Retrofit Precix 3 Axis Gantry Mill

tommylight wrote:

But transformers do feed back from DC, for a very, very short period, when powered on and off.

When powering on / off you have a changing current, that can go to the other side of the transformer, yes. But changing current means it is not DC in that moment

Usually with toroidal transformers powering on is the critical thing. The iron core can retain magnetisation and if polarity when turning off is same polarity when turning on the transformer looks like a short-circuit for up to a half wave (10ms) and that is what trips the fuse. This effect is particularly strong with toroidal transformers. wikipedia has all the details: de.wikipedia.org/wiki/Einschalten_eines_Transformators

That short-circuit type situation can't happen when turning off, so I'm puzzled. 

Thx for your time first, i really appreciate your attention.

We have frequent inrush current  issues with "B" rated breakers inhouse. our house electricians refused to put slower breakers.
So tripping goes down from my Controllber P3 (10A C-rated) then my Lab breaker (16A B-Rated) and the first house breaker ( 16A -B-Rated). All together. No FI t(guess RCCB in english) tripping, of course - walk of shame when it tripps. This ratings define how fast breakers trip A-B-C-D (not sure if you have this in US?

Old Powerbox was 110/230 mostly with 20A Fuses no breakers, a schütz for Vacuum else nothing special.

For now i just run the motor driver without Mesa Fieldbus and Logic Power
Checked the old setup in it was almost the same, also a very standard SSR to switch, but two NTC's instead of the P1 Current Limiter.
DC currents  are super low as there is no torque on the motors.
Thanks for your thought on backfeeding the toroid from the motors/Caps, this should not be possible right?

@tommylight
On the test setup the ssr is controlled by the red emergency button. This failures were mostly driven by switching via the ssr. Might be an issue switching when driver is still enabled. On Linux enabling works fine as you described.
The machine just had homing switches no min/max switches. The servo loop worked fine so far, inhibit/fault seem also ok Z-Axis has a break, X,Y run out  when driver gets disabled. But thanks, about the mosfet info . So i should put relays in between PSU and driver not! controlled by LinuxCNC?

@rmu
Also thx for your comments. My understanding so far: The primary side of the toroid induces  a voltage peak when turning it off. Maybe together with a high current? it flows backwards through the breakers. But how when switched off? The SSR must be faulty or too slow. Or only half of the sine is switched, one mosfet broken? Ok the Voltage kills the ssr, the current tripps breakers. With a mechanic switch there should be no way to flow back. But i liked the ssr And what i understand from your comments the secondary side has nothing to do with these issues. That helps.
The "Einschaltstrombegrenzer"- nice German should not have any effect as it should be switched through at the time.
Thought of using a NTC instead (switches with heat), not sure.

The safety issues are an ongoing issue in our lab.
We mostly do 3D landscapes on prototyping foam aka Sika stock is up to 300mm. We were forced to get a new filter extraction. Even though we hardly catch the chips/dust at the tip as any hose close  would mostly hit the stock. Also the machine is open on all sides - an issue - we separated the mill with a thicker curtain (transparent plastic) from the operator - what i am testing now is steel wire around the machine connected to estop switch (picture). Nothing which complies with safety standards - we also have a small robot standing around which is even worse



 

ahess wrote:

So i should put relays in between PSU and driver not! controlled by LinuxCNC?

 

No, never put anything between drives and power supply.
Cutting power to power supply mains should be controlled by the "extreme limit switches" and the E-stop.
Drive enable should be controlled by LinuxCNC, but you have that sorted.
E-stop can also be wired to be controlled by LinuxCNC, but if you do implement this, the extreme switches take priority, always.

Would like to measure the voltage spike guess it must be huge.Thanks I will try this as I am still experimenting on my breadboard.

Re. the Einschaltstrombegrenzer -- You can talk to me in german too (probably better in the german subforum) as I'm a fellow austrian.

The magnetic flux in the transformer will try to keep any current flowing while it is collapsing, and can generate more or less unlimited voltages, so turning off inductive loads with SSRs that are not designed to do that can be problematic -- the stored energy in the magnetic field needs to be dumped somewhere.

The tripping of multiple breakers in a row is really suspicious. A B-type breaker takes at least 3-times the rated current for the magnetic disconnect, so 48A. To limit current, you could wire a light-bulb (100W incandescant) in series to the transformer, but I'm not sure if it will survive the shut-off surge so take care. Can you upload your schematics? Maybe the SSR really is the problem even when turning off. Try using a contactor ("Schütz") or for test purposes a manual switch or even bridge it.

Re. safety, maybe you can use a light-curtain ("Lichtgitter", ~500-1k€) in your e-stop loop (with "Schlüsselschalter" to bridge it) -- but those can be problematic if chips can disrupt the beams.

The Servo Power SSR was originally planned also to be controlled by LinuxCNC E-Stop. But as this inductive problems came up i thought maybe leave the toroid on for E-Stop cycles.
The big caps would run the motors for quite some while in case of mosfet shorting out?
So you mean just to add the extreme/min switches to the E-stop. right?