Member postings for Marcus Bowman

I have read the original articles, and have the research material, but I am puzzled by the exact detail of the roof-mounted cooling arrangement. The published drawings for the model seem to me to represent this incorrectly, or at least too simplistically for my taste. Do you know of any really good photos or drawings which show full detail of the coolers?

Yes; but my objection is that it is widely recommended for use in the as-supplied condition when making lantern pinions. If the surface is polished, that reduces the diameter, which then means the specified matching drilling diameter is wrong. Lantern pinions have merits of their own, but, when used instead of cut gears, the blued pivot steel is used as a convenience, so one would expect the drilling diameters to be correct for the finished diameter of the rod. That diameter is variable, if the rod is polished or turned.

More modern materials may be more expensive, but are dimensionally more accurate, and have a surface finish which means they can be used as-is. They are also harder than pivot steel. I guess pivot steel was the most convenient workable hard material 100+ years ago, but I am inclined to move with the times, except when restoration demands the use of age-appropriate materials.

Marcus

Does your edition have Fig 114 showing the bracket in which the hammer is to be pivotted? The bracket is to be fixed to the front plate with 2 screws. The 3 sections of the bracket can be pressed together or soldered.

The trip pin on the minute wheel should be approximately 4mm (5/32" in from the edne of the wheel, andf aligned with the minute hand. Make the lifting piece over length, then trim it so that release occurs exactly on the hour. I would take care to allow enough material to remain so that the arm can be given a final polish but still release on the hour. Longcase clocks often have bent pins to compensate for wear...

Fig 116 should show dimensions of the lifting pin. The pin is held on the shaft using an M1.8 x 0.35 (10BA) screw, bearing in a flat on the arbor (also shown in fig 116).

The tail of the hammer arm rests against the stop block on the bracket (shown on the un-numbered fig. entitled 'bracket components' and shown in the photograph Fig 115a. I am guessing that photo may have been a later addition, as it shows the assembled parts. Fig 115 itself shows the Parts of the "one-at-the-hour" striking gear.

Marcus

As a motorcycle enthusiast, I have had many British bikes,and I certainly agree about the lack of sympathy in any restorer who attempts to convert British threads to metric simply to avoid finding or making a proper replacement Whitworth or BSF thread. Very nasty.

BSF threads are the same threadform as Whitworth threads but have finer pitches. BSF threads are specified in the same British Standards document as BSW threads, as the fine-pitch version of the Whitworth series. Hence the term 'Whitform' which covers all the thread series which have the same thread outline (55 degrees, rounded crests, tolerance classes, etc). You are right: 1/4inch BFS has 26tpi, whereas 1/4inch BSW has 20tpi. BSF extends the Whitform series by providing finer pitches for the corresponding diameters, compared to BSW.

I always felt the BSF fasteners where a more elegant thread series than the visually coarse BSW bolts. But that's just me. BSW is probably more appropriate in aluminium or cast iron, while BSF would be my preferred option for steel.

Marcus

Yes; the real world often deviates from tight standards, for practical reasons. It seems to be common practice in the USA to truncate the crests of the Whitworth threadform, presumably to remove the need to accurately form the crests of the male threads. That removes the problem of interference between Whitform crests and metric/UN roots, British Standards defines the dimensions of truncated Whitworth threads, saying that 'the rounded crests at the major diameter of the external thread and at the minor diameter of the internal thread [are] removed at their junctions with the flanks'. I did read somewhere that the original reason for truncating the Whitworth crests was because the standard of gauge-making in the USA, in the early days, lagged behind the accuracy of British gauge-making, so truncating the crests removed that problem from manufacturers. Truncating certainly makes a lot of sense. The clearance between mating threads depends on the tolerance class of thread (class of fit) but most commonly available threads are of the 'normal' class i.e. the sloppiest fit, so its no surprise that some 55 and 60 degree threads of the same pitch can be mixed. It's a bit like mixing BSW and BSCycle if they both have have 26tpi.

Marcus

There is an error in my last post, and I can't see how to edit it (logged out, then back in, and as noted by others this removes the chance to edit my post).

The area of a circle is pi x (radius squared). Swept volume = 3.14 x r x r x stroke. So scaling dimensions by 1.5 scales volume by 1.5 x 1.5 x 1.5 (one for each occurrence of radius and one for the stroke).

Given that the boiler volume scales in the same way (and I don't know if that's the best way to scale a boiler, fire grate etc), the cylinder and boiler volumes might still roughly correspond. Clive's answer is still the most practical, I think.

Marcus

Clive's answer makes good practical sense. But if you are scaling the bore, will you not also be scaling the crank throw, along with all the other dimensions? In that case, the swept volume will change as you scale the bore. It will change again if you scale the crank throw. Swept volume = 3.14 x diameter x stroke (or throw), so scaling by 1.5 multiplies the swept volume by 1.5 x 1.5 = 2.25

If you just want to scale the volume by 1.5, you would either need to keep the stroke the same as the 5" model, and multiply the bore by 1.5 or leave the bore the same and multiply the stroke by 1.5 (which you probably need to do if you are scaling the other linear dimensions). Intuitively, it seems wrong to scale one dimension without scaling the other. Then what about the dimensions of the valve gear? Surely that requires everything, including the stroke, to be scaled by 1.5. Then there is the power, which must depend to some extent on the swept volume, suggesting the bore and stroke should both be scaled. Surely the volume of the boiler will increase as its diameter and length are both increased, suggesting you would need a scaled bore and stroke to cope. It starts to get complicated.

Which is why Clive's advice seems sound.

Marcus

British Standard Cycle threads use a 60 degree threadform, with a shape which visually resembles the Whitworth form. The 60 degrees makes the thread 'flatter', but the crests are rounded, like the Whitform threads, although the radius is different. The height of the BS Cycle thread from crest to crest is 0.5376p as opposed to the Whitworth at 0.640327p.The radius of the crest is p/6 (= 0.167p), as opposed to the Whitworth crest radius of 0.137329p.

So; BS Cycle threads are not Whitworth threads.

BUT, the author may have taken account of the fact that BS 811 : 1950 (still the current standard) provides the following note:

'It is customary practice in the cycle industry to use a 20tpi series of Whitworth form threads as an alternative to the cycle form thread series...'

That's not the same as saying that the BSCycle threads are Whitworth; and the practice seems only to apply to 20tpi threads.

Although, in general, BSCycle threads are 26tpi, the smaller diameters (1/8, 5/32 and 3/16 inch) are not. 1/8" is 40 tpi, while 5/32 and 3/16" are 32tpi. It would not surprise me to learn that the 1/8" BSCycle thread is sometimes used in place of the 1/8 x 40tpi imperial Model Engineer thread, especially where the focus is on the diameter and 40tpi specification in a mixed bunch of older taps or dies, but it is not the same thread, and should not be a fit for the proper Whitworth form ME thread. Mating the two will result in damage to the thread flanks even in a short hole.

Marcus

I have just noticed the 'smileys' the site has added to the sketch. Wasn't me: honest.

If someone can tell me how to upload a sketch as an attachment, I can do that later tonight. If not; the smileys need to be edited out.

Marcus

Edited By Marcus Bowman on 22/01/2018 19:13:33

Testing the RTC is quite straightforward. Here's a sketch (below). As indiated in the comments, you will need to instal the RTC library from Adafruit (link inside the sketch.

// Date and time functions using a DS3231 RTC connected via I2C and Wire lib
// This example taken from Adafruit. Info at https://learn.adafruit.com/adafruit-ds3231-precision-rtc-breakout/arduino-usage

// Wire.h is part of the Arduino IDE
#include <Wire.h>

// RTClib.h comes from https://codeload.github.com/adafruit/RTClib/zip/master
#include "RTClib.h"

// Connections:
// RTC SDA to Arduino A4
// RTC SCL to Arduino A5

RTC_DS3231 rtc;

char daysOfTheWeek[7][12] = {"Sunday", "Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday"};

void setup () {

#ifndef ESP8266
while (!Serial); // for Leonardo/Micro/Zero
#endif

Serial.begin(9600);

delay(3000); // wait for console opening

if (! rtc.begin()) {
Serial.println("Couldn't find RTC";
while (1);
}

if (rtc.lostPower()) {
Serial.println("RTC lost power, lets set the time!";
// following line sets the RTC to the date & time this sketch was compiled
rtc.adjust(DateTime(F(__DATE__), F(__TIME__)));
// This line sets the RTC with an explicit date & time, for example to set
// January 21, 2014 at 3am you would call:
// rtc.adjust(DateTime(2014, 1, 21, 3, 0, 0));
}
}

void loop () {
DateTime now = rtc.now();

Serial.print(now.year(), DEC);
Serial.print('/';
Serial.print(now.month(), DEC);
Serial.print('/';
Serial.print(now.day(), DEC);
Serial.print(" (";
Serial.print(daysOfTheWeek[now.dayOfTheWeek()]);
Serial.print(" ";
Serial.print(now.hour(), DEC);
Serial.print(':';
Serial.print(now.minute(), DEC);
Serial.print(':';
Serial.print(now.second(), DEC);
Serial.println();

Serial.print(" since midnight 1/1/1970 = ";
Serial.print(now.unixtime());
Serial.print("s = ";
Serial.print(now.unixtime() / 86400L);
Serial.println("d";

// calculate a date which is 7 days and 30 seconds into the future
DateTime future (now + TimeSpan(7,12,30,6));

Serial.print(" now + 7d + 30s: ";
Serial.print(future.year(), DEC);
Serial.print('/';
Serial.print(future.month(), DEC);
Serial.print('/';
Serial.print(future.day(), DEC);
Serial.print(' ';
Serial.print(future.hour(), DEC);
Serial.print(':';
Serial.print(future.minute(), DEC);
Serial.print(':';
Serial.print(future.second(), DEC);
Serial.println();

Serial.println();
delay(3000);
}

Compile, then upload. Then open a Serial Monitor window. There is a pause before the first time reading appears, then an update every 3 seconds. Works very well.

Marcus

Updates are a pain. Updating to the current version caused the IDE to refuse to compile anything correctly, so it will not run on my MAC. Won't compile on XP of course, but does do the job nicely on W7.

Yes; I know my Mac OS is 'no longer supported' as is XP, but it is annoying when things like this are not at least backward compatible to the extent that they continue to work with existing sketches.

Marcus

You might look at AccelStepper, available at: **LINK**

That is a PWM library, designed for steppers, but giving a PWM output nevertheless. It accelerates up to a maximum, then, as nears the end of a sequence, decelerates. All, or part, of that might be useful.

Marcus

There are lots of ready-made programs for clocks and watches, but although using one of those MIGHT have saved time, we would not have learned as much as engaging directly with the problem and the code. I certainly learned a lot by studying the RTC chip structure and its internal registers and counters; which is something I had glossed over in other projects.

Thanks for the excuse for a mental workout!

Marcus

I agree with all that John says. These are running shafts, under a bit of load, so the holes should be tapered and burnished, and the shafts polished and burnished. A traditional method of lubrication would be to add a small oil sink on one side of each of the frames (the outside, usually). Adding an oil sink at the front would mean it was quite visible, and, on this kind of clock, you may not wish to do that. The BHI Journal recently had an interesting article and discussion about oil sinks on the inside of the frames, a method which has been found on some existing clocks, and prompted some debate. I'm not so sure they would be sufficiently accessible there, or that the oil would not be smeared out of the sink by end movement of the fusee shaft, in this clock.

There has been some discussion, on this forum, about the shape of the back cock; and a discussion, too, about the gearing (DP or module sizes). My book is a 1998 edition, so you probably have a more up to date version.

Lovely clock, and a nice project.

Marcus

So is that a bit like what happens in a radio controlled clock, where the hands run to 1200 on startup and wait for a first time signal, then go to the appropriate time? In that kind of clock, loss of signal can result in constant cycling, so the ability to do a check, as DoubleTop suggests, is a good idea. Re-usable points of reference are always good, I think.

I, too, am keen to see what evolves from this project. Sounds a lot of fun.

Marcus

Here’s a thought which interrupted my beauty sleep at 5AM.

It’s a bit broad-brush, but the application is at the end.

It is interesting to reflect that this problem has been solved quite effectively already, in a mechanical clock. As I am sure you know better than me, a clock has two distinct parts – time, and strike (or bong, as we are calling it). The time side does just what the name says. If we ignore, for the moment, the arrangements for ‘warning’ (i.e. the just-before-the-strike arrangements preparing for the first strike in a particular sequence to be delivered swiftly) at the appropriate moment (quarters or hours), the time side trips the strike sequence. While the strike sequence is doing its thing, the time side carries on. There is also no possibility that a second strike sequence can be triggered during striking. After the trigger event, the strike does its job independently, including using a rack or a count wheel to determine how many strikes of the bell take place. When the strike sequence has finished, everything is reset, and the strike side waits for the next trigger.

One clear demonstration of this kind of signalling by the time mechanism to an independent strike mechanism is in a musical clock with lots of bells. Generally speaking, the whole of the musical side can be removed independently, leaving the time side just doing its own thing.

So; in the Arduino system, the RTC provides the time, and the Arduino can work out what that is in seconds, minutes, hours and so on, using an interrupt.

Perhaps that same method can be used to control an independent electronic strike ‘mechanism’.

One way is to try to have the Arduino fetch the time, decide when to trigger the bongs by looking for an appropriate time, and control the timing within the bong sequence. That’s a combined effort, but the time and strike ‘mechanisms’ are separate functions. It is true they run using just one processor, but that’s a convenience allowed by software. It does lead to the kinds of problems you (and we) have been experiencing, because we humans have to work out how t interleave what are actually separate functions hanging off just one processor.

A second way would be for the Arduino to fetch the time, decide when to trigger the bongs by looking for an appropriate time, then signal the number of bongs to another device which will sound the bongs and deal with their timing. The Arduino takes nothing to do with that, and just carried on dealing with the time. That’s pretty similar to what a mechanical clock does. You may say that is inefficient, but in a nod to established mechanical solutions, it can be implemented at low cost. Add a Nano or something similar, to handle the strikes.

Then, the Arduino repeatedly reads the time from the RTC and decides when, and how many, bongs should sound. It makes available a sound_the_bongs flag, and a number_of_bongs.

The second device does a somewhat similar job to the Arduino, to deal with bong number and duration. It waits for an interrupt generated by the Arduino sound_the_bongs flag, reads the number_of_bongs, and deals with the timing of signals to relays etc.

So the time and strike ‘mechanisms’ are essentially separate, and function much as in a mechanical clock.

The elements of most of the code required has already appeared in earlier posts.

If you have an Arduino and it is connected to an LCD display, for example, the LCD display essentially handles turning data into visible characters. The Arduino doesn’t do that part. All the Arduino does is trigger an update and make data available to the chip on the display. It’s the same in most computer systems, with the CPU passing signals to a graphics or sound or network card.

In theory, cunning programming should allow a single Arduino to handle both time and strike functions (see earlier post commenting on co-processing). In practice, I wonder if it is not simpler to separate them and use two processors.

Alternatively, fetch the time using an interrupt, then control the timing of the bonging using an internal timer on the same Arduino. That makes it appear as though the Arduino is doing two things at once. The wonder of speedy execution and juggling of electrons. It’s a mixture of hardware interrupts and software timers.

Then I fell asleep again, until I was interrupted by the alarm.

Marcus

James,

Ah, but that was work...

I thought for a moment you were going to say you used to go to a gym and could bench press 500Kg, but then you stopped and haven't gone for years. Now you find you can only do a bit, and you have forgotten how.

Welcome back to mental gymnastics...?

Marcus

James,

Looking again at snippet 1, and working on the basis that it is always useful to know why something has gone wrong, I can't see where you do a LOW digitalWrite. Surely once the led is turned on, it needs to be turned off again after a period of time. Otherwise, it will stay on until you go around the loop again and turn it on again....and then it will never be off.

Snippet One
void Strike_the_Hour () {
unsigned long Present_Each_Hour_Strike_Time = millis(); // Sets the present time
Elapsed_Each_Hour_Strike_Time = Present_Each_Hour_Strike_Time - Previous_Each_Hour_Strike_Millis; // Calculates length of time that has passed since last event

while(number_of_strikes < 5){

State_of_Hours_Activator_LED = HIGH; // Activates the solenoid.
digitalWrite(Hours_Activator, State_of_Hours_Activator_LED);

:::::::so this turns it ON, but for how long? And where is it turned OFF again before you go around the loop again?

Even if it did go off, it will go on and off so quickly you will not see the off states.

SOD's program sets a specific delay time for that purpose, so that's one way of sorting the problem.

number_of_strikes ++;
}
}

Just a thought. What has been proposed so far are good workable bits of code.

You mentioned earlier that your were reading the time from an RTC module. Does it, by any chance, make use of an interrupt on the Arduino? I ask because the interrupt will interfere with the millis() function. At the very least a long ISR will interfere with the count required for the value returned by millis().

Which library, if any, are you using with the RTC module?

Marcus

I have a friend who has just solved the roof element of this problem. His workshop is in a detached single garage with a cenent/asbestos type of roof (as mentioned by HowardT). It dripped badly and the whole place was a rust-generating environment. He has just replaced the roof with a proprietary product consisting of panels which have a ribbed metal outer skin, a thick foam layer, then a ribbed metal inner panel. The roof is now leak-free and doesn't drip. Instead of insulating the walls, he runs a small tubular heater all the time. Seems to work. One benefit being that the workshop is never chillingly cold. The roof was replaced in a day or so, and the panels seem easy to work with. They look the same as the panels used on many newish industrial buildings. Not cheap, but good things seldom are.

If it was my workshop, I would insulate the walls. Speaking from experience of two previous detached workshops, insulating the walls and ceiling (or roof) makes a huge difference. I had a large garage-sized wooden workshop which was lined and insulated with Celotex. Then I had a small detached single-brick outbuilding which I strapped and lined, insulating with rock wool. I also installed a night storage heater in that one, and I could run down the garden and stand in my workshop at any time of the day or night, with the temperature at a constant 64 degrees. Not even a hint of rust. Joy.

Marcus