Gordon H

No argument there. But that is the point which most supporters of these are ignoring - this compromise need not have been made. The cost and complexity of the kludges and compromises so many users are now being forced to adopt to make these operate in a sensible manner far outweighs the few extra pence it might have cost to produce the signals that way in the first place - and everyone would be happy. So do I, which is why this discussion went the way it did in the first place. Control systems are what I do for a living and as a hobby, though in totally different fields. These could have been the right signals for almost everyone (operationally, if not in detail) if they had implemented the control in the way that has been suggested.

Simply using a 9V battery has already voided your warranty, I suspect. Will the mod work with the recommended (insisted?) supply of 16V AC?

I'm afraid they are totally different situations. It is nothing to do with the use of momentary contacts as such. For point operation you would use two momentary buttons, one for each direction. If the same principle was applied to these signals - one button for 'clear' and one button for 'danger', all would be well. It is the use of a single button with a toggle action that is being questioned.

Sadly, the numbers of modellers who take the trouble to construct/adapt/scratchbuild UK style OHLE to the extent of producing actual usable wires remains so low that I suspect this thread contains contributions from them all - certainly all those that I know of, and I started in the field myself over 30 years ago. Perhaps things will start to change with the introduction of the 85 - but then our group had the same thoughts when Hornby introduced the 86 and nothing further came of that for a very long while. I will only believe the corner has been turned when I see a few 1st/2nd generation AC EMUs being produced - not just fashionable 'cash-ins' like the Javelin.

Yes, it was an unfortunate turn of phrase used in the heat of the moment, but I still find it hard to credit that so many appear not to care about such a fundamental limitation of a brand new product in these days where widespread acceptance of deterministic electronic control systems (not just DCC) has just taken hold. I have no issue whatsoever with Dapol's other efforts - just this one which really sticks out like a sore thumb. I have already shown above that Dapol themselves could have achieved this aim if they had chosen to do so. The figures I quoted speak for themselves. Why would I want to contact them directly when this Forum is here for the purpose? At least here it is all in public view. As a signalman, would you not expect your lever positions to directly reflect the signals you control with no doubt or uncertainty? Wouldn't it have been so much better if Dapol had designed their gizmo that way in the first place for negligible price difference, if any? The operating method I have been advocating is no more 'difficult' to implement than the existing one. A similar 'dirt' cheap' switch is used - and it has the advantage of showing you directly which way the signal is set without needing to look at it, assuming you can. I think we have to give the majority of modellers a bit more credit than this with regard their electrical capabilities. If things were really that difficult they would surely have given up at the first attempt to get a point to move under electrical control. Unfortunately for some, this hobby is largely reliant on electricity, so some understanding of general principles is bound to be required, though even that is not necessary in this case. Does this mean that 'prototype' signalmen have always had line of sight to every signal they controlled? I find that hard to believe. It is increasingly clear that Dapol are happy with their sales of these signals, so it is understandable why they are reluctant to doing anything to improve them. However, as the title of this thread is 'Dapol working signals review', it is inevitable that any shortcomings are going to be pointed out, and that is all that I have been doing - and giving reasons why, and suggesting possible alternative strategies for consideration. Sorry if that offends some people, but when an issue as fundamental as this one becomes apparent, surely it is better to highlight the situation and provide explanations for those who may not realise the consequences?

Copied from a thread in the DCC section of the Forum, but more relevant here... I doubt whether true enthusiasts would see it that way. All the enthusiasts I have spoken to on the subject think it is a major blunder. The tragedy is that it could have been so much more to so many more men by a relatively trivial change to the design. As to where you get the figure of 50% more to make it work sensibly, I can only suggest you review your component sourcing processes because the reality is that it would cost nothing like that sort of figure. With an initial signal purchase price of about GBP 25, that would mean a price differential of more than GBP 12.50! For comparison purposes: The mechanism stays as-is - no differential A double sided PCB of identical size and similar complexity is required - no differential The same manufactured switch contacts can be used - no differential. Now we are down to pure electronic component content - and even some of this is likely to coincide with that used already. Basing the components on a design that is known to work and fit in the same enclosure, essentially a small microcontroller, a voltage regulator and two half-bridge drivers plus a few discrete components. Taking 4000 off as a working figure, total electronic component cost is GBP 2.43 - including VAT! No idea what the cost of the components in the existing design might be, but as a wild guess say GBP 1.00 So, the true differential cost is likely to be less than GBP 2 for a properly controllable signal that satisfies a whole lot more people and maintains Dapol's reputation for innovation. I think that would have been worth it.

Ironically, I have little or no use for these signals at all myself. My concern from the start has been that Dapol failed to foresee the obvious issues that would arise from their selected control method. Chances are that no individual would ever be in a position to place a special order for a sizeable batch of 'mechanism only' signals. It would take a consortium or another business concern to do that. Mind you, if that did ever come to pass, perhaps it would persuade Dapol to seize the opportunity and make the corrections for themselves. For interest, the MERG replacement PCB mentioned previously has been successfully demonstrated, so it may only be a matter of time before these become available.

The unfortunate thing about the retro-fitting idea that you still have to buy the whole thing from Dapol at the going rate, even if you are going to throw their PCB away. Thus there is little incentive for Dapol to make any changes themselves to correct the shortcomings that have been identified. If Dapol can be persuaded to supply the mechanism only, so much the better, but I doubt whether they would even consider it.

And that is the critical point, of which many appear not to appreciate the significance - each pulse on the pair of wires merely moves the signal to the 'opposite' position to that which it is currently. It doesn't say 'go to clear' or 'go to danger' as you might reasonably expect (or demand) of a semaphore - it just says 'change position'. As such it is indeterminate as far as any properly conceived control system is concerned. To me this represents a fundamental flaw in Dapol's concept.

So, if I understand it correctly, the slider cuts off power to its own motor, but then remains in position and sends the same power on to the next one as a continuous feed? This being the case, I am struggling to understand how it lends itself to use with the Dapol signal switching scheme - other than switching power to the signal as a whole. Where do you get the pulsed switch closure from to make the signal change state?

Just wondering... Do Tillig point motors give out a pulse for this cascading feature? Can't find any details of their operation at present.

There is no particular difficulty in making these signals do 'something' under the control of a DCC decoder or from a Lever Frame. The problem is that you cannot guarantee what that 'something' is, or tell what has happened other than visually. With this design, no amount of relays, latching or otherwise, can tell you whether the signal is actually showing danger or not. The innovation as such is good, but the execution on the control side is poor in today's terms. It would have cost little or no more to make it work in a sensible manner by design, which would have satisfied a whole lot more people. Whilst there is no doubt they are likely to sell well, it will be interesting to see how many ever appear on exhibition layouts, where correct operation is a 'must' (if the signals move at all). If manufacturers included that aspiration as a design aim, their products would sell even better as there is generally no better recommendation. Let's just hope that Network Rail don't buy any scaled up versions!

To operate properly (to my way of thinking) would still only require an on/off switch, using just the two wires as now - and equally no worry about which way round they are connected. Switch On = Clear, Switch Off = Danger, simple as that. It was pretty obvious in the original message that it implied DCC Accessory use. Ah, but that's the point, isn't it? When you pull your lever, how can you be sure that the lever position reflects what the signal has done? Inadvertent double activation or switch bounce could easily cause the lever and signal to get out of step. To minimise the chances, you will have to religiously ensure that all levers are left in the same position between operating sessions. Move a lever with the power off, and that signal will immediately be incorrect next time round. See above. You could easily find that your signal goes to danger in front of the loco and clears when it has passed if the signal gets out of step with the pulses that it receives.

It's not just DCC users who will be frustrated by the ill-conceived control method these signals use. Anyone wanting to use a lever frame with them will have similar problems, A toggling mechanism where you can't specify clear or danger directly from your control switch can hardly be described as a step forwards. Nice as these signals might appear, if you can't control them properly you may as well use static ones.

So Dapol actually specified a mechanism with indeterminate position control? Or was such a fundamental consideration simply overlooked? If the same approach was taken with locomotives, there would be a huge outcry if users could not predict which way it was going to move each time it was placed on the track. Also, it would seem you were expecting these mechanisms to work with 12V DC. Otherwise, why would you have tested them that way? Does this mean that the intended supply voltage range was omitted from the spec too? If not, surely the Chinese manufacturers should be taken to task for producing something which fails your tests? Dapol are certainly to be congratulated for taking the initiative in this field, which has long been neglected by mainstream manufacturers. The signals themselves look pretty good to me, albeit as a non-conoisseur of semaphores and their details. However, this appears to be yet another case where it would have been be no more difficult or expensive to produce something with an interface which would have satisfied virtually everybody's requirements, including established modellers and/or anyone using a control system such as DCC. I really can't see how the availability of working semaphore signals will have any effect on the number of youngsters taking up the hobby. Concepts such as DCC have far more appeal in the digital age. If anything, DCC users will be put off using these signals simply because their control system won't know and can't know which way they are set whenever you switch the system on.

And that reason is? Although longevity is always an issue with electromechanical devices like this, it is the fundamental method of operation that appears not to have been considered adequately. A simple toggle action each time the switch is closed is completely unacceptable, especially in this day and age where expectations of operational accuracy and the capabilities of control systems to make it happen are much higher than they used to be. Just wondering... were Dapol simply presented with this mechanism by their Chinese suppliers as a case of 'this is what we've got', or was a requirement specification produced?

The control signal is not a square wave as such, but it is a logic level 'digital' pulse of defined length (as noted) sent periodically, typically 50 times a second (20ms intervals). Also, 'wavelength' is not really the correct term to use because that too has a specific meaning in electronic 'circles', particularly those involved with RF. 'Pulse Length' is more appropriate here. Sorry to be pedantic about this, but correct use of terminology is important to avoid confusion and ensure everyone is talking about the same thing. The 'servo twitch at start-up issue' is quite widely known and reported, and there are several aspects to it. The essence is that the moment power is applied to the servo it doesn't really know what to do until a pulse is received, so the internal servo circuitry plays its own part in what to do in that situation. It might tell the servo to go to a default position, probably one or other endstops - or perhaps do nothing (which would be preferred). It could also be that the current servo position does not corresepond with the length of the first received pulse - the servo may have moved between sessions or not have been restored to a default position before power was removed last time. The circuit creating the control pulses may take a finite time to start producing valid pulses - this is part of the basic design of a Servo controller. The cause of start-up twitching seems to be combinations of these effects, and a common 'one size fits all' solution for all circumstances may not be possible. The usual first attempt is to add a pull-up or pull-down resistor between the pulsed line and one or other supply lines. This at least sets the pulsed line to a known state while the controller sorts itself out. This is equivalent to Steve's final comment of 'sending the control signal before turning the power supply on'.

Can't disagree that it looks the part, but the downside is the price you pay to have joggle controls permanently fitted - which theoretically you would only ever need to use once during installation. The resulting £12 per servo control (excluding the servo itself) seems a bit over the top. Interesting also that there seems to be a facility for connecting a setting box to it, going by the board legend.

The usual trick to overcome this is to add a couple of forward biased rectifier diodes to the output to drop the minimum back to zero.

Ah! Much better... Does make me wonder though, whether it would be equally effective to use the LM317T itself (or possibly a second one) as the output stage. After all, the rest of the components are only there vary the voltage at the base of the ZTX450 to set the overall output level. A similar effect could be created by using the LM317 ADJ pin. You would get the additional benefits of the LM317's own internal protection capabilities too. Another more readily 'do-able' alternative would be to replace the ZTX450/2N3055 combination with a single TO-220 packaged Power Darlington transistor such as the BDW93C. Much more convenient to mount than a TO-3 package. Just a thought...

The electrolytic capacitor symbols are drawn the wrong way round, though the '+' signs are shown in the right place. The 'empty box' half of this symbol variant should be the positive, which is intended to obviate the need for a '+' sign at all. Sorry to be pedantic about it, but a similar situation arose in a recent Railway Modeller article diagram, and this month's edition carries a correction for it.

Here are some pictures of panels made some time ago for the Nottingham Club layout 'Carstairs'. The first panel shown, with the complete track diagram, is used to generate serial messages to send out to the local area panels. The internals of one local panel are shown in the second picture. The third picture shows a small area on the large panel, indicating the general appearance. The diagram was drawn using AutoSketch, and printed onto self-adhesive polyester sheet material, then covered with self-adhesive laminate film. All from the Rapid Electronics 'Quick-Laser' labelling range of materials.

Rather than choosing according to 'Size', a better approach is to choose by 'Rating'. The type of insulation used on the wire also has a bearing on this, but for the most part we would be dealing with PVC insulation. 7/0.2 is equivalent to 24AWG, and is normally rated at 1.4A 16/0.2 is equivalent to 20AWG, and is normally rated at 3A Note that the current values shown at the Wikipedia link are Fusing Currents, i.e. the current at which the copper wire itself would get hot enough to melt and go open circuit . Not recommended for model railway use! The normal ratings are much lower because they take into account the limitations of the environments for which they are intended.

Did the kebab van arrive late today? Only I see there isn't any smoke coming out of it yet...

Just unearthed one of my Lima 87's fitted with a cross-arm pan for a few photos. Must be well over ten years since it last saw active service under a wire, so it's all a bit grubby at the moment having had no maintenance in that time. Unfortunately, the elastic springs have perished, but otherwise it would still be serviceable with a bit of attention. The only proprietary parts used were the head (Lima, came with the loco) and the insulators (Sommerfeldt). One of the most awkward aspects of the construction, which a kit version would also need to overcome, is that the lower arms feature a square tapered cross-section. Mine were hand-filed, having first drilled a 0.5mm hole at the 'thin' end for the upper arm pivot. I believe this was one of six pans assembled as a batch, once the necessary assembly jigs were constructed.