Member postings for Andrew Johnston

From "Definitions & Formulae for Students - Metallurgy" published by Pitman in 1942 at 6d!

Terneplates - steel sheets coated with a mixture of tin and lead

Probably not obtainable now due to the lead.

Andrew

It doesn't look like a brushless controller to me. The board looks a little rough on the upper left side of the two preset pots top left - almost as if something has leaked out? There are also several light brown areas on the right side of the board. Are these real or a trick of the light?

Andrew

The maths seems fairly straightfoward, at least in principle. While I'd agree with the statement that no algebraic equation exists for the involute I also agree with John, why? Equations for the involute are simple (albeit transcendental) and can be calculated to a precision far in excess of any practical need. That is what I did when designing an internal gear. Just calculate a series of points and let CAD join the dots. I can't remember how many points I used but it was significantly more than that shown in the example Bezier splines.

I suppose it would be quicker for CAD to calculate a Bezier spline than an involute. However, I think most gear manufacturers will have their own set of tweaks and adjustments anyway so won't be using just a basic involute. In summary an interesting article but I won't be using the results in anger.

Andrew

It will depend upon the involute, which in turn depends upon the number of teeth on the gear. Looking at the extremes the fit will be perfect for a rack but awful for, say, a gear with 10 teeth than needs undercutting. Another problem is how to assess the quality of fit. Least squares may be approproiate?

Andrew

Go to Drill Service - more types of drill than you can shake a stick at:

Left Hand Drills

Andrew

Just as well I'm not a model engineer. For the crankshaft governor pulley I used a hydraulic copy unit with a hand filed pattern:

To keep the traditionalists happy the governor end pulleys were done with files:

The pictures of full size engines that I have show that both ends of the governor belt have crowned pulleys.

Andrew

They work fine for me in bronze and brass.

Andrew

Definitely not. I'm trying to get some PCBs assembled for work and all sorts of components are suddenly unavailable or on very long leadtimes, like 26 or 52 weeks. I think inventory got used up last year and companies didn't make new stock as many places were in lockdown so it seemed that sales would be low. Now everyone is playing catch up.

Another issue is that the medical profession have failed to notice that a side effect of having Covid is incompetence. I've just ordered two items from a professional electronics distributor. One item was advised as being out of stock. Surely I wasn't daft enough to order something listed as out of stock? When I called to cancel I was told it was listed as in stock but when the picker looked in the warehouse there was no stock. Delivery of the second item can explain that. I ordered one pack of two small 12V batteries. What I got was a box of 10 packs of 2 batteries, ie, 20 batteries in total. I gave up listening to recorded messages this morning saying they couldn't answer the 'phone. So I've emailed them to ask if they want 18 of them back. If it involves more than 5 minutes of my time they can go whistle.

Andrew

Much of the last few weeks has been a nightmare of PCB layout for work. One lesson learnt is that components are now so small that the size of the PCB is dictated by the room needed to fan out the tracks rather than by the size of the components.

However, I have also been working away on what I call the singling valves for my traction engines. These allow high pressure steam to be applied to the low pressure cylinder, temporarily converting the compound to a single using the low pressure cylinder. Here are the parts and one assembled valve:

The body is silver soldered from several bronze parts and all threads (apart from that on the U shaped piece of brass) were screwcut. The valve is not part of the model drawings, but has been scaled from the full size works drawings. I've tried to be faithful to the works drawings, right down to the hole and slot on the end of the stainless steel valve rod. For scale the hole is 1mm and the slot is 20 thou wide:

Here's the valve in situ on the cylinder, with the regulator rod in the background complete with studs and split pin holes as per full size:

This is the internal shape of the body, on a scrap part, again faithful to full size:

Andrew

S275 is a hot rolled low carbon steel similar to EN3B. Like EN3B it is a bit gummy and has a propensity to tear at low feeds and speeds. Drilling, tapping and reaming are fine and it grinds ok as long as one avoids loading the grinding wheel.

I use a lot of it. Most of my traction engine motion work and many other engine items are made from hot rolled steel. Like these slidebar clamps:

I don't worry about the mill scale. I just remove it in the course of normal machining. Anecdotally it seems less prone to rust than EN1A.

Andrew

When were you there? I was at CUED from 1980 to 1986, but don't recall your name.

Andrew

Prepare to be amazed! What about Gaussian pulse shaping? Analogue filter theory used to shape a digital pulse stream.

Another topic to mull over is QAM - quadrature amplitude modulation. It's amplitude modulation but mostly used for digital information, eg, digital terrestial TV in the UK. Each symbol can have many levels, so more than 1 bit can be transmitted in one symbol, 64 and 256 are common values, as the constellation is normally square. Where's the dividing line between continuous modulation and modulation of many discrete steps?

Andrew

That's interesting. At which university did you teach?

Andrew

Precisely, which is why I asked about FSK, as in it's simplest form it consists of bursts of two sine waves, but is considered as being a digital signal.

The definitions of digital signals above are fine as far as they go, but are limited. The question of noise is an interesting one, first analysed by Shannon and Hartley. But consider a GPS signal which can be at least an order of magnitude below the background noise level, but can still be decoded. Speaking of decoding, a comparator deciding whether a signal is high or low is an example of hard decision decoding. Newer algorithms, such as Viterbi decoders, use soft decision decoding based on the reliability of each code. Logic levels at the gate level are normally described as above a given level is 1 and below another given level is 0. Zero is normally included in the lower range rather than as a discrete logic value. The precise voltage is unimportant provided it is above, or below, the specified levels. There is normally a voltage range between the levels where the signal is neither 1 nor 0. Of course more than two levels can be used, although the only common example is a third state of high impedance aka tri-state logic.

Andrew

I must have skipped that class. You're not the first new member to wade in guns blazing and telling people that they're idiots and I don't suppose you'll be the last. Generally it reflects badly on the new poster and while some quieten down many disappear without trace.

As a doctor I'd advise you to visit your GP, as breathing cutting oil fumes seems to have affected some cognitive abilities.

Andrew

That's a new one on me. I've never seen a signal, analogue or digital, that aimed to avoid social embarrassment.

It's convenient, and may be logical, to think of a digital signal as being one of a small set of values. But in reality, however it exists, the actual signal is analogue. Would a FSK signal be analogue or digital?

Andrew

It was started by Taylor and White in the early 20th century and test methods are now encapsulated in ASME standards.

I think that figures of 15 and 20 minutes come from ASME standards for testing of single point tools. The tools are run at a range of speeds and feeds such that tool life is assessed as being 15 or 20 minutes. This allows results for different tools, and from different laboratories, to be compared. It does not mean that the feeds and speeds are recommendations for use in a production environment. That's up to the individual planner. I suspect that the cost of changing a tool (labour and machine downtime) is larger than the cost of the tool itself. So there is some incentive to maximise tool life by reducing speeds.

I have made measurements of surface roughness that show that, for some materials, high speeds are needed to get a good finish with inserts.

Andrew

Hmmm, not an electronics engineer I presume? Electrons most definitely are fussy, and can be extremely annoying when they don't flow where you expect.

The USB signal (1.x and 2.0) is differential, transmitted over a twisted pair with a nominal impedance of 90 ohms. If the impedance is wrong and/or the lines are not terminated properly then reflections and overshoots will occur, ultimately leading to loss of signal integrity.

A digital signal is really an analogue signal. In some ways digital is more challenging than analogue as the fast edges imply a larger bandwidth than might be suggested by the data rate. If the transmission channel is not linear, ie, non-dispersive, then distortion of the signal will result leading to problems in decoding the signal. There's a vast literature available on the transmission and decoding of digital signals, with some pretty hairy mathematics.

Having said the above a quality USB cable will not be expensive. The cables linked to in the OP are nothing to do with technical ability; they're intended to separate fools and their money.

Andrew

Neither am I.

I've wandered all over the place buying turning inserts, but have settled on Korloy from Cutwel for CCMT and CCGT inserts, plus parting off ones. I've buy metric threading inserts from Cutwel but have to be more flexible for fine pitch Whitworth as they're becoming difficult to get.

Andrew

Go on then, start the ball rolling.

Andrew