* itself, that was before the MFX protocol was introduced. Or is it a result of the load? (There are quite some consumers on the tracks - trains, coaches, braking circuits). (And especially the latter are suspect - the capacitors there may load the booster too much).

I can do two things:

- Find the cause of the fluctuations. If it is the load or my wiring, I might repair it, if it is a design issue of Delta and/or the protocol when loaded, bad luck.

- I might build voltage regulators in my coaches after all. Because everything else - the train motors, the braking, works perfectly fine.

Any comments welcome.

/rudi

Hans Martin,

I'll play "Devil's Advocate" here.
Are you absolutely certain the "missing edges" are real? They could be caused by too low bandwidth in acquisition, or the limited resolution of the display. Unlikely, I know, but worth considering.

As regards regulators, that would seem the obvious solution -- actually a bridge rectifier (4 x diodes, large value electro cap) followed by a linear regulator. Good linear regulators are available in SOT-23 and SC-70 packages, along with good SMD high-value electrolytic capacitors. I know that working with SMD devices manually is a royal PITA, but it can be done. Also, you can get "prototyping" PCB boards with a selection of land patterns you can cut out.

The main thing to check is that the maximum input voltage is >24V or so. Vin max = 28V is becomming common, and automotive applications usually specify 42V. Also check the power dissipation in the regulator for the expected bulb load. In this instance, LED's would appear better.

Hi friends,

Hmm, modifying a Delta 4F to full wave rectification is not quite easy: you'll need two transformers for each booster (or one transformer with two 16 V output windings). Remember the one-way-rectifier of the Delta 4F produces +18 V and -18 V from one 16 V AC input, impossible with a full wave rectifier.

I suppose MS/CS are better in this respect, as they have full wave rectification ? As a consequence I would expect that light flickering is reduced.

Damon, I made some further tests. This time I have also displayed the current flowing from the CU to the loco on the 2nd oscilloscope channel. As you can see: current peaks and missing voltage edges coincide perfectly, and you can nicely see the coach lights flicker with the motor humming sound. Acquisition bandwidth should be o.k. (100 MHz), display resolution is lower, of course (~20 kHz).




If you examine the osciloscope plots very carefully, you also find a superimposed 50-Hz-fluctuation of the voltage. As was discussed above, this come from the one-way-rectifier of the CU and makes the flickering problem even worse.

However, I think a voltage regulator is not a bad idea. And it reduces the brightness of the coach lighting, which is usually too bright for my taste.

Hans Martin

Hi,

I think Jorge gives the good example. After giving opinion/ ideas on the question, we could instead give our own solution for avoiding coach light flickering.
Rudi,
maybe you should open another topic titled "Who solved the coach light flickering problem using incandescent lamps, and how?"

Here is one answer:
Track:C
Max lenght between 2 feeders: 2m
CU: 1x 6021+home made booster
Number of locos simultaneously: 5 (but most of time 1 or 2 are stopped at signals.)
lights: Standard Marklin bulb lighting, directly fed by the track voltage
4 coaches; 1 pick up shoe for the whole coach consist.
coupling: mix of Marklin fix current conducting stick and Viessman cc coupling.
Ground: 1 Marklin boogie contact per coach.
Result: No noticeable flickering.

I really think that the booster is the key component.

Have fun!

Fred

Thanks Hans, for additional testing, and Fred, Jorg, and others, for commenting. Sufficient track feeds are important and might yet be another reason why flicker occurs, but I am not yet sure this is the very source in my case.

To understand the origin of the LF fluctuation, I tried to do some more systematic testing this weekend. To this end, I built a simple test circuit: a circular track with one loco, a power controls (either an Delta 4f, an old Delta, or an MS), and different light bulb configuration (connected directly, via rectifiers, or via voltage regulators). This way I removed any uncertainties possible related to my layout, the use of Delta’s as a booster etc, these is the simplest layout one can build.

I did 3 tests for each configuration:

1. Light bulb directly connected to the tracks

2. Light bulb connected to one-way rectifier (1n5419 Schottkey diode plus 100 uf capacitor). Two variants:
2a. Current flows from slider to wheels (i.e. loading the positive flank)
2b. Current flows from wheels to slider (i.e. loading the negative part of the signal, which is on average present more often than the positive one)

3. Light bulb connected to one-way rectifier (1n5419 Schottkey diode plus 100 uf capacitor), and then regulated by a 7812 voltage stabilizer. (I changed the bulb voltage here to adapt to the lower output voltage).

For the lights, I used a chain of 3-Volt tube lights connected in series. I changed the number of lights in the chain to deal with lower voltages (e.g. for the 7812 test). The series of lights draws approx. 50 mA. All loco tests were done with an 34060 Delta locomotive.

I did not try full rectification, as my layout has a number of digital braking circuits (which are by themselves ingle rectifier-based) and I want it to work well over there too. Because flickering issues attract more attention in these stopped trains than when they are driving. Also, earlier test showed that full rectifiers do not solve the flickering per se.

Test A: Feed by a Delta 4f (66045), powered by an ‘old blue’ 40VA transformer.

In configuration (1), with no trains, the light shines fine. When the loco is started, the light starts to flicker. Overall light level remains the same, however.

In configuration (2), there is constant flickering. Even extra bulbs that are directly connected to the rails, flicker. In (2b), the flickering is somewhat worse than in (2a). Too bad, because 2b is the situation that will occur in digital braking sections.

In configuration (3), the light is always free of flicker, regardless whether trains drive or not.

Test B: Feed by an old Delta (6604), powered by an ‘old blue’ 40VA transformer.

Basically, the same as test A. The overall flickering seemed a bit less than in A, but occurred in the same situations.

Test C: Feed by an 1.2 amp MS, , powered by an ‘ugly black box’ (the 18VA transformer that accompanies it in the start sets).

In configuration (1), no flickering is seen. Driving the train does not introduce flicker, but does lower the overall light intensity quite viewable. Possibly, this is due to a voltage drop when this little transformer is loaded.

In configuration (2), the lights flicker, independent of whether a loco is driving or not. In the latter case, the intensity lowers but the same level of flicker remains. I did see little difference in (2a) and (2b).

Conclusions

- None of feeder tests (Delta 4f, Delta or MS) are entirely without flicker in all three configurations.
- Both MS and Delta units start to flicker when faced with asymmetrical loads. This is bad, because the braking circuits in my track by definition create such loads.
- In addition, the Delta units flicker as well when train loads are added.

Hans’s observations that these fluctuations are related to power draw are quite plausible – I’ll try to confirm, though I do not have a memory scope (though I do have an current probe for my oscilloscope).

I do not now how other CU’s or the CS behaves. I don’t have these (might once try them….). Possibly they have better circuit designs and deliver better results. In any case, my measurements are done with simple means (some bulbs, a train, a diode, a cap) so many of you could repeat them, if you like to, an a CS for instance.

For the time being, I see the addition of voltage regulators to all coaches as the only. After measuring the (true RMS) AC voltage on the tracks in my layouts in different situations, as well as the resulting DC after rectification, I decided I will settle on a 7812 regulator – with this one, there is always the necessary 3 volts drop for the regulator, even in the worst case. With a power draw of 50mA, it works fine without cooling, also when the voltage is higher. Since I use a series of tube lights, I can vary the numbers of lights (or make to chains) to adapt to other voltage levels.

I tested this regulator connected to different parts of my large layout as well (also in braking sections and under different system loads), it seems fine. I will soon try how it really looks when installed in a car.

Sorry to be so long again…..
/rudi

PS. I kind of decided not to use LED’s for lighting. I did several projects with them, one where a coach has its own microcontroller, which controls the nine LEDs individually (each is above one seating place). I reported it in this threat : https://www.marklin-user...archTerms=ultimate,light (sorry moy browser here does not allow to include real URL's)

Though this car works great, it is far to time-consuming to make for all my cars, and by design (draws tool much power to be used in large quantities (current flow by each LED). Finally, I experimented with several different makes of white LED’s (some with ‘yellowish’ diffusers) but I do not like their light and light coverage pattern as much as I do with tube lights, especially when you put – say – seven of these evenly over the length of the train. And series of tub lights can be easily soldered together and then installed with just a few bits of Scotch tape to the inside of the roof (or better: use black music stage tape). This is so lean that you cannot see them – even if you look in from below. I have one 37265 M* ET-87 train with integrated yellow LED’s and they are – well what can I say – way to yellow. Unnatural really.

PS2. Larger capacitors did not change much in the test. I tried up to 1000 uF, and there was little difference (well, some, but not close to elimination of flicker at all). Larger values than 1000 uF seem unfeasible to me, as I they need to be 25 Volts model at the least and these ones are simply too much to put invisible in a car.


this threatthis threatthis threat[/url][url]

Hans,
I did some tests with that. When I loaded a 7812 (i.e. one amp version) with one chain of lights (50 mA) it did not warm up at all. When I added four light chains, something like 200 mA, and run it for an hour or so, I felt it warming up. Not alarming, but still quite a bit. Not surprising, as the stabilizer has to dissipate the voltage drop over it, times the drawn current. Of course you are right that there are higher current versions (78S12 etc), but also they have to dissipate the same energy in one way or the other.

I want to be on the safe side, that is, if track voltages are higher than in my current test situation, I do not want burning stabilizers. And I do not have place for cooling bodies.

But it is no problem to have one 7805 per car. They are very cheap, and easily can be taped to the coach roof without being seen. I will do a single rectifier in the first car (i.e. the one with the slider) and bring the (unstabilized) DC (wrt the wheel ground) by single pole conductors from car to car.

Best, Rudi

Very clear "schema" and explanation Rudi,

Thanks!

Timaximus

Hi Richard,

The light bulbs I use are simple incandescent lights in the shape of a tube. There is one thin wire at each end (axial). They are approx. 1.5 cm long and the filament is almost as long as the light, alowing them to light coaches very well (I also came accros similar lights but where the filament was much shorter than the tube.)

A bulb similar to the one I use is shown here, though I use bulbs of 3 Volts, 50 mA instead.

I believe they are most often use to illuminate needle meters and (analog) frequency scales in radio's, etc. Even at there specified voltage, they seem to burn not to bright, certainly not at over-voltage. This suggests that, burning that way, they lifetime is very long. I use four (in series) to light a single coach. Before that, I used five or six (with a higher voltage), allowing also the balconies to be well lit.

I buy them from a local electronics store. I don't know where you live, but it should not be too hard to get some. In the US, Radio Shack would be the first place to look.

best, rudi

Hi Hans,

You raise a valid and interesting point. In principle, such a situation might have effects.

In my case, however, I only run complete sets of (usually four) cars. At the outer sides of the sets, there are normal (not conducting) couplers. Also, I have the feeling that the conductors of the Viessmann couplers I use do not touch the metal parts of non-conducting couplers (though I did not check that).

But assume such a short circuit does happen, I wonder what the effect is. Possible, the diode will give up. On the other hand, the diode can stand 1 amps continuously and dissipate probably even a bit more, over a short time. Chances are that the circuit protection in the booster will activate before the diode dies, especially in situations where resistance in many rail-to-rail connections limit power as well.

But I prefer not to test that ;-) And anyway, next time I open the cars I might folow your suggestion and add a resister, which is a good idea as such (though it probably would have to be a 0.5 Watt version at least).

Best, rudi