What Did You Do Today 2020

[Paul LousickParticipant @paullousick59116]

The antique engine and model clubs that I am a member of are still closed because of the lock-downs and it's getting a bit boring at home on my own so decided today to go on another One Man Rally with my traction engine.

Paul

**LINK**

[SillyOldDufferModerator @sillyoldduffer]

Posted by Iain Downs on 05/07/2020 18:28:59:

…

First thing this morning, I found out why my Arduino power feed code wasn't working – after quite some time puzzling. It turns out that Aaduino interrupts and float point don't play nicely. Migrated to long integers and it just worked! …

Iain

Probably not simply that floating point and interrupts don't play nicely as such, more likely it's a timing issue. The thing about interrupts is they suspend whatever the processor is doing, jump to new logic, and then restart the original process. If the original process remotely time sensitive, then it can be severely disrupted if whatever the interrupt routine does takes too long. Particularly confusing when an interrupt takes so long it's interrupted itself!

The general rule is to ruthlessly minimise the work done during an interrupt, getting out of it as fast as possible, ideally within microseconds. One way of doing this is to have the interrupt set a flag telling loop() to do the work outside the interrupt.

Floating point is done in software on an Arduino and – in computer terms – it's painfully slow. Arithmetic done with integers is much faster, but adds other booby traps like overflow. Floating point can be used in interrupt service routines, but keep a close eye on how long it takes. Same is caution with any other time consuming logic like waiting for serial input or a device to respond.

Good job solving it. Interrupt problems are very hard to debug because they're usually time related and the code looks fine. Event relationships are wrong rather than the program logic. One way of pinning them down is to put an oscilloscope on an output pin (or two) and toggle them with debug code when things are supposed to happen. If the oscilloscope shows toggling unexpectedly, suspect something taking too long in the code is disturbing the 'when' rather than the 'what' logic.

One advantage of the Nucleo compared with Arduino is they do floating point in hardware, pretty much as fast as integer arithmetic without the booby traps.

Dave

[Howard LewisParticipant @howardlewis46836]

One thing that no one seemed to mention in the discussion about managers for a practical environment.

If the manager does not have an appreciation of the process, he is a lamb to the slaughter with anyone keen to swing the lead..

"Can't do that, the toggle won't clear the widget arm" and he doesn't know enough to say "Stop talking nonsense. Do it this way. Shouldn't take you more than twenty minutes" BUT it has to be correct!

The word soon goes round that the boss knows his stuff, (even if it really isn't in detail ) So fewer try it on with him, because they are not sure just how much he does know..

And if he keeps his eyes and ears open, he will learn lots more useful stuff for the future.

Once I had a boss who, among many other things, didn't know what the function of bearing was. Full of management speak, but practically useless. He was a nice bloke, but an embarrassment!

Howard

[Paul LousickParticipant @paullousick59116]

"If the manager does not have an appreciation of the process, he is a lamb to the slaughter"

I worked for a company that enlisted a manager like that. It was a successful multi million dollar business but because of bad decisions and not knowing what he was talking about, sent the company broke.

Paul.

[EmgeeParticipant @emgee]

Nice running engine there Paul, not a lot of traffic to contend with.

Emgee

[Iain DownsParticipant @iaindowns78295]

Dave – After working out that it was the floating point, it still wasn't quite working and I ended up creating a new sketch and mucking around with it – with a scope attached to one of the pins.

I was surprised at how poor the timing was. From memory (it was last night now) with nothing happening apart from an interrupt on a timer (PORTD ^=D00001000 or something) I couldn't get more than about 250kHz or so as the blink – so 500K calls to the interrupt per second. 32 clock cycles. Hmm. Not that bad, I suppose, when you do the sums.

Iain

[Paul LousickParticipant @paullousick59116]

Thanks Emgee,

The streets around my home are very quiet and not a lot of cars and people and complying with the social distancing regulations are not a problem.

Driving around the block where I live is easy and I would like a bigger challenge and have loaded the trailer with extra supplies for the trip.

Have I missed anything else that I may need ?

Paul.

Edited By Paul Lousick on 08/07/2020 00:05:04

[Iain DownsParticipant @iaindowns78295]

This has been a frustrating few days.

I've been building a power feed for my new mill (there's another post somewhere) and, to my surprise, the hardware and electronics have been going pretty well. I'm just applying the final coat of paint (to the metalwork).

The software not so much. Which is embarrassing as I have very limited metal working skills, but over 40 years of development experience.

I finally worked out how to get the arduino part driving the motors (help from several members here), but the part that's frustrated me is the user interface. Being a digital person, I wanted to drive it from a touch screen (slide left for faster left and so on + have some present movements).

My original idea was to use one of the various small tablets I've got as the UI. This largely failed because the tablets (early Nexus 7) don't support 'OTG' for hardware reasons. At least I think so. Opinion is not always clear on the web.

I carried on looking at this approach and tried my newer (58 quid) tablet from bangood. This does support OTG and I can drive the Arduino from it with a PlayStore serial terminal.

now to write a program. In c#, please my Java is rusty and I never learnt raw android dev (iOS, yes and Windows Phone yes, not not Android).

Now it turns out that there is limited support for USB serial on android with c#, but some nice man has ported a java library and I could run his sample program on my tablet and control the arduino with simple commands.

However, he'd written the android part in android language and not in the 'cross platform' mode which I understand. I have no desire to learn yet another coding environment. In truth, I'm having enough trouble remembering the ones I've already used. And of course I don't want to use my cheap tablet for this. As well as the expense to buy a replacement, it's really too big – I want something more phone sized!

Aha! I remembered my old windows 8 phone (Lumia 735)! Perfect size and I've written programs for it…

Hunted for it for a while and, unexpectedly, found it in one of the 4 places I thought it might be (last of the 4, naturally). Charged it up and went about doing some background research.

Oh, dear, that one won't support OTG (serial USB) either… But a Lumia 950 will which I can buy for 30 quid from eBay. Cheaper than a new tablet, I suppose.

OH. In the meantime, I'd already looked at using a Raspberry PI, but my Pis are first edition and don't play well with LCD displays unless I get one from China which takes ages and who knows if it will work or pay quite a bit more here – much the same learning curve exists there, of course, though it would be cheaper.

So I went to bed despondent, worrying how to make this work (on the cheap).

Zing! I had an idea! I could swap the USB lead for Bluetooth. Which the windows phone will support and I seemed to think I could get for Arduino. Popped downstairs to find Amazon and Google are my friends. I ordered an d8 quid UK based HC-05 for the Arduino as I'm fed up waiting rather than a 2 quid version from China (I know. A bit hypocritical) and feel I might have a way forward.

Until the next obstacle.

Phew!

Iain

[Joseph Noci 1Participant @josephnoci1]

Hi Ian,

You should take a look at the Nextion displays – sizes from around 2inch up to 7inch – maybe more these days – a serial ( ttl level) interface, available with touch screen, 5v supply.

I use these extensively for non-dynamic graphics and text displays, esp suited to machine control, Radio panel display, etc.

You use a vendor supplied Screen page editor to create the structure of the display ( layout, placement of buttons, colour of buttons and colour change when activated, etc). To create the actual graphics you can use any graphics type editing package, photoshop, etc, I use a freebie – Inkscape – which is hugely suitable for this stuff. Save the graphic as a .png file, and load that into the screen editor mentioned, and define where it fits, how many screens you have, what is on each screen – create sliders, 'rotary' knobs and controls, etc. They just work…the image below has user buttons in the bottom row, and a screen brightness slider center below.

[Adrian RawsonParticipant @adrianrawson27430]

I've done lots of Raspberry Pi stuff and recently CircuitPython on Adafruit ItsyBitsy M4 using OLED and ePaper displays but have never come across this Nextion solution, so thanks for the post. Having had a quick look at their site it would appear that the editor is PC only. Maybe by the time I need to use one they'll have a Java solution that will run on a MAC.

[Joseph Noci 1Participant @josephnoci1]

Seems you might be able to –

Nextion editor on Mac

But not going to get into the Mac/Windows/Linux/Java thing…

[Mark RandParticipant @markrand96270]

"Write a program in C# because my Java is rusty."

Damn I'm old! My C was approved by messrs Kernighan and Ritchie and only just progressed to the new ANSI style.

I did, at one point have 65,000 lines of code in HP Rocky Mountain Basic and 24,000 lines of HP Pascal in use in the company at one time. They weren't interested in selling the programs to our customers and only missed out by a few million pounds in the process…

But at least I turned my nose up at FORTRAN and COBOL (apart from a few programs I had to translate)

[Nigel Graham 2Participant @nigelgraham2]

First things first, before any engineering…

… exhuming the neighbour's house stop-cock from a deep infill of damp soil and woodlice, to try to establish the direction of its inlet side. We couldn't conclusively but it does seem to be from the street.

This was because occasionally I hear water flowing strongly through pipes when I am in my own home, but have no taps running or cistern filling. My fear was that for some reason the feed to my neighbour's property was tee'd from my own, concealed under concrete paths, downstream of my meter! This practice from Victorian or Edwardian times, sometimes survives.

It is strange, though I know from a study at work some years ago, that sound travels very well for quite a long way through the water in pipes. This is used for calculating where along its length, a buried pipe is leaking. The noise of the leak is picked up by sensors at each end of the section under test, and transmitted to an electronic "correlator" that displays an amplitude-distance graph indicating the proportional distances from sensors to leak. Our work was to assess extending the range of pipe diameters that can be tested, above existing limits.

So it is feasible the noises I hear are from quite a way away, but it's odd neither of my neighbour hears them.

'

So unto the workshop.

'

All work for the workshop: a bit of painting, an extra shelf.

Mainly though, assembling the crab on the overhead crane. I should have trained as a car-designer then I would have been even better at designing awkwardness into assemblies – and been paid for it!

It is now useable, serving two-thirds of the workshop. I have still to erect the rails for the last third of the length but it covers the heavier machine-tools and the steam-wagon's assembly area..

The lifting-tackle is suspended from a bar held in the crab by grooves located in keyholes in the two side-plates, enabling changing things. I aim eventually that this bar is the fixed axle for a set of sheaves, but so I can swap these for separate blocks as shown (the untidy tackle is what I happen to have available), or a chain-block, without altering it.

The red bag dangling from the hook holds the line.

Test load? I grasped the cross-bar and lifted myself by arms, and it seemed perfectly happy with my 1cwt 1stone mass!

'

The tatty, part-dismantled horizontal mill in the left foreground of photo 3 is a Denbigh H-series, awaiting refurbishing and putting into service. The vertical mill is a Myford VMC – ooops, should have taken that (empty) set of drawers off the table before taking the pictures!

[Iain DownsParticipant @iaindowns78295]

Hi, Joseph – that's interesting, but I think I will try the last solution mentioned before I switch gain!

Mark – I started off with Pascal before C became popular, then C++, then c#, with a whole bunch of others along the way (most of which I've totally forgotten).

And I forgot to mention that my (12 year old) laptop I was planning to use for test purposes on this also died at the weekend. At least I think it's died. It seems to have got slightly worse after I've tried to fix it. Once among the most powerful laptops available, it's now an ugly bookend. Technology advances, does it not?

On a positive note I did get the drive installed, hooked up to the electronics and driven by a windows tablet. The motor moved the mill table (result!), but I'd got an earlier version of the arduino code (C++) so it peaked out to early.

Iain

[AdrianRParticipant @adrianr18614]

Today I made an M3 counterbore, came out great, cuts wonderfully. Except for one small rooky mistake.

[IRTParticipant @irt]

I finally got around to making a lathe carriage stop:

I still need to replace the bolt with some threaded rod.

[mechman48Participant @mechman48]

Over the weekend.. converted a small grinder to a buffer polisher; had ordered left & right handed pig tails but received both with 10 mm bore so had to drill & tap for 12 mm. Bought 12mm left hand tap & die ( CS ) to suit, not that I'll ever need to tap M12 left hand again that I can foresee. Also made up a pair of new aluminium vice jaws to replace well battered steel jaws on vice.

George.

[Iain DownsParticipant @iaindowns78295]

The carriage stop is like the one I made, but much much nicer.

Iain

[mechman48Participant @mechman48]

I made a similar one, yours is much nicer looking.

George.

[SteviegtrParticipant @steviegtr]

Posted by IRT on 20/07/2020 13:31:14:

I finally got around to making a lathe carriage stop:

I still need to replace the bolt with some threaded rod.

Very nice work.

Steve.

[Nigel Graham 2Participant @nigelgraham2]

Re-erected an ancient and very battered Elliott 'Progress 2G' bench-drill.

I had been falling over it until I had completed a travelling-hoist for the workshop so I could assemble the machine then lift it onto the special-to-purpose timber bench I had to build for it.

A bench sized carefully to take minimum floor-space whilst creating a store for two welding-sets below it, and to allow limited off-table drilling of items too tall for the machine itself. A neat "accidental" touch that emerged as I made it, was that a plywood baffle closing the inner faces of the two legs not only reinforces them, but gives the possibility of a small, very shallow but still useful cupboard in the resulting enclosure!

It was a good test for the hoist-assembly, with a 'Clarke' chain-hoist suspended from the crab, and I watched it very carefully for any signs of distress as I eased the machine assembled except for its motor and table, off the floor.

The hoist does have a couple of snags. The long-travel wheels are a bit too under-gauge, allowing the whole frame to twist or move laterally enough on the track to strike the rails' fastenings in places. I also noticed that under the heavy load of the drilling-machine, the crab wheels were sliding rather than revolving.

'

Another snag is not of my doing… for a change.

I used M8 sleeve-anchors and 10mm concrete-screws for securing the bench to the floor and wall. The former use standard ISO-M nuts, so an 13mm A/F spanner.

The latter's required hole diameters are in regular mm sizes, so you'd expect a 10mm screw's flanged-hexagon head would take a 17mm spanner even though not a conventional bolt/ set-screw. Oh my word no. Their hexagon size is random even in ISO-crat terms, and the only spanner that fitted them vaguely properly, was not even metric, but 5/16-BSW!

'

Anyway, the result when I finally finished for the day, was that I could walk around the workshop without having to squeeze round all manner of obstacles!

[Colin HeseltineParticipant @colinheseltine48622]

Had a hard drive failure in my NAS drive. Luckily having set it up as a three x 3TB drive RAID5 with a 3TB hot swap spare, it carried on working but will take around 24hours to rebuild itself. If I have a drive failure now will lose all data. As is usual in these cases the drive that has failed is obsolete, but been informed by manufacturer tech support that any 3TB SATA 7200rpm drive will be okay. New Toshiba drive now on order for delivery tomorrow.

I run two RAID5 NAS drives and one copies files to the other, they are both on a UPS.

Fingers crossed the replacement arrives okay and works okay.

Colin

[LimpetParticipant @limpet]

Colin I wouldn't worry too much as raid 5 over 3 disks will give you 1 disk redundancy anyway. The hot spare is so you can immediately go back to 1 disk redundancy. You only lose data if a second disk in a 3 disk raid 5 fails

Lionel

[Iain DownsParticipant @iaindowns78295]

Some years ago I had a server with 3 (large) drives in RAID 5, based on hardware on the mother board. I chose this because I got more storage than with mirrored drives.

This worked nicely until the motherboard rusted. Yes. Rusted. I turns out that this server was in the cellar under the water main inlet which had developed a slow leak. The motherboard being drenched (relatively speaking) did not cause any immediate problems but over time the tracks or some chips or something corroded and it became, well, broken.

The easy way to do this was to replace the motherboard with the same model and stick the drives back in. A few cheap encounters with second hand (non-functional) boards from eBay put paid to that idea and I was told by my supplier (who properly know their stuff) that it was that board or nothing.

In the end I had to find some software that de-striped the individual drives into an image and then burn that onto a fresh drive. This took a full man week to get sorted, during which time I was pretty much out of action.

The result of this is that I no longer use RAID 5 and all my drives are mirrored (A 5TB, 4TB and 1TB set).

Iain

[Henry BrownParticipant @henrybrown95529]

Yesterday actually! I picked this VEM 600W output 8" bench grinder up locally yesterday for £50. It had been used by the chaps grandfather from new for sharpening wood chisels and came with a set of O'Donnell rests and a huge magnifying lens that has some dodgy looking strip lights in that I will replace with LED's once I have polished the light scratching from the lens. It had white and pink wheels fitted and came with some other wheels including a coarse new grey wheel that I will fit along with the nearly new pink wheel.

As can be seen it has cleaned up beautifully, only a wipe over with an oily rag, there was very little grinding residue in the guards and they don't look to have been connected to any extractors. It even came with the original hand book and guarantee card that was not sent off.

The O'Donnell rests can go on ebay to offset the cost a little!

[Author]

Posts