By continuing to use this site, you agree to our use of cookies. Find out more

Member postings for SillyOldDuffer

Here is a list of all the postings SillyOldDuffer has made in our forums. Click on a thread name to jump to the thread.

Thread: Old School Drawing Exercises and 2D CAD
04/07/2020 11:52:57
Posted by JasonB on 03/07/2020 19:54:05:

How did I do? one line produced from a 1.5R semicircle

daves latest.jpg


Fair effort, the curve is fuller than mine, I give it a B+, same mark I give my attempt at scaling and aligning our two drawings!


Here's how my book does it:


  1. Find the origin of the blue 1.5" radius outer semicircle by drawing two 1.5R arcs from points A and B
  2. Draw the blue semicircle and then several perpendiculars to it from the 60° line AB
  3. Then draw connected horizontal lines of the same length, the ends of which define the required curve.

A CAD package should include a spline curve tool to span all the construction ends and draw a clean curve. The easiest way to join them manually is with a French curve tweaked to blend nicely by eye. Not as easy as it sounds to draw really smooth curves by hand, my attempts usually have obvious joins! And see Georgineer's post about being marked down for blending his curves too well in a City & Guilds exam!

Comment on the book method - unlike bending a spline, or picking a random French curve out of the box, it generates accurate reproducible dimensions for the maker. Whether or not the result more artistic than, say, Jason's fuller example is a matter of opinion. Might just be fashion, but some curves, like a Spitfire's wing, definitely look better  to most people. I've no idea why!



Edited By SillyOldDuffer on 04/07/2020 11:59:29

Thread: What Did You Do Today 2020
04/07/2020 10:14:27
Posted by Sam Longley 1 on 04/07/2020 08:59:50:


One question I sometimes posed to degeree students was:-

I have to dig a hole of volume 1 million cubic millimetres. My lorries carry 8 cubic metres of earth. Allowing for earth bulking 25% when dug, how many lorry loads will I need?


A very odd question to ask either Quantity Surveyors or £80k Managers. What was your purpose in asking it?

Look at it from the point of view of the poor old candidate. He has to decide if the question is:

  1. testing mental arithmetic with an artificial problem using wildly dissimilar metric units, or
  2. deeply subtle needing careful analysis, or
  3. looking for a robust response to daft questions, as might be needed to rein in a bumptious apprentice, or
  4. seeking a polite response, as when dealing with a naive customer, or
  5. testing his negotiating skills.

Did getting the answer right or wrong make any difference to getting the job, and if so why?

I'm genuinely interested because selecting candidates is remarkably difficult and error prone: most methods don't work well. Fortunately, most people most of the time are adequately competent however they got the job!

The worst performers are untrained interviewers with no criteria - they look for Old School Ties, firm handshakes, and warm feelings. Assessment Centres are most effective. At them candidates are carefully put through several scenarios by a team over a few days, and assessed continually - even at lunch. Assessment Centres are very expensive, and although better on average they're far from perfect. They still miss exceptional talent and select men of straw for top jobs!


04/07/2020 09:18:57
Posted by Michael Gilligan on 04/07/2020 08:59:53:

Where’s the poetry in that, Nick ?


No such thing as Imperial poetry - it's always been written in metres, ho ho...


Thread: Lathe dogs
03/07/2020 17:52:29

Yet to use a faceplate on my big lathe or turn between centres myself ! I've not had the need to use either straight or bent dogs because if a job will fit in a chuck I'd much rather use that. In the good old days dogs were much used because chucks were expensive. Less call for dogs now, though they're still useful for some types of work, which might be your thing of course. However, if unsure about straight and bent dogs, maybe they aren't needed at all? Yet! When one is needed, you'll know exactly what to buy!

As Bo'sun explains the bent ones are meant to engage in a hole in the faceplate, which is fine and dandy if the two fit together. (Old lathes had a specially slotted plate for bent drive dogs, but general purpose faceplates are far more common nowadays.) A straight dog might be easier to fit if it has to be driven via a bolt or some other arrangement. Both types of dog might need to be fiddled into position.

Made some different dogs once for the one and only faceplate job I ever did on my mini-lathe. Clamps fouled the job so I had to shape some metal specially to suit. Never needed them again...


Thread: Old School Drawing Exercises and 2D CAD
03/07/2020 16:06:10
Posted by Neil Wyatt on 03/07/2020 11:57:51:
Posted by Michael Gilligan on 03/07/2020 11:33:39:

What remains of my brain is at risk of frying ... So can anyone help me, please ?

Taking Dave’s triangle as an example:

We know that

  1. the centre of the inscribed circle is located at the intersection of the ‘bisectors’ of the three angles
  2. the centre of the circumscribed circle is located at the intersection the ‘perpendicular bisectors’ of the three sides

So ... There must be some elegant relationship between those two facts

But what is it ?

A geometric demonstration would be appreciated


As a rider...

There are actually THREE points defined solely by the triangle itself that all lie in a straight line.

Michael has mentioned TWO, what is the THIRD?

And what are their correct names (I had to look these up)?


Michael and Neil have caused a complete meltdown here. Now I know I know nowt about triangles!

A guess; Neil means the centroid, incircle and orthocentre, all of which I had to look up and read four times. And there must be a million other triangle facts out there - my brain hurts.

Anyway, noticed this coincidence not mentioned by Wikipedia's Triangle entry. The inner circle is located by bisecting the angles (point I), whereas the outer circle (point O) is located by dropping verticals from the centre of each side. The coincidence is the white bisectors and green mid-verticals meet together on the perimeter of the outer circle when they're carried on outside the triangle (Points D, E and F).


More! The orthocentre is found by dropping the red verticals at a right angle into the triangle's opposite corner. They cross at the orthocentre (Point H). If the red and green lines are also carried across the circle, they too meet on the perimeter.

So all three centres I, H and O are all related to each other, and Point H is also the triangle's centre of gravity. There's deep logic in this I don't understand. Arggh!


03/07/2020 14:15:11
Posted by Gary Wooding on 03/07/2020 13:45:41:

How about this?



Pretty close, your curve goes slightly deeper than mine:


I'm not complaining!   Fusion did better than I thought it would!


Edited By SillyOldDuffer on 03/07/2020 14:16:39

03/07/2020 13:55:13
Posted by Neil Wyatt on 03/07/2020 12:56:48:

Posted by SillyOldDuffer on 03/07/2020 12:23:07:


With CAD the curve is best done as a simple bezier with a control point at each end. Trivially simple.

Manually just as easy using French curves, the hard bit being finding mine if they turn out not to be where I put them years ago...


CAD may make drawing Bezier curves easy thank goodness, but are they trivially simple?


Doing this manually, I'd use a French curve at different angles to fair the curve, but it would be lucky to find an exact match. One of mine comes close, but not at both ends:


two or moredsc06285.jpg

And someone designed that French curve on a drawing board in the first place! How was it done?

Re Duncan's comment, I'm afraid the curve isn't drawn with arcs. (Maybe it could be?)

Once the approach is explained, easy enough to do manually, with Qcad, or with any other competent 2D CAD package. Not sure about FreeCAD & Fusion360's sketch tools for this particular job. They might not have the 2D primitives needed to construct the curve. (I haven't checked.) They can both do spline curves, but in this example it might be easier to construct the outline in 2D CAD and import the answer rather than sketch it.


03/07/2020 12:23:07

While I ponder Michael's problem and worry from his video if Thales is pronounced Thay Lees or Talez, here's another hard one. Well I think it's hard!

Replicate the curve of this moulding by whatever method. Clue - it's generated from a 1.5" radius semicircle, but the semi-circle's origin isn't on the 60° line. Whilst a CAD package can draw the spline curve given relatively few points to follow, producing a 'fair curve' manually needs several.


Nigel complained about hard-to-make drawings and this might be a good example! But shapes like this are common in the real-world. Streamlining and decorative work demand tricky curves, and made cheaply too! My stair bannister is an example, and dining room coving is another. No sympathy for the poor old tool-maker - the curve has to look right!


Thread: My new lathe a Warco 918
03/07/2020 09:33:59
Posted by Martin Connelly on 02/07/2020 15:16:20:

... It does require start up current to flow through the diodes though so heavy duty diodes would be required. I wonder what would be specified. Don't think my version needs manual resets. I've drawn the switches open in the operated position when they should be closed (nc contact) which may be why you think it needs resetting.

Martin C

Big current diodes aren't my thing but many automotive types look good, this MBR10 from Farnell does 10A (surge 150A) up to 100V for under a quid. I doubt Ron needs a heatsink for this application because the diodes only have to work until the switch takes over again - milliseconds, unless there's a jamb.

I thought your circuit needed resetting because I didn't think it through! Slapdash, sorry...


Thread: Old School Drawing Exercises and 2D CAD
03/07/2020 08:56:29

Posted by Michael Gilligan on 03/07/2020 00:07:14:
Posted by SillyOldDuffer on 02/07/2020 12:33:00:
1. Drop the vertical line AD
2. Draw an arc from the corner ABC that crosses AD (in Blue on Diagram)
3. Draw two circles from the each end of the blue arc where it meets AB and BC (Yellow on Diagram)
4. Draw a line from B to the intersection of the yellow circles. (Light Blue on diagram) This line bisects the angle ABC, and where it crosses AD is the centre of the biggest circle that will fit inside the triangle.

With respect, Dave ... Your Step 1 is cheating, and disqualifies you.
But it’s simply a matter of adding the ‘mirror equivalent‘ of Steps 2&3 to locate the centre point.
... This also provides a legitimate way of placing that vertical, by construction.

Posted by Spurry on 02/07/2020 16:30:38:
Ah, but that needs a set square, ruler and pencil in addition to the compasses. Is that allowed now?

To Pete, anything goes in my workshop. All's fair in Love, War and Model Engineering. Triple points for innovative cheating in my book, but don't try it in an Exam!

To Michael, no need for the 'with respect' - like Ian's comment on dividing, you are completely orthogonal...



02/07/2020 15:42:53
Posted by IanT on 02/07/2020 14:46:02:

That seems a bit complicated to me SoD.

I'd just draw two circles the same (but overlapping) size and then draw a line between their intersection points. That would give me the exact mid-point I'm sure.



You're right Ian, easier to do AND it would be more accurate! blush

Meanwhile, following Duncan's divide by 5 instructions:


You chaps are all too good at this...


02/07/2020 14:17:00
Posted by Spurry on 02/07/2020 12:54:05:

That's sneaky, stating that a ruler could be used, after posing the question. wink The answer was simple in Turbocad, but I could not work out how to measure a 14.42 diameter circle with a compass.


Yeah, sorry about that! My bad entirely - I'm in the dog-house again!

Going back to first principles again, a compass can determine linear distance by using it as a divider or a doubler. Quite easy to halve a line by setting a compass until the intersections it draws match exactly, and dividing can be repeated until the human eye can't cope.


The two red circles are obviously too small, and the two yellow circles equally obviously overlap. The two green circles are either spot on or pretty close.

Good eyesight can divide an inch down to 1/128ths, and dividing from another line at an angle to the target amplifies results allowing even finer graduations. Most struggle to read a 1/64" scale in practice. These days 1/32" is as good as this poor old duffer can do without a magnifying glass.

Dividing scales by factors of two is useful, but decimal scales are even better. Another drawing challenge is how to divide by 5 with a compass to make a decimal scale, and then a vernier? At this rate, engraving micrometers tomorrow!


Thread: My new lathe a Warco 918
02/07/2020 13:38:11
Posted by Martin Connelly on 02/07/2020 12:44:07:

The diodes will be for spark suppression,


I thought that at first, but looking again they have a smart purpose.

If the limit switches are hit they open and stop the motor. Now the clever bit! When the current is reversed with the DPDT, the diode bypasses the open switch and allows the motor to back away from the limit. The motor can't drive past the limit but it can always reverse away after being stopped.

Ron's first circuit is wunderbar because it recovers automatically from a crash. The diodes are needed and they're shown the right way round. Martin's circuit is OK but it requires manual resets after bumps.


Thread: Old School Drawing Exercises and 2D CAD
02/07/2020 12:33:00

Apologies for complicating the circle in a triangle question by allowing folk to assume they had to draw the triangle itself with only a compass! No, it's the radius of the circle I was after - 7.207592

Gold star to Gary for solving it with trigonometry, but I have to mention he's used maths tools provided by someone else! Was it a calculator or a set of trig tables that delivered 71.565° ?

Some lessons learned:

  1. The key to solving the problem is bisecting an angle, with luck this is remembered from school!
  2. Need to know how to bisect an angle with a compass, or
  3. Understanding a 2D CAD package will almost certainly have a tool for bisecting angles, in which case there's no need to remember points 1 & 2 above. CAD wins because it's quick and doesn't make mistakes, but the operator has to know which button to press! Beware of easy to use arty drawing software - it may not do techy stuff like this.

Can the entire problem be done with just a compass? Someone did in the distant past. Imagine a flat damp sand beach, a few sticks, a home-made compass and a length of string. The corners of the triangle can be found with the compass and fixed with sticks. A straight line can be marked on the sand by pulling the string tight between two sticks and twanging it on to the ground. Builders use much the same trick with chalky string to this day. No need for a hinged pair of compasses either - the 'compass' can be two sticks with a length of string held taut between them.

Here's bisecting done with a compass:



  1. Drop the vertical line AD
  2. Draw an arc from the corner ABC that crosses AD (in Blue on Diagram)
  3. Draw two circles from the each end of the blue arc where it meets AB and BC (Yellow on Diagram)
  4. Draw a line from B to the intersection of the yellow circles. (Light Blue on diagram) This line bisects the angle ABC, and where it crosses AD is the centre of the biggest circle that will fit inside the triangle.

Comment - problem looks easy, but has hidden depths! More to technical drawing than CAD buttons or pencil and paper, but CAD hides a lot of underlying complexity.



Edited By SillyOldDuffer on 02/07/2020 12:33:57

02/07/2020 11:30:30
Posted by pgk pgk on 01/07/2020 22:26:21:

70mph round a 100 metre curve? I'd guess you're tilting it 90 degrees and fitting ejector seats...


Absolutely! Engineering is about balancing cost and practicality, and it isn't easy. Given the clue that the operators wish to maintain an average 70mph over the length of the line, the engineer must pay careful attention to the design of this reverse curve. He can't assume it's OK for trains to take the curve at 5mph; the curve has to be optimised for speed. As building railway line is too expensive to let practical men experiment on the job, this was done with drawing boards, slide rules and maths in an office.

So what is the maximum speed a train can be safely taken round a well-designed 100 metre curve? Transition curves and super-elevation (banking) allowed.

In the history of railway building, it was common for lines to be routed cheaply at first because money was short. They put up with slow tight curves, gradients, weak bridges, and long diversions even though these hack profits down year on year. When money was available later inefficiencies were often eliminated by building cut-offs. A good way of getting rid of a reverse curve is to connect both sides with a direct tunnel or viaduct, but these have to pay for themselves by increasing the railways efficiency. No place for practical men in this type of engineering either - it's all about cost estimating. Practical men have the skills needed dig tunnels, but the hard part is making the railway pay.


01/07/2020 21:57:37
Posted by DC31k on 01/07/2020 20:49:10:
Posted by SillyOldDuffer on 01/07/2020 17:40:18:

Score double for solving it with only a compass.

If you can solve it using a magnetic thing, surely that deserves considerably more than double.

If you can solve it using a pair of compasses, that also deserves more than double as drawing the straight lines that form the angle bisectors and whose intersection defines the inscribed circle centre is quite challenging using an instrument for drawing circles.

Referring back to the original challenge, I am dubious that any railway would be constructed from simple circular arcs tangent to straight lines. AIUI, transition curves are always necessary (and drawing those, whether manually or on a computer, is somewhat tricky).

I plead not guilty to both charges, M'Lud.

A compass is a drawing instrument, also known as a Pair of Compasses, but 'Compass' in this context is correct.

When there's plenty of room to lay out a railway line there's no need for transition curves and few early railways had them. My example didn't give a scale, but if the radius of the tight curve is a mile, or even a kilometre, I suggest there's no need for transition curves.

But the point is a good one. Laying out a practical railway the design of curves is complicated - it's necessary to take trains safely round tight curves as fast as possible.

Question for transition curve experts: what form should my reverse curve example take on a standard gauge line if the small radius is 100 metres? And what speed limit should be applied if the line is laid flat? Even more difficult, what improvement can be obtained by applying super-elevation, and what's the optimum tilt assuming the train normally averages 70mph?



Edited By SillyOldDuffer on 01/07/2020 22:02:10

01/07/2020 17:40:18

Another Geometrical Drawing. What's the radius of the largest circle that can be drawn inside this isosceles triangle? Score double for solving it with only a compass.



01/07/2020 17:23:12
Posted by Michael Gilligan on 01/07/2020 16:03:16:
Posted by SillyOldDuffer on 01/07/2020 14:47:48:

Pendant alert!


dont know

Dangling from a thread ?

Incidentally ... Orthogonal simply means ‘at right angles’ or ‘mutually perpendicular’


Not to us geeks : In computer programming, orthogonality means that operations change just one thing without affecting others.

What I dislike about computer-speak is the constant disappointments. Orthogonality is conceptually related to 'Loose Coupling', which has nothing to do with wife swapping. No joy with 'Back-end Penetration Testing' either, and that one is dangerous to Google!


Thread: Coronavirus, advice from ME
01/07/2020 17:07:56

Magazine arrived today and the editorial is controversial, possibly by accident, because the conclusion is sensible enough - 'get back to normal as soon as we're allowed to'. Apart from being counter to government health advice, it's apparently based on a belief that the Prime Minister irrelevantly refers to 'morbid funk' in private. I guess 'Surveys suggest that, even after our freedom is restored, many people will be reluctant to get out and about' was written before packed beaches, raves and football celebrations hit the news!

This issue got even worse when I read 'Confessions of a Model Maker' by John Moorhouse. John rubs a hundredweight of salt straight into the open wound of my guilty conscience. I am Mr Untidiness and Task Avoidance incarnate. Someone has grassed me up - I'm so obviously the Jackass John is writing about.

I was devastated long before getting to the cruel truth of 'Starting a Project is much easier than finishing.' After delivering this comprehensive thrashing John finishes off with the brutal promise: 'To be continued'. I may have to cancel my subscription for health reasons.



Thread: What Did You Do Today 2020
01/07/2020 16:12:44
Posted by Keith Wyles on 01/07/2020 15:22:09:

One issue with ignoring the cm is volume. A mm3 to m3 is a big jump, and I wonder how many can visualise either.

Imperial is so much easier to visualise!

A gallon of water weighs 10lbs so a ton of water must be 224 gallons. As 1 gallon occupies 0.1605437 cubic feet, a ton of water must occupy 35.9617888 cubic feet. Or 62141.9710464 cubic inches. Being called the Water Ton means it's only used to measure Oil.

And 0.9617888 should be a fraction: it is 86561/90000

Don't forget these are proper British Gallons rather than weedy American gallons, and a Freight Ton of water is 40 cubic feet on both sides of the Atlantic. Otherwise the sums go horribly wrong.



Magazine Locator

Want the latest issue of Model Engineer or Model Engineers' Workshop? Use our magazine locator links to find your nearest stockist!

Find Model Engineer & Model Engineers' Workshop

Latest Forum Posts
Support Our Partners
Allendale Electronics
Eccentric July 5 2018
Eccentric Engineering
Subscription Offer

Latest "For Sale" Ads
Latest "Wanted" Ads
Get In Touch!

Do you want to contact the Model Engineer and Model Engineers' Workshop team?

You can contact us by phone, mail or email about the magazines including becoming a contributor, submitting reader's letters or making queries about articles. You can also get in touch about this website, advertising or other general issues.

Click THIS LINK for full contact details.

For subscription issues please see THIS LINK.

Digital Back Issues

Social Media online

'Like' us on Facebook
Follow us on Facebook

Follow us on Twitter
 Twitter Logo

Pin us on Pinterest