The Sequel - Oh Blimey I bought a CNC Lathe (Beaver TC 20)

[awemawson]

OK I've found the Crawford specifications for these collets and it seems that my 15 microns is within specification - seems rather a lot to me 

[awemawson]

All enthusiastic I've recently re-measured and drawn up a pull stud for the tooling of the Beaver Mill, to be made on this Beaver TC-20 CNC lathe.

Going to the machine to review tooling, and powering it up there were odd and unfamiliar humming noises and the machine wouldn't go 'Ready' 

Now over the last few days we have had masses of rain and some very significant lightning - though not (I'd thought) very close.

So far I've spent a couple of days on the diagnosis trail and found a blown up memory card. Fortunately I had a spare, and now at least the controller is going ready.

At this point I've been able to load back ups of the PLC program and the other gubbins - in fact I've been through the process FIVE TIMES 

However, although I can 'reference' the machine, jog in X & Z, rotate the tool turret and select a tool, I cannot start the main spindle. You may recall this is driven by a vast Maudsley DC motor by a KTK DC drive and a propriety field current source. If I program a modest spindle speed (say 100 rpm but it doesn't seem to matter what) there is a slight increase in hum level but no rotation, and after a few seconds it errors with 'spindle speed too high' !

Putting a DC clamp meter on the field drive it's passing a reasonable 5.9 amps. Transferring it (On AC) to the feed to the KTK driver for the motor, it briefly is drawing about 40 amps (at 120 V 3 phase) before erroring.

The othere oddity is that one of the files that's been re-loaded (several times into RAM) maps PLC error and comment numbers to text, but despite this the controller is reporting the number and not the cross referenced scrap of text.

Now this controller has bits of ROM  and bits of RAM in several places and I've never satisfactorily managed to map what is supposed to be where - the manuals are as vague as a salesman's promise!

The spare could be faulty of course and explain both issues, but I seem to remember previously using it. There is another for sale on eBay at a reasonable price, but the postage and customs duties from the US are crippling :

https://www.ebay.co.uk/itm/130842358746?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2060353.m1438.l2649

(This is one of the cards that were destroyed before I got the machine by oozing lithium battery juice, and indeed is the same as the card that was the source of he leak!)


. . . decisions . . . decisions 

[Sea.dog]

Parts are rare. Buy it!! 

[tom osselton]

I’d buy it too it’s hard to say when another comes available.

[awemawson]

You chaps are a bad influence !

Things have gone from bad to worse - I've ended up with a blank screen and no activity 

I had noticed an oddity. If I manually turned the spindle, even a degree, it immediately gave the spindle over speed error message. Now the spindle is monitored by a 'measuring card' that takes input from the optical disk - I had a spare so swapped it - no difference. Now this card talks to the CPU card - I had a spare (in fact the original that came with the machine - it's frame is slightly corroded but I'm sure that I've previously proved it OK. BUT swapping it into the machine - BLANK SCREEN. Putting back the one I took out but still BLANK SCREEN 

The video system is a pair of cards - a clever one and a bank of memory - no spares.  I reasoned that it was possible that the display itself may have died. Just putting in these two cards I got a random display, proving that at least the CRT was good (I'd tried to see the cathode filament glowing but couldn't)

The power supply is slightly suspect. it gives several rails none of which are accessible to monitor except the 5 volt rail which is good. I seem to remember having to inert pins into the backboard sockets to measure the +/- 12 volts but that entails the removaval of the entire controller and it's card compliment - not a job for today.

Meanwhile I have spares of the two video cards on order, and a memory card that has iirc 32 kByte of ram, and slots for three daughter boards that have the controller firmware on them. This card is minus the daughter boards but the chances are that mine are OK.

. . . .I didn't need this . . . 

[awemawson]

A few minor developments:

My son in Houston, bless him, has bought that interface / memory card that I linked to and you encouraged me to get and will post it on to me I'm sure at a much lower price than the $192 the seller was quoting !

In addition to the two video cards, and the bare memory card (sans daughter cards) I have put in a half price offer for a CPU card which has been accepted 

Incidentally it turns out that the memory card in this 820T is NOT populated with the extra 32K of RAM - presumably other models have it.

This >should< mean that I eventually will have spares of all the cards in the controller when international couriers do their bit - all of course are used items, sold as working, but I suspect that dealers probably have no way of testing them so we will see . . .

Meanwhile I have extracted the entire 820T controller from the rack for easier working on the bench. I seem to remember that when I first got the controller to come back from the dead when I got the machine, I was able to bring it up to 'Initial Clear' mode on the bench and only wired to mains - hopefully eventually I can repeat this 

Frustratingly I cannot find my 'Lab Book' covering that initial resurrection period (I try to keep notes on the more complex bits) - my current book only starts at 7th Sept 2018 yet the controller was up and talking to me by 21st June 2018. It's somewhere, it's A4 and blue - but WHERE 

Meanwhile some pictures I took really to make sure that I know where things go when / if I ever get it going again:

[awemawson]

Pulling all the cards (except the two video cards) out of the card cage the screen is still blank, but with the ribbon cable from the video logic cable to screen unplugged the screen shows white scan lines.

(Screen is actually all white - the green bit is something to do with the camera and scanning)

According to the book of words the screen takes RGB TTL levels from the video-graphics card so I need to scope it and see whats what. No time at the moment as cottage guests due any moment.

However I HAVE managed to prove that the +5, and +/- 15 volt supplies are making their way along the back board - I actually found the original leads that I'd soldered wire wrap pins to back in June 2018 !

[Pete.]

Do you think that the fault is definitely due to a lightning strike? I was talking to my mate about this very thing recently (long-time National Grid engineer) he said that most strikes hit the top wire on the pylons which is deliberately the earth wire, or it hits the pylon itself (being a metal frame and grounded). He said when it does hit a phase it causes loads of damage and (predictably) the closer you are to the strike the more things go bang.

[awemawson]

Can't be sure Pete.

We lost power from 05:30 to 14:45 that day. But initially it was going on and off repeatedly (woken by iPhone that was on /off / on charge objecting !)

 Many flashes of lightning, and we lost one battery backed timer program out of five installed around the farm - (fortunately only the pond pump!) and this lathe electronics. However the setting for the DNC controller that connects this lathe to my network (By RS232 cable) were also wiped out - I suspect that it was currents induced in this relatively long cable that caused the problem, but how do you prove it in the absence of a smoking gun?

The lathe WAS isolated at the time so it can only be induced currents.

I know it's proving a right royal pain to sort out !

[awemawson]

Well, positive developments, but I don't know why !

I came over to the workshop this evening to set the scope up to look at the backboard bus. I have no diagrams but it's amazing what you can deduce of card activity if you are persistent.

But I thought, just because I could, I'd return the cards to the card cage and see what happened. Darn me it comes up talking to me! Now it's reporting a "Battery Alarm Power Supply" which isn't surprising as the battery is disconnected !

Plugging back the (modified and remote) battery it now comes up "Ord 1 PP-Memory wrongly Formatted" which is probably not surprising - be good if I knew what 'PP-Memory' was but never mind - I now have a means of diagnosis as I can SEE things on the screen.

. . . life is full of mysteries . . .

Tomorrow I need to perform a rigorous and methodical re-run and see what does what when other cards are plugged in !

. . . fingers crossed . . .

[awemawson]

Apparently "Part Program Memory"

(Thanks Google  )

[Sea.dog]

Oxidation on some of the pins? Unlikely, I know, since most of your kit is tucked up in a nice, warm workshop 

[awemawson]

Possibly Graham, but I'd be surprised.

The other day I carefully returned all the cards to the card cage and temporarily returned the 820T controller to the lathe cabinet. But before I did this I sorted out a height discrepancy in the runners that it sits on. Left side was about 1.2 mm too low resulting in great difficulty aligning it's screws, and as the flange for the screws is fragile plastic it needed sorting. A simple packing piece was inserted under the runner.

So with it back in the lathe  the controller was quite happy going through the entire software re-loading process without reporting any errors.

At this stage all seems normal, the machine can be referenced, jogged, tools turret operated, hydraulic chuck and hydraulic tail stock all worked as they should.

BUT issuing an M03 or M04  (CCW or CW) rotation command for the spindle with a suitable speed value resulted in no rotation, maybe a feint hum, and after a few seconds an error message reporting excess spindle speed (!)

The other oddity is that certain reporting and error messages displayed on the screen should comprise of a number followed by an explanatory text. The number is there but no text, so for example: "6000 Initialise Turret" displays as just "6000" . Now the translation is in a file called PCA.TXT which HAS been loaded - all very odd.

So trying to bottom out the lack of rotation I opened up the connections for the KTK Mentor DC Drive to get at the DC speed command input (+/- 10 Volts sign indicating direction) and was amused to see my clip links still in place from the last time - very convenient!

Measuring the speed command voltage it remained resolutely at 0v - mmm . . very odd. Now the controller is asserting the drive enable signal so all it needs to spin is a small DC voltage to say how much and which direction!

Well years back I made a little DC servo drive box - simple affair - 9v battery, suitable potentiometer, and a reversing switch. So it got a new battery and set to work.

Sure enough, I can drive the spindle quite happily from the little box until the control again reports spindle over speed and drops the drive enable signal.

This actually is quite a relief as this vintage of KTK Mentor are thin on the ground, and even if you can find them they command silly prices.

So what are we left with? Logically the lack of decoding of those numbers points to software in some form (firmware or what I loaded) and the lack of servo drive voltage could be either software or hardware.

Until a few more spares arrive I think I'm at an impasse 

[nrml]

Just playing devil's advocate here; is it worth spending the money and effort on finding and buying increasingly hard to find spares? Wouldn't upgrading to to a more modern controller if you keep getting further problems make sense given the repeated issues with ageing electronics?

[awemawson]

No that's not the journey that I'm on!

(I actually don't NEED a huge heavy CNC lathe at all, but I WILL repair this one I sincerely hope)

[tom osselton]

That would be about  $20,000 over here when I looked at it.

[awemawson]

A VERY big step forwards today:

A parcel from Germany arrived remarkably promptly (Thanks Fedex) containing the two video card spares and the card that hold the three daughter boards that have the main operating system in EPROMS. Now I really don't think these are anything to do with the issue that I am pursuing, and although I have unwrapped them, I have not yet tested them.

My Disco had to go for it's MOT today, so I got the wife to drop me off at the chap who always prepares it on it's annual outing (ex REME knows Landrovers!) and as a sweetener I took her to a rather nice pub (Two Sawyers, Pett Level) so she doesn't have to cook this evening. Lubricated with a drop of Harvey's Best Bitter I got to thinking when I returned - it's almost as though this machine either doesn't know that it HAS a spindle drive system, or that it is inhibited in some way 

The spindle drive system in block form is pretty simple (but they are VERY complex blocks!). DC motor with a Tacho and Encoder attached which is driven by the KTK DC drive. This takes it's command from a DC level set by the 'Measuring Card' that compares desired speed and direction given by the CPU with measured speed and direction from the tacho and encoder.

Does the encoder work? Well yes, once the drive has been told to turn (and failed) you can see the rotational position of the spindle via the controller. So there's a 'tick in the box'.

. . . about now a light light started to glimmer, and get brighter by the second. There is a page of 'spindle settings' tucked away in the controller specifying 'Acceleration Ramp Time, G92 spindle Speed limitation, M19 exact spindle Stop location, and Spindle Speed Limitation Everything was set to zeros.

If you limit the spindle speed to zero, it AIN'T going to spin 

Now I thought that these settings were included in the re-load process, but it seems that they are not. Setting sensible values and now I have a lathe that turns it's spindle on demand - in fact just what a lathe SHOULD do 

I'm still left with the oddity of missing text on the error messages but I bet that that tuns out to be some setting that's been zapped !

[RussellT]

 

It's very satisfying when you can work it out like that.  Well done.

Russell

[Sea.dog]

Another Mawson triumph, and only a "little lighter" in the bank balance 

[Pete.]

See I told you that things always seem better with a beer insde you

[tom osselton]

 
That’s something you wouldn’t expect you’d think it would have a minimum generic setting.

[awemawson]

Graham, I rarely resent buying spares for these things when they are available and the price is sensible. It makes diagnosis so much easier in the future. A case in point is the little battery backed RAM card that seems to have succumbed to the lightning storm - I had it because it had flitted by on eBay about a year ago. Without that spare I'd still be scratching my head 


Other examples: I have a complete set of electronics and servos for my H425 Spark Eroder in a crate in my container alongside another with virtually all the bits of a Fanuc Wirecut wire eroder.

I suppose it comes from a lifetime career supporting obscure equipment in critical applications.

Years ago we were taken over by an organisation more accustomed to supporting commodity PCs where there were usually many substitute spares available at short notice from dealers at low cost. Seeing some exceedingly rare CPU modules on the books at well over £50K a piece the new MD wanted them written off pronto to make the books look better. When I pointed out that they supported a fair proportion of the National Grid infrastructure including Nuclear Power Stations that we had very tight contractual commitments to, and that they could no longer be made his face was an absolute picture

[vintageandclassicrepairs]

Hi Andrew
Well done on getting the lathe electronics sorted, It is always a good idea to take some time away from the problem,
I have found solutions to problems appear in my mind at the oddest of times and places (if that makes sense?)

John

[Sea.dog]

The tube system was also reliant, and not so long ago, on obsolete equipment. For all I know, it still is.

[modeng200023]

Which tube would that be? 
John