The circuit works on the bench at a (fixed!) threshold of around 58V. Problem with that is that the OVP threshold on the DMM Tech servo drives is lower than than. When I turn up the deceleration in the CNC controller, I get overvoltage shutdowns, usually on the Y axis which has the greater mass (both the table and the saddle).
As I can't easily adjust the threshold using hard wired zeners, this solution either works or it doesn't. In this case it doesn't, so I'll need to revert to a more complicated solution.
Here's the DC bus voltage captured during a series of rapid moves. The scope is AC coupled so I can resolve the voltage at a decent level.
The overvoltage transient reaches about 8V above steady state, which is a nominal 48V for the X and Y axes. So the overvoltage threshold must be somewhere above 56V. Naturally the DMM Tech documentation doesn't specify what their threshold is but I'm guessing around 60V from what I've seen.
What now then?
As I have only 15V and 27V zeners available, it's back to the TL431 and a pot.
This looks about right, with the pot chain sensing the bus voltage and the 431 protected against the full voltage by use of series zeners:
However, the clamp voltage is still dependent on the IGBT's gate threshold voltage, which makes a mockery of the 431. I should just stop trying to cut corners and do the job properly.....
The classical "braking module" concept uses a hysteretic controller to switch an IGBT on and off, to maintain a bus voltage within a window. What I am doing here is using an IGBT in a linear mode, to mimic a zener diode. So instead of implementing hysteresis, I'll use closed loop / negative feedback to regulate the bus voltage. This requires a PNP transistor to translate the 431 cathode signal to a gate voltage:
This looks about right, although the actual tuning of the error amp (the 431) may need to be optimised, as I can't easily determine the circuit parameters in my system. I can play with C1 if the thing is unstable. If I make it too big, I may start to see an overshoot.
In order to adjust the threshold, I'll make R8 a 100k pot. Varying its value between 100k and 10k (easy to change in the simulator - and 10k is close enough to zero) results in a range of ~44V to ~63V. That should do the trick.
Now build the bugger....
Found some stripboards and the required compts in my "collection". Off we go...
There. Not actually that complex and only took 20 mins or so to build.
And bizarrely, it actually worked first time out. I fitted a cardboard bezel and calibrated the knob using a DVM.
Then refitted it into the cabinet:
And bugger me, it actually does the business. I can now turn the feedrate up to over 10,000mm/min (400" per min). That should be good enough for now.
I may actually have to try out the combined plasma / CNC thing soon at this rate.
No comments:
Post a Comment