Digital solenoid point motor

[96RAF]

I see Gaugemaster has released its solenoid point motor with a detachable decoder that runs from either DC or DCC using just 2 wires. bEing detachable allows the decoder to be mounted on the solenoid or remote. Tad pricey at just under 20 quid, but given individual prices of a solenoid and a decoder maybe not, however slo-mo motors are not much different in price.

 

Could Hornby not adapt their solenoid motors with a similar retro decoder. I have been asking for a DCC point and/or point motor for years on the wishlist.

[Chrissaf]

Something RAF forgot to mention about this new product is that the decoder also has an integrated CDU as well.

.

Want to know more, then click here

[96RAF]

At 2200mFChris it must be a one shot device, but thats all that is required - throw me but make sure I go.

 

For comparison the R8247 PAD has 4 x 4700mF caps.

[Chrissaf]

Yes Rob, I had noticed the relatively small cap value. 2200uF is not a lot. But as you say a bit of oomph, if even slight, is better than no oomph at all. The downloadable instructions state that the product works on DCC voltages as low as 14 volts, so presumably the solenoid coils and the integrated CDU components are matched for reliable operation.

.

Pity the product doesn't include any frog switching. It might be possible to interface the switched LED outputs to a low power relay, possibly with a simple transistor driver circuit. But then that starts to increment the costs.

[96RAF]

@Chris

I have just been trying my relay system to prove the R8247 problem in another thread (I'm waiting over there for your answer to big electrical sums posed) and I can throw 2 paired R8234 surface mounted solenoids (low resistance, high current) continuously and reliably as fast as I can press the switch. (2 x 2 0hm solenoids connected in parallel to the output from a GM CDU supplied by H&M2000 15vAC output..

[Chrissaf]

Rob,

I did see your post containing......

.

........can one/more of our exceedingly clever electronics chaps tell me the numbers.......

.

The maths is exceedingly complicated as the formula's need a University Degree in Calculus.....alas in my case, I have an appreciation of the concepts, but not knowledge of the Calculus detail. I did do Calculus briefly at college, but to be honest I struggled with it, and whatever I did learn about it has long been forgotten.

.

The measured 2 ohm of the R8243 is the DC resistance of the coil, but the coil is an inductor. The 'Henry' value of the coil is also unknown. When a voltage is applied across an inductor, then the current build up lags behind by about 90 degrees of phase (or is it the voltage lags behind the current - I can't remember). Again this maths is more appropriate to an alternating rather than direct voltage waveform, thus the maths is only somewhat relevant in a broad sense. In the same way that ohms law cannot be directly applied to an AC circuit.

.

What I am trying to say in a cak-handed way is that the amount of current a pair of 2 ohm solenoid coils in parallel will draw will be based on Impedance Z rather than Resistance R. Because of the current/voltage lag due to inductance, the value of Z will be higher than R - but it is not easy to estimate by what factor as the unknown 'Henry' value of inductance would be needed.

.

15 volts across 1 ohm (2 x 2 ohms in parallel) in a pure DC resistance environment would infer 15 Amps based on Ohms Law. As Z (due the the coil inductance) would theoretically be higher than R, then the max current of two R8243's in parallel should be less than 15 Amps. At a very rough and ready guess, maybe half or a third of 15 Amps. That is 5 to 7.5 Amps.

.

Capacitors if sufficiently large enough can supply very high currents but only for very brief pulses. This, of course, is what makes them so ideal for use as a CDU.

.

The bottom line to your original query is probably that it will require physical experimentation, as a purely theoretical mathematical modelling solution will be too difficult.

.

The GM CDU powered via a H&M2000 15vAC output that you have mentioned should, in theory, be far more powerful than a R8247. I would be surprised if it couldn't throw 2 x R8243's in parallel.

[Nick_]

Rob, I have a DCC Concepts ADS-2fx that is 100% reliable in switching a pair of Peco surface mounted point motors. Looking at it it has a 25v 2,200mf cap. I am unable to reliably switch the same pair of Peco surface mounted point motors using an R8247. This is vey strange if you recon the R8247 has about 8 times the size of caps!

Nick

[Chrissaf]

I too have found the CDU output of the R8247 very weak indeed.......I have three of them.

.

I appreciate that this following statement doesn't fit in with the physical R8247 recharging time issues that have been reported. But I have always been led to believe (maybe wrongly) that the four 4,700uF capacitors are used one per output port. In other words the R8247 contains four individual CDUs not one shared one.

[96RAF]

@Chris

 

Thanks for the explantion and yes induction is different from resistance and outwith my bailiwick of knowledge for sure.

 

OK - magnifying glass on, peer into R8247 and engage guessing mode:

 

The 4 caps are ganged in parallel, so no Chris they are not 1 per port specific.

 

Output from the main and only processor (PIC) feeds 8 x mosfets** which in turn link to each side of the 4 ports (each marked + & -).

** For those who don't know  - metal oxide semiconductor field effect transistor (a switch).

 

The common of these ports (each marked C) feeds back to the +ve side of the 4 x parallel ganged 25V 4700mF caps, which are charged from track voltage via a couple of diodes (are these protection for the controller against feedback?). The -ve side of these caps feed back to the input side of the PIC via another mosfet.

 

There is aso a diode between each side port (+/-) and the linked common ports © to isolate each port side from the CDU discharge.

 

There are other bits in there feeding the PIC input side, but by searching the numbers on-line I guess they are to do with voltage regulation, stabilisation, etc, although there is a resonator in there as well (clock maybe?).

 

So I guess this is how it works - the bank of 4 caps charge in unison from track voltage and, as and when commanded by the controller, the PIC selects the appropriate one of 8 mosfets, which triggers the full CDU bank into the selected port. The diodes prevent this discharge affecting any other port outlet. Recharging begins ready for another selection.

 

Thats as I see it, but I am open to rock throwing as usual.

[Chrissaf]

Rob,

Further to my earlier reply on theory. What I failed to make clear is that the theory about Impedance only relates to the inductor initial powering up and charging cycle. That is at the instance of switch on and shortly afterwards. Like a pulse from a CDU for example. In a DC circuit an inductor will charge until it reaches magnetic saturation. Once it is saturated and the voltage across it is stable, then in effect Ohms Law comes into play based on the coils DC resistance.

.

In a CDU circuit, the CDU pulse current will probably start to decay before the solenoid coil becomes totally magnetically saturated.

[2e0dtoeric]

Also, you have to include the length, gauge, type of metal, resistance, voltage and current drop of the connecting wires between CDU and solenoid!

A very long thin steel wire wastes more power than a short fat copper one!

If you really get fussy, you have to include the DURATION of the pulse into your calculations, as well - not forgetting the mechanical friction of the solenoid core inside the coils, and the mechanical resistance of the point spring and tie-bar.

[96RAF]

I'm real grumpy now - Twice I've tried to post a detailed answer and twice its disappearred so I give up. No I didnt take the advice to copy it before posting, as finishing it and sending it was automatic and it was gone before I realised.

 

@chris - thanks for the responses.

Sorry but the 4 x caps are ganged in parallel.

 

As I said I had written up how I think the R8247 works based on observation of its internals but now I've lost the will to write it out for a third time.

 

No matter lets try again..

 

The bank of 4 x 25v 4700mF caps is charged from DCC track voltage via a couple of diodes (presumably protection for the controller against feedback).

 

This side of the caps circuit is also connected to all the output port commons (marked C).

 

The negative side of the caps circuit is connected to the input side of the main processor (PIC) by a mosfet (i.e. a switch).

 

The output side of the PIC is connected via 8 x mosfets to the + & - connections on the ports.

 

Between the positve and negative sides is a set of diodes (again x8 - one in each output line).

 

 

There are other bits in there like a voltage regulator and a resonator (internal clock maybe).

 

Anyhow this is how I think it works:

 

The cap bank is charged from track voltage and upon command from the controller the PIC tells the associated mosfet to open the selected port which allows the cap bank to fire into the correct side of the correct solenoid.

 

Recharging begins awaiting the next command.

 

Open to rock throwing as usual...

 

[2e0dtoeric]

I'm guessing the two diodes will give a rough dc supply from the ac voltage on the track. As they are just charging up a couple of cap's, the unsmoothed dc doesn't matter. If they were driving an audio amp, they would release a horrible buzzzzz at 100hz.

[Chrissaf]

@chris - thanks for the responses. Sorry but the 4 x caps are ganged in parallel.

.

I'll take that as gospel then, as unlike me, you have opened one up and looked inside. That makes Nick's comment even more relevant. If the R8247 internal CDU has 4 x 4,700uF of capacitance (18,800uF) why is it so limp wristed in operation**.

.

Edit** a rhetorical question.

.

Edit2 - Good point Eric about taking other physical factors into account such as connecting wires and pulse lengths. Brings home the point I made about difficulty of trying to construct an accurate mathematical model.

[96RAF]

@ all

The Peco motors must be low resistance as well. I have a couple so could meter them tomorrow for comparison, but if low resistance means high current broadly speaking then I'm as clueless as everyone else as to why an R8247 struggles but Nicks kit doesn't on a fraction of the horsepower unless his is running higher voltage of course, kind of nitrous oxide - overclocking points decoder scenario.

 

The pucka voltage regulator must be looking after the PIC.

[96RAF]

Its done it again...

[Chrissaf]

Write it in Word......

[Teedoubleudee]

Write it in Word......

 

Or Notepad if you don't have Word.

[Fishmanoz]

Just on the maths, it's even more complicated than anyone has indicated so far.  Start from the fact that impedance is dependent on frequency for inductors and capacitors.  At DC, the inductor, ignoring stray resistance, is short circuit and the capacitor open (I can't remember whether current leads or lags volts either Chris, but whatever it is for one, it's the opposite for the other).  At very high frequency, the inductor is open circuit and the capacitor short circuit.

 

So now hit the inductor with a current pulse, coming from the capacitor. The leading edge of the pulse is very high frequency but, if sustained, the frequency drops.  Figure that.  Now remember the CDU capacitor and the inductor are connected in parallel and you have a tuned circuit that will start with a high voltage then decay at the tuned frequency.

 

Finally, as soon as the magnet starts to move in the coil, it acts as a generator producing an opposite reactance in the coil.  So yes, lots of calculus, which I've forgotten too but the result should be a big enough pulse to throw the solenoid, as we knew in the first place.

 

PS.  Managed to write and post this before being logged out RAF, but the edit box is big enough the Cancel and Save buttons are over the top of it, as is the time signature for my post.

[Nick_]

Just to be clear, my tests were carried out using the same points, same point motors and same wiring. The R8247 was removed and one channel of the ADS-2fx wired in in its place. Both use screw connectors so no soldering necessary. I have an eLink with a 4A power supply and use a bus to distribute power.

[96RAF]

I put the meter across all three point motors and these are the figueres I got - much different from those quoted by Hornby:

R8234 6 ohms

R8014 8.5 ohms

Peco PL11 sidemount 7.3 ohms

Make of that what you will.

The inside of R8247:

/media/tinymce_upload/36d563362e16c977af3980a0b22c0fc3.JPG

 

you have to look under the caps to see they are 25V 4700mF. For some reason the picture rotated when I uploaded it. The top end equates to left in the next picture.

 

/media/tinymce_upload/5a434768a053708f010d7f33bf7380a5.JPG

 

The big blobs of solder in the centre are the cap connections. D's 1-4 are diodes 'regulating' the track voltage going to the caps(per 2-eric's post above), Diodes D's 5-12 are inbetween the caps positive line and the mosfets line and Q's 1-4 are the dual mosfets that switch the command from the PIC U1 to allow the caps to discharge into the solenoid..

[Chrissaf]

Just to clarify leading and lagging and refresh my memory.....I looked it up this morning.

 

With an inductor - current lags voltage.

With a capacitor - voltage lags current.

[Chrissaf]

Adam, could it be "discharge".

.

EDIT: Probably not, as this reply has not been blocked - unless the " " fooled it.

[96RAF]

Hi RAF96,

 

Both your posts (excluding the one with images) were automatically blocked by the profanity filter. I haven't been able to identify which word the profanity filter incorrectly blocked but I'll look into it. Apologies.

 

That's OK Adam - I thought I had a time out problem or even worse a 'Man-in-the-Middle' hacker lurking on my pc monitoring my traffic looking for something they could use nefariously.

Rob

[Chrissaf]

An interesting discussion...........who would have thought a simple solenoid would be so complicated.

.

Thanks to Rob for tracing out the R8247 internals.

.

Most CDU's have a direct connection between the capacitors and the solenoid via a metallic switch (manual or relay contacts). From what Rob has discovered the R8247 puts a semi-conductor Mosfet in the circuit. Maybe the Mosfet is the culprit that is strangling the capacitor output to make the R8247 firing so weak. Thus, why would the R8247 need so much capacitance ie. 4 x 4,700uF if it wasn't to overcome weaknesses in the circuit design.