Railpower 1300 testing

[mmagliaro]

In contrast, here's what I consider to be a well-designed DC throttle.  This is not rocket science.  It is the
"High Capacity Throttle" from Peter Thorne's 1971 book, "Practical Electronic Projects for Model Railroaders".

I also added a zener in the circuit that feeds the base of the first transistor in the circuit so that it behaves exactly like Thorne's did, except that at maximum throttle, the total pulse amplitude never gets above about 13.5v, while preserving all the pulse behavior at lower settings.  Without that, the pulses can hit around 17-18 volts.  And while that never hurt anything, it isn't necessary.  This is shown at a 400mA load - the same as I did for the 1300.

My main beef about all this is that letting high amplitude pulses go out to the rails smacks of being "cheap".  It's not that hard to keep those under control.

[Point353]

Although these sine wave pulses don't look anything like a DCC command, who's to say that they still cannot damage a decoder?

How would that happen?

[mmagliaro]

How would that happen?

I have no idea.   What happens if you put pulses of some other type on the rails, and they are over 22v?    Maybe it could just harmlessly confuse a decoder. But maybe not.    All I'm saying is that the MRC pack's pulses are over a spec limit, and that is not a good thing.

As far as I am concerned, I am done here.  We are not going to conclusively resolve this without buying a bunch of decoders
and testing them exhaustively on a bunch of power packs to see which ones fail.  My main purpose in this was to see if there was something about the 1300  that made it stand out from other power packs, and there is.

What I don't like about either side in this dispute between MRC and Rapido is that neither one has published any scientific data explaining why "the other guy is at fault", leaving people like us to try to decipher it.    Well, I've done what I can.




 

[Point353]

And here's the 1300 schematic everyone's been waiting for.
I did this in a circuit simulator, so I could also "run" it and confirm that I had the schematic right.  Indeed, I get the
same pulse pattern that the real thing does.  The absolute peak in the simulator is not up to 25.  It is at about 23.5.
But that can easily be caused by the transformer or the Darlington not being simulated exactly the same as physical ones that are in the unit.    The transformer is modeled as 27v peak to peak, which is about what the MRC pack does.

Note:  The 4 diodes ARE in fact wired up as a full-wave bridge.  The funky half-pulses are created by that 10k resistor that they tapped off the AC input from the transformer.

My main beef about all this is that letting high amplitude pulses go out to the rails smacks of being "cheap".  It's not that hard to keep those under control.

How much more expensive would the 1300 need to be to eliminate the "high amplitude pulses"?
Possibly not much, if at all.

Suppose that you remove R1 from its present position (connected between one side of the AC input to the bridge rectifier and the base of Q1) and, instead, connect it between the junction of the +ve output of the bridge rectifier and the collector of Q1 and then (in series with) the top end terminal of the potentiometer speed control R2. 
Would you be willing to rerun your simulation for this modified circuit configuration?

[mmagliaro]

Point353...
SURE!  I'll change the circuit as you suggest and see what it does.

Curiously, I already tried putting a 20v zener between the base of Q1 and Gnd.  This allows the pulse pattern to act exactly as it always did until we get up over the zener voltage on that base and it starts to conduct.  The end result is that at high speeds, the peaks get clipped to around 18v, which is a big improvement over 25, and because it's in the base of the transistor, it is very load-independent (i.e. the pulse pattern remains the same whether you are drawing 50 mA or a half amp.  I had already thought, "Wow... it would cost pennies to include a zener in there".

But your suggestion of just moving an existing component would cost ZERO.  Ha ha...

I will try it and see what happens.

[peteski]

Max, thanks for that 1300 diagram. What circuit simulation software are you using?  Also thanks for all the time you invested in this quest. 

Yes, they are injecting the AC voltage to the transistor bias circuit through the 10k resistor. That is a fairly clever design.

The easiest way to keep the throttle's output voltage within specs would be to use the appropriate voltage transformer.  I wonder if the transformer used in the 1300 was designed back in the day when they uses selenium rectifiers which drop several volts.  Then MRC reworked the circuit to use silicon diodes, but kept using the same transformer?

[MK]

So once Max does the repositioning of the resistor that Point353 suggested, and if that works, can someone take some pictures of how to do this?  I would like to open up my 1300 (in Athearn clothing) and modify it such.  Or at least draw the corrected schematic.  (Before and After)

[mmagliaro]

So once Max does the repositioning of the resistor that Point353 suggested, and if that works, can someone take some pictures of how to do this?  I would like to open up my 1300 (in Athearn clothing) and modify it such.  Or at least draw the corrected schematic.  (Before and After)

I would be willing to do this to the 1300 I've been experimenting on.  But be warned.  No matter how similar that Athearn pack appears and even if the circuits are in fact identical, it might not look all that similar when you open it, at least not so similar that you could just follow instructions like "snip this here and solder this there".

Peteski: The circuit simulator... heh heh.  I was waiting for somebody to ask that.  There is a professional grade circuit modeler called CircuitMaker, by Altium.   Years ago, it was a stand-alone application that you had to pay big money for, and it went through incarnations like "CircuitMaker 2000".  But back then (and this is the 1990s), there was a free version available for personal use known as "CircuitMaker Student".   Anyone could download and use it, with the understanding that anything you did with it could not be for commercial purposes.

That's what I use.  Even though it is an old application, it runs just fine on Windows 7.   The editing and simulation features are really terrific.  And although the library of SPICE models for devices it has isn't huge, it's big enough, and you can add your own devices to it if you know the SPICE parameters.

Years ago, they pulled the plug on the free version and you can't get it anymore.  And now, Altium has a new "free" version of it that is somehow hooked into creating an online account and uploading anything you design with it to a shared library so that the other CircuitMaker users can have it.  I guess the thinking is that if they give you the tool for free, you should be willing to share anything you make with it for free.

I have no beef with this, and I'd even use it that way, but their entire web site and system for using the software was so convoluted that I gave up and went back to using the nice, simple, local executable for the Student version that I have.

[MK]

I would be willing to do this to the 1300 I've been experimenting on.  But be warned.  No matter how similar that Athearn pack appears and even if the circuits are in fact identical, it might not look all that similar when you open it, at least not so similar that you could just follow instructions like "snip this here and solder this there".

Absolutely understood.

[mmagliaro]

How much more expensive would the 1300 need to be to eliminate the "high amplitude pulses"?
Possibly not much, if at all.

Suppose that you remove R1 from its present position (connected between one side of the AC input to the bridge rectifier and the base of Q1) and, instead, connect it between the junction of the +ve output of the bridge rectifier and the collector of Q1 and then (in series with) the top end terminal of the potentiometer speed control R2. 
Would you be willing to rerun your simulation for this modified circuit configuration?

Now that I'm reading this, I realize I don't understand what you want to try.

You mean:
1) from V+ out of the rectified, to R1, then to the collector of Q1 ?
2) what do you mean "and then inseries with" ?

However I try to interpret this, it seems like either R1 is completely bypassed, or else it is in series with the + to the collector, in which case we are going to get almost no output from Q1.

Can you restate this, maybe another way?

[peteski]

Now that I'm reading this, I realize I don't understand what you want to try.

You mean:
1) from V+ out of the rectified, to R1, then to the collector of Q1 ?
2) what do you mean "and then inseries with" ?

However I try to interpret this, it seems like either R1 is completely bypassed, or else it is in series with the + to the collector, in which case we are going to get almost no output from Q1.

Can you restate this, maybe another way?

That is a nice circuit simulator Max!

I also don't quite understand what he means.  The AC signal fed  through the 10K resistor to the transistor's base is not creating high voltage pulses. It actually produces the alternating lower amplitude pulses. It could even be totally elimianted and then all the pulses will be the same size while the maximum voltage would still be just as high as the rectifier output (minus voltage dropped inside the transistor). I think he has it backwards.

As I mentioned, the easiest and cheapest way MRC could have modified 1300 to produce lower voltage out of that throttle would be to use a transformer with a lover output voltage.  No additional components or circuit modifications needed - just one component (transformer) replaced with a different one.

[Point353]

Now that I'm reading this, I realize I don't understand what you want to try.

You mean:
1) from V+ out of the rectified, to R1, then to the collector of Q1 ?
2) what do you mean "and then inseries with" ?

However I try to interpret this, it seems like either R1 is completely bypassed, or else it is in series with the + to the collector, in which case we are going to get almost no output from Q1.

Can you restate this, maybe another way?

Sorry, a picture would have been worth a thousand words.

First, remove R1 from its original location.
Also, disconnect the top end of the pot R2 from the junction of the +ve output of the bridge rectifier and the collector of Q1.

Now connect one end of R1 to the junction of the +ve output of the bridge rectifier and the collector of Q1.
Then connect the other end of R1 to the top end of the pot R2.

Is that any clearer?
If not, I'll try to mark up your schematic.

[mmagliaro]

I see now.  I tried it.

It does reduce the maximum pulse height, but you have to change the resistor to 2.2k in order to get a full 12V average DC out
at full throttle.  Otherwise, you only get about 9.5v and I don't think that's good enough.  With a 2.2k, the pulses are reduced to 20v from 25v, which is a nice improvement, and the maximum DC out is about 12.3v at full load, which is fine.

What I don't like about this is that it changes the fundamental waveform.  You lose the intermediate half-height pulses, which I think would improve motor performance at low speed, and I like the way they gradually "catch up" to the full pulses as you raise the speed, thereby giving it more smooth DC at higher speed, which is desireable for motors.

Instead, here's my idea of putting in a 20v zener diode.  This limits the maximum pulse height to only 18v, and as you can see, it preserves the alternating pulse height behavior.  I do admit I'm wondering if the clipped square wave shape at higher speeds will create more heat or noise from a motor.    But I rather doubt it, since it's certainly no more harsh than  the square waves from a PWM.

A 20v 1W zener costs about 50 cents, and that's a consumer price, buying only 10 of them (which I just did on eBay... ).
So we are talking pennies here from a manufacturing standpoint.

The second plot, at half throttle, doesn't show the voltage on the voltmeter.  Instead, there I was measuring the current through the zener.  As you can see, it's miniscule, so a 1W zener can easily handle this and it won't even get warm.

[mmagliaro]

That is a nice circuit simulator Max!

I also don't quite understand what he means.  The AC signal fed  through the 10K resistor to the transistor's base is not creating high voltage pulses. It actually produces the alternating lower amplitude pulses. It could even be totally elimianted and then all the pulses will be the same size while the maximum voltage would still be just as high as the rectifier output (minus voltage dropped inside the transistor). I think he has it backwards.

As I mentioned, the easiest and cheapest way MRC could have modified 1300 to produce lower voltage out of that throttle would be to use a transformer with a lover output voltage.  No additional components or circuit modifications needed - just one component (transformer) replaced with a different one.

Yes, but the catch there is that they would probably have to use a better transformer.  Transformers all experience some voltage drop as the current increases, and the closer you push a transformer near its maximum current rating, the harder it drops (as of course, you know).  So to get a 1/2 amp output with a fairly stable output voltage would probably require using a better transformer than the one they have in there, and probably one with more like a 1A current rating.  I'd venture that a zener for a few pennies would be more desirable to somebody in the accounting department than a better transformer, cynical though I may be.

[DKS]

Max, this has been an outstanding thread. You've gone above and beyond to determine exactly what's going on, and furthermore, proposed--and demonstrated--a practical workaround to address the problem.

I'd vote this as a "best of" thread for the DCC/Electronics section.