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

[russ57]

Russ

[awemawson]

Thanks Russ

I started the day determined to bottom out that 24 volt monitoring relay issue that I linked out temporarily. The way it works is that CR10 is located on a remote card in the Chuck hydraulic control cabinet, and has it's coil straight across 24 volts. It's normally open contact is brought back to the big control cabinet in the rear of the machine.

Similarly CR20 is located in the Tailstock hydraulic cabinet on a similar remote card with it's normally open contact brought to the rear of the machine.

These contacts are wired is series (according to the diagram!) the common point being 'Terminal 100' the other side of the CR10 contact is wired to 24 volts and the other side of the CR20 contact is wired to a digital input where the voltage is monitored.

Now I knew that Terminal 100 was at 24 volts (so the CR10 contact is closed and working) but the other side of the CR20 contact at the digital input was at 0 volts. I had just wired the digital input to 24 volts temporarily to get round the problem.

So - open up the Tailstock box and pull the card out to check the relay. The card is located on the common nylon PCB pegs  - it pushes on and a barb pops sideways to lock the card in. Easy to remove by pressing the barb back - except that I couldn't get at them. OK there is actually a proper tool for this job, essentially just a suitable diameter tube that presses onto the peg and releases the PCB. No, I didn't have one, but I have a box of bar ends and a lathe, so yes now I do - made it SO much easier 

Then I gingerly powered up with the card free floating on it's wires and check the relay contact. Sure enough - closed powered up, open with no power. So if it's not the relay it must be the cable. Belled the cable out - no it doesn't go to Terminal 100 - it goes to Terminal 99 - and nothing else goes to terminal 99, it's all on it's lonesome doing nothing !

However, unlike the diagram, there are two extra wires on Terminal 100 that are not on the diagram, a Green and a Violet, so at some time I'm going to have to trace where they go. But I took a chance and moved the wire from 99 to 100, and guess what, it works. Now as it was the CR20 contact can never  have been in circuit and it never worked - very very odd as I certainly haven't moved that wire 

OK knock that issue over (and created another mystery) so back to trying to initialise the Turret.


Later Edit: well of course I couldn't leave it at that so I chased those wires. The Green one goes into a cable form labelled 'Zero Ref' and the Violet one goes into a cable form labelled 'Position Switch' presumably providing power to these devices 

[hermetic]

Sort of points to the possibility that some other accesory, now removed, was connected between 99 and 100 to modify the sequence, tracing the two non standard wires may give a hint!
Phil

[awemawson]

OK Turret Initialised 

I spent some time trying to master the Siemens Sinumerik implementation of MDI - MDI being a way that you can give the machine a few manual commands to move or spin or change tools. It is also used to send initialisation codes to the Turret Controller (a Baldor SMCC 602090-102) which is a microprocessor controlled servo system looking after turret rotation, clamping etc.

The trouble was that I had initially miss-understood a command line in a crib sheet that I'd been given, and having entered it the machine (correctly as it was wrong) refused to run it, but I could find no way to delete the wrong information.

Much experimentation and button pressing and we are initialised.

So now I can jog all axis, select tools, start the spindle and it seems all systems are 'go' I've not actually yet found how to extend the tailstock ram or spin the powered tooling, but progress continues 

[vtsteam]

Just catching up....exciting day yesterday! 

An you're moving right along today, with all axis moving. Good man! 

[awemawson]

As is the way with these things it's one step forward, one back !

Last night I was experimenting with MDI and tool changes. I selected tool 10 and the machine did it's thing rotating the tool turret. It pushes the tool disk forward, rotates to location, then moves it back - I assume that there is a Hirth index disk (*) which it then precisely locates on. I then told it to select tool 5 (where we had been before) - it completed most of the move but failed to finish, and the tool disk looks to be 4 mm too far forwards, as the coolant pipe no longer reaches the port on the rear of the tool disk by that amount.

As a slight consolation I did get the tailstock moving out and back under program control !

So major investigation into the Turret. I removed as many covers as I could and managed to gain access to the internal terminal box where the 'position sensor' leads terminate - these are proximity sensors running off 24 volts (which is there) but oddly all four proximity sensors are saying 'no detect' - I'd have thought one or two would be active.

While climbing all over the turret I had opportunity to examine the Turret Crash damage that I'd known it had suffered in the past. Seems to be just to the tin work, and a bit of tin-smithing should sort it out.

Likewise the swarf cover above the turret was somewhat rusty so got cleaned up.

All a bit of a set back - but these things happen - I suspect I'm going to have to get at the inside workings of the turret - just not sure how yet 

(* Hirth Index disk looks like a flat bevel gear with teeth facing forward, and mates with another identical one in any of N discrete locations dependant on tooth count)

[awemawson]

Well an extremely frustrating day today 

I managed, by re-wiring hydraulic solenoids, to unlock the turret, allowing me to turn it manually and set it in the correct position, thus being able to initialise it - hooray. I could then select tools at will, jog where I wanted, and spin the spindle at great speeds forward and reverse.

However, once turned off and back on I couldn't even jog, the control reporting a fault with the KTK Mentor. Now I've been through EVERY wire into and out of the Mentor, documented them and inspected any that might report a fault. None do, and the drive says it's ready on it's front panel.

BUT - if the power to the Mentor is dropped and remade by dropping out it's isolator the fault very occasionally clears

Once this fault is cleared I can jog as I wish. So two issues to bottom, the Mentor issue and the Turret issue - I did at least manage to winkle circuit diagrams for the Turret SMCC servo card out of Delta Tau in California today 

I think tomorrow I will log  the voltages on all 41 of the Mentors interface connections with the fault 'ON' and try to clear it and repeat the logging - hopefully that will show how the controller knows that there is a fault 

. . . meanwhile I'm going to finish off this bottle of Pinot, I'm sure things will look better after that 

[charadam]

Andrew, I'm curious.

Is this normal for machines of this vintage?

I mean, apprentices in a cold shed throwing components from 5 yards and installing them where they fall?

Or is the component layout and circuitry designed with no concept of maintenance or fault finding.

[awemawson]

You have to remember that this machine has sat un-powered and unused in two locations over the past 10 years. Things deteriorate, connections corrode and problems are bound to emerge as it is brought back into commission.

It's not untypical that one fixes something only to find another issue emerge. Hopefully as I progress and a few hours are put on the clock (which isn't working by the way !!!!) the reliability will increase.

[awemawson]

Another day going round in circles, but I've reached a tentative conclusion. But before I reveal that I took a break from diagnosis (to preserve sanity!) and did some metal bashing.

You may recall that when I pulled the turret apart, one of the complicated covers had suffered a collision with the tail stock when at the University. I was quite badly crumpled and unfortunately someone had welded up a crack on the corner while it was in the crumpled state. This of course makes a repair far more difficult.

I'd thought that with a good heating from my 'Rosebud' tip on the oxy-acetylene torch I could get most of the panel a dull red and tap in back with a hammer on the steel welding bench. No way  I did manage to get the folds 'pointing in the right direction' so I decided to use the 60 ton press to flatten it. Worked quite well. I may well cut off the welded corner, and weld in a new folded piece - but I can't do that until the rest of the tin work is back to give me something to measure to.

[awemawson]

Before the tin bashing I had come to the suspicion that the drive fault that is being reported is actually the FXM-3 Field Coil current driver - clue - it's little red light was not on and should be !

But before this I thought that the best approach was to try and decode the PLC 'ladder structure' to see what inputs lead to the generated error output. There is a software package that Siemens make available called 'Step 7' intended for generating PLC code and ladders but apparently also able to read the PLC code from an existing machine.

But no - you can't just down load it - you have to jump though all sorts of hoops and sign all sorts of disclaimers before you can down load it. The Siemens web site was a nightmare today - I had to reset my password five times as it kept locking my account. Eventually I spoke to a human being in Manchester who greased the path and expedited my 1.8 Gb download. That was finished at 13:08 today, and the self installing program has JUST finished seven and a half hours later - I don't have the energy to look at it yet  (But I only have a 21 day free trial !!)

So, I don't know why but I pressed the 'Spindle Jog' button on the controller, and the main spindle motor growled but didn't turn. Ah I though either the armature or the field isn't energised, but one of them is ! I rigged up my Fluke clamp ammeter and repeated the experiment on both the armature (shows 50 amps DC) and the Field coils - zilch - nowt, nothing. So that reinforces the fact the issue is quite probably the  FXM-3 - this is the little card below the KTK Mentor, on which that poor 0.1 uF capacitor committed Hari Kari the other day - so this is probably collateral damage.

So off to try and source a spare tomorrow

[RotarySMP]

How does a farmer get to be an experienced Mechatronic engineer? You are very good at this troubleshooting.
Mark

[awemawson]

Answer: he didn’t start life as a farmer!

Qualified in Applied Physics. Employed for years running a support organisation for Process Control computers. Hobby mechanical engineering. Retired early and bought a small farm.

Easy really 

[PekkaNF]

I did 10 years of S5 and S7 and always needed some support manauals! No matter if the printed probram was only 150 pages of ladder/blocks or 10x that, there were plenty os stuff.

Simple I/O signal is possible to debug without commented program, but when it has plenty of internocks and sequences, it will become easily convoluted. Many alarms and interlocks are grouped toget as flags and those would be nice to have commented.

Pekka

[awemawson]

Another jolly old day wandering aimlessly around the maze 

I've located a service exchange FXM-3 field coil drive card and agreed a price with the vendor but the weekend has got in the way.

Firstly, the Postman brought me a parcel from the USA - a spare Baldor SMCC card. This card is a servo card that drives the tool turret. I've been having problems re-initialising the turret and a working spare would help diagnosis - EXCEPT - the card delivered is not the card I bought ! I very carefully made sure that the vendors image exactly matched my card as there are apparently several variants. This one is missing an EEPROM, has an extra LED character display, and many option sockets that are not on mine.

As I understand it, when you initialise the turret by issuing an M80 command from the Sinumerik 820T is sends a load of data to the SMCC card which effectively is it's program. You then send a tool command to it that tells it which tool position is currently presented to synchronise both systems. This data that is sent I believe to be retained in the EEPROM.

With no documentation I'm not prepared to try it and a pained email has been sent to California asking pointed questions !

The turret position is read  by the system  by an internal Euchner four way proximity switch block, and on one occasion I did notice that one bit wasn't being read reliably, so I reset the gap and it now reads ok.

Under certain circumstances, and I've not discovered exactly what, the Turret will initialise, and once initialised seems fine for tool changes until you power off again. I then have to thrash about trying to re-initialise.

At one point, quite repeatably it was fine if the turret had to rotate clockwise, but failed if the required tool meant an anti-clockwise rotation. This lead to what was probably a red herring, the Simodrive AC servo system that drives the X,Y, and turret servo motors - more of that later.

Here are the two variants of SMCC that I now have

[awemawson]

Now the oddity about the tool change failing if it involved an anti-clockwise turret rotation sort of fitted in with my observation that if manually unlocked and rotated by hand it was more difficult in one direction than the other. There seemed to be a slight bias aiding turning anticlockwise. Maybe the servo null point needs adjusting.

So venturing into the Simodrive cabinet the aim was to identify card part numbers and google any setting up details that might be floating in the Ether.

Taking the front panel off you are faced with what at first looks to be a confusing mass of connectors and wires. As far as I can tell, there are two two channel servo cards for X, Y, T and an unused one, and three power amplifier boards. Centrally there is some sort of supervisory board, and below a Power Supply with a lethal pair of 275 volt DC buss bars.

Eventually I managed to extract one of the two channel servo cards for photography and identification. Unfortunately I can find no setting up data for this card -  loads for sale but no documentation.

There is a daughter card facing forwards with a DIP switch bank of options and three twiddle pots per channel - I bet on of those is the offset null !

My first port of call was the Siemens support web site - for the 7th time in two days I had to reset my password - they have some very odd fault on that site ! No data on the servo card though.

So I put it all back together, had another session thrashing about trying to find a sequence that would re-initialise the Turret system, and then darn me it happily did tool changes in both directions as long as I wanted - still wouldn't survive a power off / on cycle so I left it for the night !

[awemawson]

Another interesting journey this morning !

The Turret has seven proximity switches. Four in a Euchner block as described above, one for 'Clamped', one for 'Unclamped' and one for 'Zero Reference'

I had been able to prove all working except the 'Zero Reference' one so this morning, moving the turret as far as I could towards the chuck without tripping the over travel alarm, I could just see it right at the top back surface, pointing inwards towards the toothed dog clutch disk. I don't like sticking my head in places like this with the power on, as if the Z servo was to have a fault I could be pulped, but the power has to be on to unclamp the turret to rotate it to see the notch it is supposed to detect. Strangely this corresponds to somewhere between tool positions 7 & 8

With my 'scope on the output of the sensor as detailed in the circuit diagram, I rotated the turret back and forth either side of the notch. Permanent 24 volts - nothing detected. Just to be sure, I took the cover off the terminal box mounted on the turret, checked that the sensor had it's 0v and 24v supplies, and it's output was permanent 24 v. OK faulty sensor - unscrew it. Well it's right behind the turret. By lowering the X axis as far as possible it's possible to start unscrewing it, hindered by the clamp and unclamp sensors whose cables get in the way.

Eventually, it's out on the bench for testing. Rough and ready - bench supply set to 24 volts, meter on the output, spanner in and out of range. Works perfectly 

Actually, this is  good thing as these particular proximity sensors are far from cheap, but whats going on. Back to the terminal block - hang on still showing 24 volts as in detecting  Then back to the other end of the cable where it goes into the SMCC card - why's that wire a different colour ???

Turns out that the output of the sensor has been connected to the famous 'Terminal 100' - it's the extra green wire that caused confusion before. All very odd.

I can see that the software doesn't need a Zero Ref input when it already has a four bit tool count as the tool disk rotates - so is this an official mod, or did someone bodge it in the past - no way of knowing

Can't put it back just yet as I need to smarten up - the wife's playing a gig in a pub this afternoon, and it would be rude not to support her - oh just by chance it's a Harvey's  Beer pub - the only bitter worth drinking - not that I'm biased 

[vtsteam]

....
 Permanent 24 volts - nothing detected. Just to be sure, I took the cover off the terminal box mounted on the turret, checked that the sensor had it's 0v and 24v supplies, and it's output was permanent 24 v. OK faulty sensor - unscrew it. ....

Isn't that inverted compared to your test?

[vtsteam]

Is there a continuous magnetic field in the area? Something magnetized?

[PekkaNF]

I got the impression that this proximity switch has open collector output, they can be ORed, that (or other output) would burn totem pole output. Also I got impression that it is and inductive sensor, those are not that sensitive to magnetic fields (HAL is a different animal).

Pekka

[awemawson]

Well chaps what I SHOULD have done is disconnect the three wires and test it 'in situ' using my Proximity Sensor Tester, that I had completely forgotten that I had until a lull in the music this afternoon 

Would have saved so much time. When I put it back, even though it is obviously not in use, I'll use the tester to set it up so it can work should the need arise.

The wire connected to the output is held firmly at a nominal 24 volts DC - so no way the sensor will have any affect !

It's detecting (in the original concept I think) a notch in a very slow moving disk of steel, so I'd guess it's a simple induced magnetic one, but it bears no markings.

[awemawson]

....
 Permanent 24 volts - nothing detected. Just to be sure, I took the cover off the terminal box mounted on the turret, checked that the sensor had it's 0v and 24v supplies, and it's output was permanent 24 v. OK faulty sensor - unscrew it. ....

Isn't that inverted compared to your test?

i was turning the turret back and forth across the notch where we should go from detected to nothing and back to detected. In practise of course nothing was detected as the sensor output was nobbled !

[awemawson]

  Well developments 

I put the Index proximity sensor back. In order to start the thread on the device I had to unsheathe the other sensors to get the alignment right. Then  I set it up using the little test box, and then got to wondering about the odd wiring. A multi-core cable takes the four outputs from the  Euchner Tool Position Sensor directly to digital inputs. But it also provides 0 volts and 24 volts to the sensor and carries the Index transducer output to Terminal 100 on the Violet wire. Then from this terminal a green wire goes back to the SMCC card to it's Index input. Now this is illogical, as Terminal 100 is permanently at 24 volts when the system is 'up' due to the CR10 / CR20 'Volts Healthy' relays.

I reasoned that this is a cock up, and that the Violet Index output and the Green Index Input should have been joined using the adjacent terminal 99 so that a genuine Index signal is provided at one point in the rotation of the Turret Tool Disk due to the notch in the clutch disk in it's innards.

. . .so take your life in your hands and alter the wiring . . stop fussing   

Well I did, and do you know - since then the Turret has performed faultlessly 

Now actually this is very odd, as that Index signal is usually at 24 volts, and when the Turret is initialised it doesn't rotate the tool disk so the Index signal will remain at 24 volts just as it would when it was connected to Terminal 100

But let's not get carried away, it works. It just might be with all the dismantling and reassembly I have made good a poor connection, but, fingers crossed, the Turret does what it is supposed to do 

On the Spindle Motor Field Coil driver front, my chap has gone very quiet and is not answering texts or phone messages. I've actually ordered all the semiconductors on the board as a fall back solution, but would far rather have a known good service spare as had been offered. We'll see which arrives first. Certainly the component approach would be far less costly.

[awemawson]

So to celebrate I at last got round to changing the hydraulic solenoid 'Tell Tales' - originally there was a DIN 43650 socket fitted to each hydraulic solenoid valve, with a 24 volt 'Pea Bulb' to indicate the state of the valve.

Most of the pea bulbs had failed, and their heat had made the plastic of the sockets very fragile - I'd not been able to source replacement bulbs of a low enough wattage, and it turned out that the complete modern version with a red LED instead of a bulb was not much more than the non existent bulbs anyway.

There are ten hydraulic solenoid valves on this machine - handy as the tell tales come in five packs! So this afternoon I've installed the first five in the tail stock hydraulic cupboard.

Should make fault finding a bit easier - still five to go in the  chuck control cupboard.

[Pete.]

We have those on some of the demo robots at work. They ain't half handy when something doesn't move when you tell it to, cuts down the diagnostic time drastically.