[Grbl_Esp32 Issue#244] Spindle PWM invert option

#1 – bdring 于 2019-10-08

I’ll take a look at that later this week to see if it can be easily and cleanly added.

#2 – bdring 于 2019-10-11

I think I have this working. I am doing an overall cleanup of the spindle at the same time, so I need to do some more testing.

Previously, if you did not specify a spindle pin, you still needed to define a bunch of spindle values to get it to compile. Now, if you don’t specify the pin, the other values are not needed.

#3 – owenduffy 于 2019-10-11

Ok, sounds good Bart. Happy to test when you are ready.

Please see issue #247 which relates to spindle control and user customisation… it is a very simple enhancement probably to the same module if it fits with your ideas on user customisation.

#4 – bdring 于 2019-10-11

I saw that issue. I wanted a clean PR on the invert issue before I look at that other issue. There is also the #define ENABLEPIECEWISELINEARSPINDLE feature of AVR Grbl that was never implemented in GrblESP32.

How are you measuring spindle speed?

How well does it hold RPM under load?

#5 – owenduffy 于 2019-10-11

Measuring speed:

All tests are unloaded on a CNC6040 1.5kw 2pole induction motor VFD. The VFD controller displays VFD supply frequency. Induction motors have slip which is load dependent (it is the mechanism by which motor current increases to meet the load requirement), and I have verified the spindle speed unloaded reconciles with freq display using an optical tacho on the shaft (ie unloaded slip is very low as expected).

I would argue that unloaded speed is the target, as cutter burning on light cuts is related to near unloaded speed. In this context, I think you suffer loss of speed under load as an essential characteristic of the motor.

This particular VFD is not very stable unloaded at low rpm, and I expect even less so when loaded. There are practical limits to speed control.

I expect that slip at high speed is probably small, but at the lowest frequency / speed where winding reactance is lower relative to winding resistance, the slip curve is probably less ideal. Did I say there are practical limits to speed control?

The exit idea I proposed does not limit the compensation to a piece wise one, though I saw the warning to keep the code path length short.

#6 – bdring 于 2019-10-11

The changes have been uploaded to the devt branch. I will leave them there for a while before merging with master.

Any testing you can do or feedback you have is appreciated.

#7 – owenduffy 于 2019-10-11

I should be able to do that today (UTC+11).

#8 – owenduffy 于 2019-10-11

I have tested it initially with a scope on the output.

Verified that it is inverted, but there is an issue at speeds below $31 speed in that I expected PWM of 0% (inverted) but it is providing 1.56µs (I assume PWM=1/256) and we really want PWM=0 when speed is below S31 don’t we?

If I set speed S2000, I get 10.15µs (inverted fine), and them M2 and PWM=770ns, we need 0ns to turn the motor off.

Have I been clear?

#9 – bdring 于 2019-10-11

I didn’t test that. I never liked the way that works. It seems like it is not very robust.

Is the purpose of that to make sure you run the spindle at a minimum RPM or a minimum PWM?

I think it should just change any non zero Sxxx below your minimum to the minimum and send it through the same calcs as any other S value. Otherwise, unless SPINDLEPWMMIN_VALUE is perfect you have am odd transition from below minimum to the minimum.

Also

I think S should be tested and fixed in gcode.cpp. If you you send S100 it would change it to S200 (if that is the min). Now the $G reporting and overrides, etc work on the right value.

#10 – owenduffy 于 2019-10-11

Ok, I think the system needs to send PWM 0% (non-inverted or inverted as per config) whenever the motor is stopped, program is stopped, or Sxxxx

#11 – owenduffy 于 2019-10-11

I might explain that the reason I think the spindle should be stopped for sxxxx

#12 – bdring 于 2019-10-11

I am not sure how it saves the edge finder if you set the RPM too high.

I suppose if you are really worried about all this stuff, you will set it up correctly. Maybe I’ll mostly leave it alone.

How about this…

Example:
settings.rpm_min = 200
settings.rpm_max = 1000

S0 = 0 PWM
S100 = SPINDLEPWMMIN_VALUE
S200 = 200 RPM
S1000 = 1000 RPM
S1200 = 1000 RPM (and reports 1000)

#13 – owenduffy 于 2019-10-11

Let me try to say it more clearly.

The system should send PWM=0% (inverted or not as configured) for:
-Sxxx

#14 – owenduffy 于 2019-10-11

>
>
> I am not sure how it saves the edge finder if you set the RPM too high.
>
> I suppose if you are really worried about all this stuff, you will set it up correctly. Maybe I’ll mostly leave it alone.
>
> How about this…
>
> Example:
> settings.rpm_min = 200
> settings.rpm_max = 1000
>
> S0 = 0 PWM
> S100 = SPINDLEPWMMIN_VALUE
> S200 = 200 RPM
> S1000 = 1000 RPM
> S1200 = 1000 RPM (and reports 1000)

S100 = 0 PWM

Reasoning:
do not spin a tool faster than requested, it may be unsafe for the tool, work, operator.
do not spin the motor slower than rpm_min (unless you stop it).

#15 – bdring 于 2019-10-11

I disagree.

Reasoning:

Never accept an RPM but not spin.

Here is “original” Grbl.

“
else if (rpm <= settings.rpm_min) { if (rpm == 0.0) { // S0 disables spindle sys.spindle_speed = 0.0; pwmvalue = SPINDLEPWMOFFVALUE;
} else { // Set minimum PWM output
sys.spindlespeed = settings.rpmmin;
pwmvalue = SPINDLEPWMMINVALUE;
}
“

The alternative is error, stop job, require reset.

#16 – owenduffy 于 2019-10-12

Ok, I understand and accept.

I am about to do a cal run to feed the py script for piecewise linearisation. I am running it on the devt version, and a dry run looks good. The optical tacho will log the actual spindle speed (unloaded).

#17 – bdring 于 2019-10-12

Great.

Do you think a SPINDLERPMRANGE_ERROR should be added?

#18 – owenduffy 于 2019-10-12

I don’t understand.

The cal run looked good with the devt branch. Iie I am using the inversion.)
!line_fit

#19 – owenduffy 于 2019-10-12

I have been going over this, and I think there remains an issue that when the spindle should be stopped, there is a very short (790ns) down pulse in each PWM period.

I have found that if I set sxxx>0 and analogWrite(SPINDLEPWMCHANNEL, (1<PWMBITPRECISION) – 1);
grblanalogWrite(SPINDLEPWMCHANNEL, (1<PWMBITPRECISION));