Starting 3 phase electronics after long period?

Wow.

When I were a a lad I did things like that - nowadays I'd just shove a bezoar down it's throat [*], ie connect up a BOB and an old PC running emc2 or mach3.

Maybe feed a rotary converter, perhaps with a 3-phase resistive load in parallel, from a variac?

Feeding an inverter from a variac might be kind of all over the place ..

Don't think I'd do it one phase at a time tho'

-- Peter Fairbrother

[*] a Harry Potter reference

Hmmm, is this an AWEM-typical lathe, big enough to rout out the Grand Canyon if needed?

Are the servo drives 3-phase?

-- Peter Fairbrother

Well not quite that size - it was based on the castings of a Colchester Chipmaster, however the servo system is 'generously proportioned' - as far as I can see the GEM psu takes in all three phases, and the long and cross feed servo motors are the same SEM units that Bridgeport used on the Interacts. The spindle motor has been replaced by the biggest servo motor that I've seen.

AWEM

I dealt a lot with A-B on inverters, having purchased them in the 10's at a time. Their tech support guy said to just power old stored inverters for about two days before using them. That was enough time to re-form the capacitors or whatever. These were all three phase,

480V.

Pete Keillor

...and if you want to be really sure, disconnect any loads from the inverter output, then power up with incandescent lamps (eg: 2 x 40w in series) in series with the 3 phase lines. Even better is a 3 phase variac, but not so thick on the ground.

Have seen otherwise reformable electrolytics explode if the current isn't limited in this. The inverter types in large cans make a very big bang :-)...

Regards,

Chris

Yes I didn't think Pete had been given very good advice there, as it's the applied voltage to a depleted capacitor that does the damage. The anodising layer on the plates dissolves back into solution somehow, and reduces it's insulating properties. By bringing the voltage back up slowly the layer has time to re-form. If you just apply the full voltage it makes little difference if the load is on the inverter or not.

AWEM

I'm not sure if this idea makes sense so laugh if its just nonsense.

What about using the single phase variac on 'all' three of the phase inputs at the same time? i.e. strap the three input terminals together.

Obviously you want be able to run the load wired like that, but at least you can reform the electrolytics slowly.

Ian

That's a good point, since inverters generally have a 3 phase bridge feeding one or more caps in parallel. A single phase variac feeding across 2 of the

3 phases would probably be enough volts to reform the caps. If the inverter has a neatral terminal, then use one or more inputs with the neutral, though at half the volts.

Taking the cover off the inverter should reveal the bridge and cap connections on the input circuit, just to make sure...

Regards,

Chris.

Yes I'd given that serious consideration as the only true three phase load is a hydraulic pump that can be disconnected - the rest is electronics through rectifiers

AWEM

"Andrew Mawson" wrote in message

If you charge them slowly with a lab power supply you can see the leakage decrease as the layer reforms. Set it on constant current at maybe 10mA and watch the voltage rise slowly. When the cap reaches the voltage setting the current falls off.

On my power supply when I turn down the voltage a little the voltmeter shows the cap discharge rate. Some may not handle this reverse voltage and should have a series diode on the output to protect them, and a separate voltmeter on the capacitor.

Alternately you can disconnect the cap from its load and charge it through a high value resistor. Use a 1/2 or larger package because

1/4W resistors have a fairly low voltage limit due to their lead spacing.

jsw

I didn't have any problems with the A-B's. These were fairly new Powerflex 700's. May not have worked for others.

#

Yes - my stuff has quite possibly been powered off for ten or more years, which probably puts it in a different league. I need to get round to mapping out the servo and power electrics side of this lathe, - putting it off as accessibility is atrocious. Power supplies wired up and mounted down inside chambers where you can see them, but no way can you get at them without dismantling everything else. There is a whacking great three phase transformer posted into another tunnel so there is no way to get at it's terminals. You can be sure that the designer never had to fix anything. I need to adjust the wiring anyway, as it has both a three phase feed AND a single phase, which I think is very bad practise.

The GOOD thing is that today I got the Micon 850 controller talking both ways to a PC so I can download and upload G code, and the fact that the repair of the tape deck is at the moment defeating me thus doesn't really matter

AWEM

Thanks for all the useful suggestions. Today I took the plunge and sorted it. The main issue was a big power supply, with three phase input giving out a nominal 150 v DC at 100 amps. The reservoir capacitor that I was concerned about is 6000 uFD at 200v working according to the book, but the supply is not at all easy to isolate or access. I worked out that by removing the three GEC GEM servo drivers, I could get at the 150v DC busbars, and with croc clips attach a variable voltage current limited lab supply. The highest voltage one I had was 60v, so I attached that, with current limited to a few milliamps, and let the volts ramp up to 60 over about 5 minutes. I left it on for a few hours, then took the plunge, and connected the 415v three phase with the servo amps still removed. The DC supply went up 165V off load, and (thank goodness) there were no dramas!

Now I can consider re-integrating the controller into the lathe, and you never know, it might all work (and pig might fly)

AWEM

That deserves serious congrats for effort and tenacity. Not only have you analysed and fixed the electronics, but also the mechanics along the way.

Renaissance man would have approved :-)...

Regards,

Chris

Thanks Chris,

Today the controller was re-connected to the lathe, all be it on extender cables so I can monitor things, and now I can move in both X & Z axis and control the massive spindle motor with G&M codes.

Just need to sort the erratic hydraulic 12 station tool changer that is controlled by a sub-system based on a 6502. It's electronic construction leaves much to be desired being a mother board with daughter board on spacer pins, and relying on the bodies of capacitors to prevent the two boards touching! I've no documentation on it, and if it proves too time consuming I may re-engineer it. In essence reasonably simple - it takes a BCD value from either panel thumb switches or the main CNC controller, and drives four hydraulic spool valves (Clockwise, Anti-clockwise, Lock, and Unlock) and takes a coded position value from the tool turret to say where it is. Doesn't help it's unboxed and rather exposed to oil vapours in the lathe base - no doubt poor connections are the root of the problem.

AWEM

Greetings Andrew, I don't know how your turret codes the position but the turret on one of my lathes lost position last year. It started to position erratically and then totally lost position. The turret uses an optical encoder that outputs 8 unique codes, one for each position. The problem ended up being the coupler that connected the encoder to the turret. Even though the setscrews were hard to turn they were not tight against the shaft. I fixed that of course, timed the encoder position properly, and now everything works great. Maybe your turret use a similar system and could something be loose. Cheers, Eric

Eric,

I think something similar is very likely. The microswitch in my other post is used for lock / unlock detect, and the pair are operated by two disks, threaded onto a shaft and located by grub screw. One erroneously is free to rotate, so in the locked position the switch may or may not operate dependent on whether the disk has unscrewed. (Initially the fault was dependant on if the tool changer had approached position clockwise or anticlockwise) Currently tightening the grub screw is defeating me!!!! It's approx a 3.5mm thread down a 20mm long hole and someone has buggered the thread above it and also the hex socket. It's looking increasingly likely I'll have to drill it out in situ which I don't relish as it will be hard steel

AWEM

I hate it when that happens. Eric

OK it turns out that there were two faults: a/ the loose disk giving erratic locked signals, and b/ the phasing of the coded disk with the physical angular locations where the lock is able to drop in. The coding is remarkably simple with only two channels. One goes logic true when Tool 1 position is reached, the other changes state on each tool position. So the logic finds Tool 1 when it is first called to change tools, then counts from then on keeping track of where it is, until next powered down. The lock / unlock microswitches in my other post are infact fully functional but not normal switches but optical ones in a microswitch style housing a bit smaller than V3 but with V3 spaced mounting holes. RS could come up with a functional equivalent, but too large physically, so it's as well they are working! I blush to admit that I bodged the grub screw issue, as there was no way to remove it without drilling. I ground a fine screwdriver to approximately the size but a bit bigger, and tapped it in with a small hammer while turning. It worked but the next chap along will curse me.

AWEM