Trouble at Mill, Bridgeport Interact 2

[beery]

Hi All,
This is the ongoing saga of the repair and maintenance of a Bridgeport Interact 2 Series II, fitted with a Heidenhain TNC150 control system. I could have named the thread 'A Bridgeport too Far'.

My day to day work is designing electronics full time (lighting controls). We have our own manufacuting facilities and often get called upon to use my repair skills to fix factory machinery such as SMT placement, pasting machine, AOI machines etc.
Some of the machinery can be classed as vintage and this is the case in the metalwork dept.
As you migh guess, most faults are PSU based due to dried out caps. But the Bridgeport has been quite a bit more involved than that and thought it might make for interesting reading.

The mill repair came off the back of finally fixing the CNC turrett press (I plan to write up that saga later).

The CNC mill had been in bits in the workshop for about 6 years or so (longer than I have been in my present workplace) and indeed the person who first took it apart had left not long after that. The Mill had been fitted with a 'Spare' servo motor for the quil drive (Z axis) which had the encoder cover left open. When fixing the turret press, I pinched that motor as the mill was basically scrap anyway at that point.

However, once the turrett press was happily working again, attention turned back to the mill. Could I put it back together and make it work?
I like a challenge, but it took a while to get permission to spend some time on it. BTW, it is working again, for now!

More to follow...

Cheers
Andy

but it took a while to get permission to spend some time on it

That sort of thing gets my goat, someone willing and able to do a job is hampered by sheer bureaucracy (the love of inventing, filling in, and filing of forms). I do particularily like the sign on machines "Only for use by authorised personnel", I just use them and when asked "who authorised you?" reply "me", there is never a list on file!

[barrymagrec]

Long long ago we were involved with making cable harnesses for Bridgeport - thick, thin, multiway, flexible conduit.

May still have the odd connector about if you get stuck - no idea if it was this model of course.

[Station X]

Originally Posted by merlinmaxwell

That sort of thing gets my goat, someone willing and able to do a job is hampered by sheer bureaucracy (the love of inventing, filling in, and filing of forms). I do particularily like the sign on machines "Only for use by authorised personnel", I just use them and when asked "who authorised you?" reply "me", there is never a list on file!

Please let's not turn this thread into a discussion about bureaucracy.

-----------------------------------------------

Andy/beery as a dabbler in CNC machines myself I'll be following his thread with interest.

[beery]

Hi all,
This is indeed not a complaint about bureaucracy. I'm just setting the scene. The factory in not even in the same town as my office, so considering that they had managed without the mill, it was hardly a priority. However, as we shall see, events eventually conspired to make fixing the mill a priority in the end.

Back to where I left off.

Our Metalwork section is focused on punching and folding aluminium sheet. Milling and turning are for one off jobs like production jigs etc.

The Metalwork manager had found the mill not working and attempted to power it up after re-fitting the quil motor. Upon hitting the CNC motion enable button, there was a loud bang and then the machine went totally dead. He found that a fuse had gone. Replacing the fuse got the the CNC control system to power up again, but he did not push the motion enable button this time. The machine was then turned off and then left for another 5 years before I looked at it.
That is how it was when I found it. Where to start?

More to follow...

Cheers
Andy

[mark_in_manc]

I look forward to this! I was defeated second time by the ancient Anilam 'Wizard' DRO (just read out, no CNC) on our works Bridgeport (first time was a PSU fault which I found using some confidence picked up on this forum and a lamp limiter, likewise!). Second time faulty it came on, but did nothing sensible; and being packed full of chips my heart sunk and we bought a new DRO. So lead on!

(If anyone needs a broken ancient Anilam and 2 glass scales for a standard manual Bridgeport, FOC, PM me, soon ).

Please let's not turn this thread into a discussion about bureaucracy.

Oops, sorry. Best to stick to the machines.

[Lucien Nunes]

Will read with interest. We had a very similar machine with exactly the same NC. Never gave us any trouble until I changed the backup battery and didn't lock the cover in place properly, so when the power went off the configuration parameters were lost. I think about the most technical intervention I had to give it was to clean the head on the microcassette data recorder.

[beery]

Hi All,
I'll start with a basic explanation of how CNC machines worked in the 1980s and 90s.
The mill has a conventional 3 phase AC motor to drive the milling spindle, albeit through a rather complicated electro pneumatic reduction drive.
The X, Y and Z axis are driven by DC servo motors which are fitted with tacho generators and position encoders. The servo motors are rated at up to 180V DC.
Note that most CNC machines are based around 115V 3 phase AC working often using a 3 phase step down transformer. Single phase peripherals such as small motors and power supplies usually operate from 115V AC. Relays and contactors as well as switch inputs to the control system operate from 24V DC.

The machine has a control system in the form of a custom computer which has an operating system similar to a PLC, a data entry panel and a display unit.

The logical operation of start and stop switches, limit switches and relays is dealt with by a PLC. Sometimes this PLC is incorporated into the main control system, as is the case with this particular mill.

Position measurement is achieved by using incremental optical encoders built on to the end of the motors. These encoders also provided a reference pulse once per revolution. The encoder outputs from the motors are connected to the control system. The control system for this machine is a Heidenhain TNC150. As with other Heidenhain control systems of the time, the outputs of the optical position encoders are analogue sine waves (taken directly from the photo diodes). These sine waves are amplified and converted to digital pulses within the control system.
Because the encoders are of the incremental rather than the absolute type, the machine has to reference itself against fixed reference points whenever it is first switched on.

Axis motor drive is achieved thus. The control system has 3 analogue outputs, X, Y and Z. These typically have a range of +-15V. A positive voltage drives a motor clockwise and a negative voltage drives it counter clockwise. An output of 0V means that an axis motor should be stopped. These analogue voltages go into servo drive modules, these are separate from the control system and often made by a different company. Servo drives may be by PWM, full transistor bridge or in this case by thyristors. The servo drive units attempt to match the analogue control input +-15V, with the voltages being generated by the motor tachos, to keep the motor speed constant irrespective of load.

Machine parameters tell the control system the limits of travel and speed etc. These are stored in volatile memory, with battery backup. Failure of the backup battery and loss of the parameters renders the machine useless.

Now, about capabilities. The Heidenhain TNC150 control system was introduced in the late 1980s. It was a considerable improvement on what went before because it had a proper VDU screen enabling the user to see a lot more lines of program on the screen at any one time. The big limitation it does have though is that it can't interpolate 3 axis simultaneously, so you could not for instance mill a perfect dome shape from radial coordinates. It does however have a special function for milling a helix. The lack of true 3D movement means that 3D cad drawings can't directly be translated into code that the mill can understand. The TNC150 also can't be drip fed commands and has to download whole programs at a time via a serial port either from a PC or a special cassette tape drive. This means that programs are limited to 1000 lines and more complicated machining operations may require more than one program. None the less, the ability to interpolate two axis movements and thus mill perfect circles and circular pockets was a big thing at the time and still very usefull today.

Other Bridgeport Interact 2 mills that I have seen had been upgraded to TNC151 or TNC155 controls, which allow full 3D milling, the ability to accept code translated from G code (the same as used in 3D printing) and the ability to execute commands drip fed from a PC via its serial port (this enables almost limitlessly long CNC programs).

With that over, I'll get back to the story in the next part...

Cheers
Andy

[McMurdo]

Wedgwood pottery had one the same in the mould making department. I went to it a couple of times due to the AA batteries failing (accessed through the front panel). With the power off the parameters are lost and the machine wont work til they're reprogrammed.
In wedgwood's case I obtained the figures from bridgeport on a fax and I had to type them in by hand. The department is now a housing estate but I still have the figures somewhere I'm sure.

[beery]

Hi all,
Here we go then...

A slight correction to my introduction is that the TNC150 control system actually dates from the mid 1980s.

Anyway, when I first found the milling machine it was in bits with two quill motors (Z axis), one of which was a spare, or was it? Both of these motors had the covers removed from the rear end, exposing the position encoders.
I was at this time concentrating on how to fix the CNC turret press, which being a sheet metal workshop was the more pressing task (see what I did there!).

When it became apparent that a replacement servo motor was required for the turret press, attention turned to the broken CNC mill what had been in bits for over 5 years. Aha, a good source of motors I thought!

Looking at the two quill motors, it soon became apparent that they were not from the same machine. One of them actually came from larger Bridgeport. This motor was in better condition and was a better fit to replace the Baldor motor on the CNC press. It was missing the box and lid that covered the connection terminals and we also needed some UNF bolts to mount the motor to the press as the original motor had metric fittings. So the broken quill motor that was fitted to the Bridgeport was removed and it donated its connection box and mounting screws to the turret press.

After the turret press was fully operational again, attention turned to the mill. Could it be made to work with the bolts we had left?

The metalwork manager had been promoted up the ranks and when he became the manager he decided to look at the mill. However no one was left that knew why it had been taken to bits. He found that it wouldn't work with the quill motor that was fitted to it as it seemed that the encoder did not work (a fault that was to come back later). He re-fitted the original motor, but there was a bang when he powered it up. The machine had blown a fuse.

I took the broken motor apart and soon realised that we were not the fist to dismantle it. Parts of the encoder were not fitted in the right order and one of the tacho generator brushes was broken. The actual fault with the motor turned out to be an insulation breakdown between a thermal overload sensor and one of the motor's brush holders.

The first photo shows the quill motor (the black one on the left) after repair, so you can see where it lives.

The second photo shows the bimetallic temperature sensor. If you look carefully you can see where the brass mounting clip has melted at the centre of the left-hand edge due to arcing.

More to follow...

Cheers
Andy

[beery]

Hi All,
The motor repair continued.
I replaced the bimetalic sensor with a new one and wrapped it in Nomex paper before fitting it in place next to one of the brush holders.
New bearings were fitted and the main part of the motor was reassembled.
It was found that one of the four main brushes was slightly different and much shorter than the rest. I could not find any 3/8" x 1/4" brushes (original Bridgeport ones go for silly money), so I filed down some 10 x 6mm ones to fit.
I powered the motor from my 150V bench supply and it ran very smoothly.

Attention now turned to the tacho generator. Remember that one of its small brushes was snapped? Worse still, I discovered that one of the tacho generator's armature windings had been damaged when the motor had been dismantled in the past. The winding showed quite a bit of damage to its insulation and there was a break in the winding at one point. Because of the way the windings overlap each other, I could not reach the end of the break. I decided instead to break the winding again, a few turns away from the original break. By joining the two breaks together I was able to save most of the winding by bypassing just a few turns which would now be out of the circuit. The repair was sealed with electrical varnish.
I could not find a brush near to the size of the missing tacho brush, so I found a much bigger one and filed it down to fit.

I built the tacho back up inside the motor. I ran the motor up very slowly and observed the tacho output on a scope to see how much of an effect the missing turns on one winding would have. Fortunately not much effect at all

The encoder was built up on the end of the motor. The metalwork manager made up a new cover for the motor terminals and painted it black. I found some 2BA screws to hold the cover in place and the motor was then installed back on the mill using four new UNF bolts.

The mill was powered up and, you guessed it, the parameters were missing . Fresh AA batteries were fitted and then fortunately the parameters were found on the PC that was connected to the mill. After downloading the parameters to the mill, the mill sprang into life and was seemingly working perfectly, for a while...

The first photo shows the tacho generator armature.

The second photo shows a close up. The bottom arrow points to the area of damaged insulation, you can see the copper colour showing through. The top arrow points down to the solder blob where I joined to the two breaks together.

The third photo shows the carbon brush that I had to file down to fit the tacho. I also had to cut most of the copper connecting braid off to make it fit.

The fourth photo shows the tacho installed at the back of the motor.

The fifth photo shows the motor on test on the bench. The motor is made by SEM, Small Electric Motors of London. Strowger fans may recognise that this company also made the tone generating ringing machines for the old electromechanical telephone exchanges.

Cheers
Andy

[beery]

Hi all,
time for another installment of the Bridgeport saga.

After working for a short while, the mill started to play up. Once up and running it seemed ok, but sometimes on power up it would fail to find its reference on the Z axis.
Referencing works thus:-
If the reference switch is closed it will move the axis motor until the switch is open.
With the reference switch in the open position, it will move the axis motor until the switch closes and continue until it sees a reference pulse. This pulse occurs once per revolution of the axis motor. The reference pulse enables the control system to set the axis measurement value to a the value that is stored in its paramters as the position for that particular axis reference. Thereafter all measurements on that axis are relaive to the reference position (remember that the encoder is simply an incremental type).
If the machine does not see the reference pulse then the axis motor will keep turing until the extreme limit stop is reached, this shuts of the power to all the motors.

This is what was happening, the motor was indeed going past the reference point and tripping that axis limit stop switch. Remember that the motor on the mill was found with the encoder cover removed? This fault must have come up before.
I had a look at the reference switch. This is a roller lever change over switch, a bit like a giant micro switch. The screws for the switch's adjustment bracket had damaged heads, someone had been here before too. Both the normally open and the normally closed contacts are used, so although the control system knew which way to turn the motor in order to reference, it seemed possible that the other contact may have been needed for it to see a reference pulse.
Going into the dianostics mode enabled me to see the digital inputs to the control system. All looked ok until I enabled the motor drives, at which point all sorts of strange flashing of random numbers apeared on the screen. The PSU of the TNC150 control system was also making loud screaming noises at this point.

Out came the TNC150 control unit. I decided to re-cap the PSU. It is an unusual design with an isolating step down transformer being fed with 115VAC which was then stepped down and rectified to produce about 40V DC.
The 40V was then fed into the flyback switched mode supply to produce +/- 15V DC. A buck regulator provided a high current 5V rail for the logic.

After re-capping the PSU, the login input diagnostic screens looked as they should, even with the motors enabled.

The machine now worked ok. But was it actually fixed? Would it continue to work?

To be continued...

Cheers
Andy

[Red to black]

Hi I have just been working on numerically controlled servos with X, Y, and Z values, not related to your application though (think much larger), we had a to have a datum point/marker sensor (or home value) for each axis to initialise the values stored within the HMI and PLC systems to suit our digitally scaled inputs.

Ps. ours used resolvers as the positional numerical counter.

[beery]

Hi all,
At last, I have a moment to write up the final parts of the story...
As I hinted in my last instalment, it was not long before the mill stopped working again.
It was indeed again the problem of not finding the Z axis reference point. It seemed to be intermittent and dependant on the ambient temperature when the machine is switched on. If you recall, when I first found the machine the cover had been removed from the encoder at the back of the motor.

Now the Z axis is the first to reference, this is to make sure that the head is in the raised position before the X and Y axis move, in order to reduce the chance of the milling cutter from crashing into a job that has been left on the machine. As the Z axis would not reference, I could learn nothing more.

By editing the machine's CNC parameters, I could change the reference order so that Z would be last. This enabled me to study in detail how it worked.
I also found that although when not referenced, the X Y and Z position numbers on the display were frozen, if I stepped through the display options I could get to the raw relative encoder data. Here I could see that the X and Y encoder inputs were active even before referencing and responded to the movement of the referencing process. The encoder on the Z axis however, was not being read.
By swapping over the encoder inputs to the control system and then attempting to reference the machine, I could see that the encoder on the Z axis motor was working ok and that it was the input on the TNC150 control system that was at fault.

Now as a slight aside, I had come across a mill with the same problem some years before. This was a Bridgeport Interact 412. This machine was fitted with a much later TNC355 control system. Unlike the TNC150 which is more or less impossible to hook up to a scope when it is in a machine, the TNC355 is mounted remotely from the CNC controls, that coupled with it all being on one big PCB meant that I could scope the incoming encoder signals. I diagnosed that the axis fault (I think it was the X axis this on that machine) was due to a problem with the encoder input on the control system. The control system on that machine had been fitted out for an optional 4th axis (such as a dividing head etc.). I found that by editing the machine parameters, I could swap the encoder inputs over. So I connected the X axis encoder into the spare 4th axis input. All that I had to do then was wire a link wire from the X reference switch input to the 4th axis reference switch input and all was fine

Now back to the TNC150, could I do the same trick? Unfortunately the parameters for the TNC150 don't have that option

So, I removed the TNC150 control system again and removed the analogue board for inspection. The control system had a Bridgeport repair date of 1996 on it. Examination of the analogue board revealed that 3 thick film units had been replaced. These corresponded to the encoder inputs for one of the axis that was working ok. Inspecting the solder joints, I could not find any evidence of any other repair work on the analogue board.
So I figured that these thick film units were likely the source of the trouble. Cold checks by the way found nothing by way of difference between good and bad encoder inputs.

Now a note about what is inside the encoder unit at the back of each motor. A light bulb (later units had an IR LED) shines through a copper disc that has slots etched in it. Four photodiodes, of which only three are used, pick up light that passes through the slots. Two of the photodiodes provide the fine position pulses, the pulses being 90 degrees out of phase with each other, just as in the mechanical encoders we often find in user controls in domestic appliances these days. The third photodiode provides a reference pulse once per revolution.
Only I was not quite right in what I said there, the photo diodes to not give out pulses, but give out pure sine waves instead. Also the positive and negative connections to each photo diode connect to their own wires in a multi-way cable that goes all the way to the control system. It is hard to scope these signals because the cable although plugged onto the encoder PCB goes all the way from the back of the motor to the control system with the only external connector being the one a the control system end, that connector making use of crimped inserts. Removing the cover from the back of the motor lets daylight into the encoder which messes it up...

I've included a photo of the encoder at the back of the motor and the pin out of the encoder cable.

To be continued...

[beery]

Hi all,
Here we go again, the final part this time. I know it has been a long read!

Back to where the encoder signals go...

In the control box the sine wave encoder inputs first go through some lovely yellow potted ferrite beads and then on to an op amp which I presume is simply used as a buffer (don't quote me on that). Then they pass to the thick film units. These units house an LM339 comparator that compares the inputs against a reference and provides nice digital square waves. The outputs of the comparators that are from the two phases of the position encoder go to logic gates that determine the direction of movement. The reference pulse thick film unit is very slightly different. I've provided a very rough sketch of the thick film unit circuit. In reality there are some extra, printed circuit capacitances on the inputs, but I've left these off. The 15K resistors actually measure 14K9 and consistently so.
The original style thick film units use a through hole style LM339 with its legs cut short and surface mount soldered on their remaining edges. This might have been the cause of the trouble perhaps. The thick film units conduct heat very well providing a uniform temperature across the whole unit. Maybe thermal cycles were stressing the solder joints on the LM339 or even stressing the chip's internal bonding wires. It is of note that the thick film units that had been replaced had proper SMT comparators fitted to them.

Well where to get replacement parts? There was indeed a spare axis and I could of tried the thick film units from there, but they were the same old style ones. Working TNC150 control systems are very expensive and strangely (not!) whenever someone is breaking one for parts they never sell the analogue board
Then I had an idea. I read somewhere in the Heidenhain documentation on encoders that if the motor was a very long way from the control system then inline analogue to digital converters were required (one for each axis) together with a special digital input version of the analogue board. As it turns out, second-hand inline converters are not to expensive at all and they contain the required thick film units.

I very carefully removed the thick film units. I did not solder them directly to the analogue board though, opting instead to mount them in turned pin sockets.
Then the machine went back together and hey presto it has been giving reliable service ever since

The first photo shows one of the later thick film units.

The second photo shows the circuit of the thick film unit. It is not very clear. I can draw it up properly in Altium if anybody want me to.

The third photo shows the PCB from the donor inline A to D converter unit.

The fourth photo shows the repaired analogue PCB.

At some point I will have to replace the capacitors on the servo drives, but that is a job for the future. In the meantime the surface grinder has failed and I'm currently working on a fix (no electronics in that one).

Next I'll have to write up the overhaul of the turrett press, which was of course my way in to fixing the CNC milling machine in the first place.

Cheers
Andy

[cliff2903]

Thoroughly enjoyed this, the Z axis encoder fault seems quite common on these and more so on the later machines where replacement is the really the only option. I've used a few of these that have been shown the door after failures (servo drives and power supplies mainly). Good machines though but with £200 call out and£90hr (if you can find anyone willing to take it on) lack of parts and given their age it doesn't take much for them to become beyond economic repair

[cmjones01]

Thank you for posting this, and well done on a successful repair to a valuable machine. It's interesting to see a branch of electronics different to the things I see every day.

Chris

[Station X]

Interesting and enjoyable read. Not the usual run of the mill stuff you see in the forums.

[Radio Wrangler]

Originally Posted by beery

Hi all,

The original style thick film units use a through hole style LM339 with its legs cut short and surface mount soldered on their remaining edges. This might have been the cause of the trouble perhaps. The thick film units conduct heat very well providing a uniform temperature across the whole unit. Maybe thermal cycles were stressing the solder joints on the LM339 or even stressing the chip's internal bonding wires.

In the meantime the surface grinder has failed and I'm currently working on a fix (no electronics in that one).

Sometimes, when the legs of ICs and transistors are sheared off, the shearing process can create a pressure transient shock-wave which travels into the device up the leg and does damage resulting in later failure.

That sharp snap when sidecutters cut a leg is actually quite risky. Trimming the legs of a device once it's soldered in a through-hole board is OK because the solder damps the transient.

Is that grinder a nice Jones and Shipman? I spent a lot of time with their internal, external and surface machines at Rolls.

David