Friday, 31 March 2017

Inrush limiting

As noted below, I've gone for the old tech solution for my servo PSU. Simple but large (3kVA) mains transformer, bridge rectumfrier, 2 large bucket caps (10mF, 200V each), nothing much else. Which is fine but if I suddenly bang that little lot onto the mains, a large lump of voltage will simply disappear and it's fairly likely the circuit breaker will trip. Wouldn't do the power switch much good in the longer term either. 

So I need a soft start circuit that won't take me a month to construct or cause trouble with The Internal Controller, viz-a-viz the cost. Industrial motor / transformer soft starts are expensive and I didn't have one on the shelf last time I looked.

There are generally 2 issues with turn-on of transformers - the inrush surge due to the large DC bus capacitance (in this case) and the occasional transformer saturation caused by the remanence (residual magnetism) that was left in the transformer laminations at the instant the power was interrupted when the power was last turned off. Transformers generally run very high magnetisation flux swings (core losses aren't generally a problem), so periodically (and the timing is essentially random), you get a very large overload at turn-on if the initial flux swing added to the initial remanence exceeds the saturation flux. If the circuit breaker / fuse doesn't trip or blow, the flux will then settle down and everything will be fine again.

I've been here before. My massive (400kg) Miller-Interlas AC/DC TIG welder is basically a giant 50Hz switched mode power supply, complete with 2 massive mag amps and an even more massive output filter inductor. It's pretty cool that the only active (semiconductor) components are 4 stud diodes - that's got to be the simplest regulated SMPS you can make. Full load current from the mains  is about 50-60A with over 200A output. The current surge at turn-on when the transformer saturated periodically was too big even to be masked out by using a 63A Type D MCB, so I had to make a soft starter for it. 


There are generally 2 types of soft starter - phase control using triacs (the duty cycle starts at zero and gradually opens up to 100%) and a series precharge resistor that is switched out by the main contactor after a short delay. The phase control circuits are pretty complex if you are after a quick solution and of course the components are f expensive. Even for a single phase version, the power components wouldn't be cheap.

The switched resistor circuit sounds dead simple and in some ways it is. For the welder, I was able to simply delay the closing of the main contactor while a small-ish resistor got the flux sort of balanced around zero over a few cycles. But if the resistor is too big, you don't get enough flux balancing to reduce the remanence enough to avoid saturation when the contactor finally closes. And if it's too small, you don't get much of a soft start. So you have to figure out the sweet spot and nobody will tell you where that is - requires some experimentation. I used a Finder time delay relay (something like this) to delay the main contactor and a metal clad resistor across the contactor at power on. 

This sounds like the right approach. I'm sure I bought a couple of spares at the time but can I find them now....

Getting my head examined

Last night I thought I'd better investigate the pneumatics for the gear selection and drawbar engagement. And the mysterious impact driver on top of the head, along with its dedicated controller module.

Gear range selector:
The head has 2 speed ranges - direct drive from the driven belt pulley to the spindle - and a 2-stage gear reduction stage, rather like a conventional back gear. Unlike the classical Bridgeport head that uses a toothed belt and a gear reduction, this doesn't cause a reversal of direction when the reduction gear train is engaged. This is a section view of the AN-S but apart from the mechanical speed variator, it's 90% identical as far as I have discovered.

On the manual machines and the more commonplace AN-S models, there is a rotary knob on the side of the head that is turned either manually or by an external electric drive. For direct drive, the splined driven shaft connects to the spindle by means of a sliding dog clutch. For low ratio, the dog clutch / gear assembly is slid the other way and the gears are engaged.

On this machine there is a rotary pneumatic actuator to make the range selection. It clearly has a direct action (no internal reduction gears), although I haven't taken it apart to look closer. The actual position is reported back by means of 2 microswitches. There are two 24V pneumatic solenoids to drive the actuator in the correct direction. Simple enough, although currently I have no compressor...






Impact driver:
This machine was specified with the optional power drawbar / toolchanger. This is not the full "ATC" carousel automatic toolchanger option unfortunately but otherwise allows semi-automated tool changes.

The impact driver is literally a modified pistol-style Bosch hand tool with the handle removed. The 4 wire for the brushes and field winding are brought out - that's necessary in order to be able to control the direction of rotation. There is a pneumatic piston that pulls the driver down against a return spring, to engage it with the drawbar. There's no microswitch to tell the controller it has done so, so I assume there is a time delay to give it a chance before the impact driver starts up. I also assume the controller checks the spindle is stopped and the spindle brake has been applied.





I dismantled the impact driver, cleaned the patina off the commutator, lightly greased the bearings, rearranged the gear box grease and reassembled it. It seems to work fine and the brushes have lots of life left.

Drawbar controller:
The power drawbar / tool changer was automatically controlled. The system controller clearly told the drawbar system how and when to operate. The low voltage (24V) push buttons on the front of the head went back to the controller and came back up via 240V relays to a box on the side of the head containing a lot of discrete components including 2 relays, an SCR and a handful of transistors, pots and caps. It's all 240V stuff, so clearly the relays in the old cabinet were to interface between the low voltage system and this high voltage box.



There are 4 wires to the impact driver on the top of the drawbar, 2 going to the 240V pneumatic solenoid that forces the impact driver down to engage with the drawbar and 3 red wires that come from the main cabinet. These 3 red wires are 240V but appear to be signal rather than power connections, judging by the circuit on the PCBA. The actual power comes in as a L-N-E cable elsewhere, via a switch on the outside of the steel box. So it seems this PCB is told to engage the impact driver, then energise it in the correct direction, depending on whether it needs to tighten or loosen the toolholder. Question is - can I reuse this module? If not, I may have to come up with a cunning plan to control the sequence of operations require to bring about a tool change - I'd rather not.

At the limit

Had a good old look at the limit and dead stop switches the other night. I need to figure out what's already in place and if it will do for my needs. I may need to reconfigure them or run some additional wires etc.

For the X and Y axes, each has a rather proper looking 4-position limit switch assembly from Omron. These contain 4 microswitches, each with NO and NC contacts. There are corresponding ramps fitted to the underside of the tables to operate them. They are screwed in place, so if necessary I could move the ones that are there or add some more of my own. They look almost IP68 in terms of sealing, although the cable entry is not sealed in this machine. So the Y axis switch was full of oil that had dribbled in from above. This was most likely (neat) cutting oil. It's handy that the previous owner never used water-miscible coolant otherwise the switch would have died long ago.


In fact, they have only connected up 2 of the switches in each module. One is a NC "dead stop" switch wired into the e-stop circuit that kills the drives if the table ever reaches that far. Both ramps operate the same switch, so it doesn't matter which direction it's going in, although of course the controller should know. The other is a "limit" or home switch and presumably signals that the controller shouldn't think of going any further - and for homing. Again, the limit ramps at each end share the same switch.



So I pretty much have what I need already. I also have the old chestnut to contend with - do I want my home switch right at the very end of travel? That would require a lengthy and possibly awkward move right to the end of X and Y travel each time I home the machine. On the other hand, if you switch the machine off when the table is "past" the home switch, it will end up hitting the dead stop switch without encountering the home switch.

The whole Z axis assembly is enclosed within the head anyway, so no need for any fancy IP68 industrial switches. There are 3 microswitches at the top of travel and one at the bottom. I assume the top ones comprise (from the top) dead stop, home/limit and tool change positions, while the bottom is dead stop. I probably only need 3 of them - dead stop at top and bottom and home switch near the top.



Wednesday, 29 March 2017

Standalone controller

Yes, as you might have noticed in the previous post, I have acquired a stand-alone CNC controller for the retrofit. I got this on my recent travels in China. It fits nicely into the original console box and leaves a lot of space for the remaining system components. 

It's a 4-axis milling controller from Newkye in Shenzhen, model 990MDc.





It contains most of the functions required to drive the servo drivers and interface with the system components such as the spindle VFD, limit switches, tool change / drawbar controls, speed gear selection, coolant control, autolube etc. Unlike a LinuxCNC or Mach 4 / Mach 4 installation, no PC, monitor, breakout board, mouse etc are required. 

I'm not very knowledgeable about CNC controllers but I'm told it's almost a clone of the Fanuc 4M(?) but costs an order of magnitude less. The user manual contains some very strong Chinglish, so I spent a fair bit of my spare time in China digesting it and translating it into something more intelligible.

I plan to fit a new blank rack front plate with a suitable cutout for this and a few controls like the VFD Remote Digital Operator(!!) and the main isolator switch.

...and in with the new?

I have some CNCdrives servo drives (160V / 35A) that can drive the DC servo motors directly, given the correct step / direction signals from the controller. Each drive is small enough to fit in your hand, whereas the original 3-axis system had 2 large racks, each large enough to house a small dog. 

Similarly, the new Yaskawa VFD is small enough to pick up with one hand, while the old VFD and braking unit were each the size of large shoe boxes. 

All in all, the new system is small enough to fit into the original console box, so that's the plan.

First I modified the original support arm that held the original telly. It used to be mounted on the top of the cabinet but that's down the road now. This looks good:



Looks as if it should fit in the old telly box nicely:


The old VFD was stripped down and donated a large finned heatsink that can be reused for the servo drives. It pokes out the back of the console unit.


A little bit of work needed now to fit everything into the box and connect it all up.....

Turn it up!

The new servo drivers can handle 160Vdc with the braking module starting to operate at 180Vdc. The motors are rated at 150Vdc, so it would be wrong to underutilise them. Higher voltage basically allows faster movement which is all the rage these days - and indeed the sign of a real man. The 160V / 35A rating of the CNCdrives modules is a pretty good match to the motors. 

Having dumped the massive 3-phase transformer, I needed another source of power for the drives. 

A nice SMPS would have been an acceptable solution but I don't have one and didn't fancy coughing up hundreds of £ for one. So I took a 110V CTE (centre tapped earth) site transformer and removed some secondary turns to leave a 100V secondary. This will give me around 140Vdc off load. The "Wall mounted" version is not potted, so easy to dismantle and modify, unlike the yellow moulded site transformers with their potted internals.





Out with the old...

At this point, the system stopped working. It refused to display anything and above all wouldn't allow any form of movement. I got it to come up once briefly and then it was gone. Now when I opened the control cabinet or turned to the console, I was looking at a pile of scrap electronics. It's possible I could have reverse engineered a non-functional servo system without any documentation but even if I succeeded, the minute anything failed I'd be presented with another challenge, possibly one that couldn't be overcome. 

Out with the old:

So I ripped everything out of the cabinet, being careful to leave the motor wiring, switches, encoders etc in a fit state to reconnect later. Finally I removed the cabinet and its fixings. It's literally big enough to climb inside although not quite big enough to sling a hammock. Without all this stuff, it looks much simpler and a large floor area became clear. Possibly enough room for another machine tool? 




The original operators console contained the control computer system, keyboard, small CRT monitor and tape recorder (for saving and loading programs!). It's basically a 19" rack system about 12U high.




Fusion 360 post processor - create a post processor for Centroid with probing??

WTF? This must the the definitive rabbit hole. I've been aiming to attempt a modification to the existing post processor that is provid...