As you know, I want to do realistic operations on my layout. One important element of many operating sessions is a fast clock of some sort. As it happens, I want to have a fast clock on my layout, and to hook it up to various features on the layout to turn them on and off at the appropriate times. I already have the perfect device to do this: a timer switch, scrounged out of a washing machine (after the gearbox blew up). It’s essentially a rotary switch with a great deal of available contacts, geared to a synchronous motor. Okay, so there’s the solution. Except, I want to be able to vary the rate at which the clock runs: for example, 1:2, (2 railroad hours every real hour) 1:6 (6 railroad hours every real hour), 1:12 (12 railroad hours every real hour), etc. The issue, of course, is that a synchronous motor, being an induction motor, can’t really be controlled by voltage or amperage variation like the DC (with rectifier) and universal motors in most of our trains. That leaves one way to control it (other than physically changing the gear ratio, which is probably not practical): a variable-frequency drive. Okay, here’s the issue with that: VFDs, it seems, are mostly commercially available for MUCH MONEYS, as three-phase only (the motor is single-phase), or, most commonly, both. So does anyone have any idea how I could build one?
Actually, I may have just thought of a solution: one of these cheap signal generators:
https://www.amazon.com/Weewooday-Generator-Function-Frequency-Adjustable/dp/B08SMB6DRM/ref=pd_sbs_d_sccl_1_1/147-2377985-0182412
Now, it generates neither sufficient voltage nor sufficient amperage. So would a some optocouplers work to transmit the AC signal onto a 120V line? Like this:
I’d tend to go a lot simpler-
Get a pendulum clock with a decent length pendulum to give “normal” time. You can probably scrounge flea markets and get a cheap 8 day or 30 day for a little of nothing, or even better find an old American 30-hour weighted Ogee type clock(those use super long pendulums, and since they are weighted rather than spring, have near constant rates over the full wind).
Even if the clock has a decorative pendulum, there’s nothing special about it other than being a specific length and having a weight to set the “center” of the pendulum. Make your own out of a brass rod with a weight that can slide the full length of the rod.
There’s some math to figure an exact placement, but a pendulum half the length of stock should run at double speed, 1/4 length should run at 4x. That’s why I say the longer the better, since you can get a pretty wide range of adjustment.
Oh, it’d be easy to build a fast clock. I could even do a digital one, which, as a matter of fact, would be even easier. But I want to use the timer switch, because I need the control–that’s the whole thing here. I do have what could be a solution: a VFD drive I was able to slap together in a circuit simulator…
The 102-ohm and 47K-ohm resistors are both variable, and the output (at the lightbulb) can be varied from nominally 60 hz to nominally 260hz.
The difficulty will be in finding components with sufficient wattage (which can probably be solved decently easy by way of Mouser), particularly the op-amp, which will need to have more than .2 amps of capacity. And the transformer might take a little searching to find.
An Arduino based system and a couple of GPIO boards would be extremely flexible.
Well, here’s a better VFD that I was able to slap together, using optocouplers and a typical Generic Chinese Co. signal generator, allowing it a wider range of frequency and, possibly, more available amperage ratings. But the tradeoff is that it has a much uglier waveform.
That’s true, but it wouldn’t fit my needs so well as the switch.
That’s not going to help him. He’s using a washing-machine timer switch, probably driven just like an electric clock at the 60 Hz line frequency. The corresponding ‘beat’ is 120pps, or in bph 432,000.
The old dodge of using a 32768 quartz movement and tinkering with the quartz frequency won’t work for him, because in a typical quartz clock the frequency divider would run a stepper motor at 1pps (to Move the second hand in ticks’). See the August 1999 article in MR to get the various step frequencies corresponding to different ‘fast clock’ operating schemes. There was an ambitious scheme in the mid-2010s to build a fast clock in the shape of a railroad watch, which i believe was to have a selectable ‘time compression rate’.
A problem may be that the motor and gearing in the timer will not like being accelerated to ‘fast clock’ rpm. Another is that the ‘time’ built into the various contacts may not correspond to the ‘fast dayparts’ when action is to be taken – and even if they are, there may have to be some kind of secondary dimmer to produce the effect of sunrise or sunset, or other action that is not on/off.
Yep, just a 60 Hz. synchronous motor.
That would actually be relatively simple to jigger: a cheap 3D printed case, and a standard small watch movement, controlled as you described above.
The motor is rated at 3 RPM, and I suspect that it and the assorted mechanisms will be able to handle increases of speed–the mechanisms I know, because they can be turned by hand and other motors at higher speeds without issue, and the motor… mostly hoping I get lucky. It will help that the actual dial of the switch will not be the clock; I’ll gear a clock face to the switch instead–meaning that a certain baseline time (probably real time) will be achieved at 60 Hz.
I’m planning to base most of the actions on the switch itself and just see what I can make it do–the darn thing has like 16 poles…
That’s pretty likely–I was also thinking of using relays, including latching, to produce additional functions. I can set up some to serve as logic gates, others to extend a function, and so on and so forth.
Even so, I wouldn’t mind looking at those tables anyway–especially if I wind up making other “fast clock” projects.
Ooh, thanks!
That’s nicer than any of mine!
If you’re going to work on railroad watches, particularly if re-adjusting them to positions, one of these is essential.
Yep–I’ll have to keep an eye out for one of those!
Neat!
I had my fill of self-correcting with the clock on the Thunderbird. This very cleverly corrected the clock regulator every time the adjustment stem was pulled out and rotated. The problem was that if you were correcting after a dead battery you’d have to let the stem back in repeatedly to keep turning it… and the clock would then massively overcorrect, with no visible indication how the regulator was set. I found it a misery.
Yet another reason rack-and-snail is superior to count-wheel, even if you can only set the hands forward without a pull off…
The factories managed to turn them out in volume by without timing machines(although most adopted them pretty quickly when they became a viable option).
A timing machine gives you a lot of information about how a watch runs that can’t be done by observational timing, but at the same time getting average daily rate really is a lot more important in the real world than instantaneous shots on the timing machine.
Also, I use a timing machine to check and occasionally tweak my work. Even when I staff a watch, by truing and poising it properly, then by watching and listening to it running, I rarely have something that needs more than a small tweak on the timing machine to get positional adjustments back in line…
Watches adjusted to positions took weeks to get through the process. In the absence of a Vibrograf-style display (analog, printed on paper tape) the watch would have to run long enough between adjustments to show the relevant timing precision, and some of that adjustment might be thrown out by another position. The only ‘mass-produceable’ exception in the era of American mass production was the Elgin Durabalance, which had no external poising weights and was timed ‘in the watch’ by special machinery – but these balances were never used in 16s railroad-watch movements to my knowledge (perhaps in part because railroad time device people would have trouble servicing them).
Basic poising is less of a concern when timing can be coarsely set with a regulator, but you can readily understand by analogy with static vs. dynamic balancing of road-vehicle tires that you will not get high accuracy getting rid of the low spots on the agate rails. Toward the end of mass 16s production, at least one maker made it easier to replace broken staffs by machining a taper into a bushing in the balance, and a comparable taper in the balance staff, so you’d just press out the broken one and press in the new. In practice… that still required re-poise and then re-adjustment.
Waltham patented the friction-fit balance staff in the 1890s; the first watches to use it were the 12 size 1896 model, and then the 16 size 1899 model(on which basically all subsequent 16 size Walthams were based). Hamilton started using friction fit staffs with the 992E and 950E, then continued the practice on into the 992B and all watches based on it. As the Waltham staffs were primarily used on split bimetallic wheels, they often do need a bit of adjustment after staffing. Hamilton almost exclusively used them on solid balance wheels and IME they rarely need any work after replacing the staff.
I learned timing on a paper tape machine and actually still prefer them to digital, even though I have both.
And yes, dynamic poising does account for the entire escapement assembly and a bunch of other factors that static poising can’t, however in practice I find that the curb pins introduce more inherent positional error than dynamic poising can correct. The curb pins really are a limit for high precision timekeeping especially across positions-in fact if you check Hamilton’s allowed variation for the 992(5 position) vs. the 992B(6 position) you’ll find that even with all of the technology used to both build the 992B, and with access to at least early tube based timing machines, that the 992B is still allowed more positional variation than the 992. There’s a reason why modern high end wristwatches have gone to freesprung balances. You can iron out even more positional variation with a helical hairspring, but no one wants to deal with a watch that thick.
Since you mentioned Elgin’s Durabalance, it’s kind of a shame that it never made it to a railroad watch, as it was one of the few mass-produced American freesprung designs…(not counting the various WWII chronomoters).
Ah, but it did! It’s just that by then the result was a wristwatch… complete with the B.W.Raymond mark of excellence.
I believe the contemporary ‘competition’ was the ETA2124/5 in the Balls (the difference being stem vs. automatic winding) and the 500 and 505 Hamilton Electrics – in both cases equipped with a darling little swan-neck regulator spring to satisfy nominal requirements. My B.W.Raymond also had this, unlike the 780 pictured above.
In general, pre-quartz railroad wristwatches haven’t gotten the interest that the late 16s pocket watches did – which is a shame.
I am old enough to remember the thrill of seeing railroad-approved Accutrons adopted for Metroliner service… and later, by some of the better ex-PRR conductors. One of the ‘regular’ men on the PJ&B ‘dinky’ gave me his when he retired, one if the greatest honors I have been afforded – it still runs and keeps time.