5i24 substitute using ethernet?

You would have to run the ethtool command by hand or reboot for the command to take effect, though the tmax times seem ok (assuming the CPU clock speed is 2 GHz or higher)

Still not sure whats going on, might be worth taking a look at the spindle analog out for a clue

The Dell 790 I have is an Intel Core i5 (2nd Gen) 2400 / 3.1 GHz

I'll get a multimeter on the spindle analogue out.

So on the analogue output of the 7i47s (with the 7i80) I get the following:

Voltage across the legs in all cases is 12.36v.
Wiper:
5000 RPM 5.06v
2000 RPM 2.07v
30 RPM 0.89v

By the way, do any components need to go into TP1 and TP2?
The board has connectors but is bare there.

No those are just test points

Nitram wrote:

So on the analogue output of the 7i47s (with the 7i80) I get the following:

Voltage across the legs in all cases is 12.36v.
Wiper:
5000 RPM 5.06v
2000 RPM 2.07v
30 RPM 0.89v

The last one is really far off unless you forgot a 0
It looks like the scaling is 1V ~= 1000 RPM so 0.89V ~= 890 rpm
Note that the 7I47S analog out will not swing all the way to ground
but only within 50 mv or so of ground which is going to make
selecting very low speeds difficult on a high speed spindle

Reinstalled the 5i24, and spindle behavior is normal. S30 M3 the spindle starts straight away (no lag) and this enables the tool change (orient cycle) to work correctly without timing out.

Retested the 4i47s analogue output with the 5i24 reinstalled.

Voltage across the legs in all cases is 12.30v. (Was 12.36v with the 7i80 installed)
Wiper:
5000 RPM 4.97v (was 5.06 with the 7i80)
2000 RPM 1.99v (was 2.07 with the 7i80)
30 RPM 0.033v (the floating decimal point is very subtle, so it is highly likely that the 7i80 reading was 0.089v)

25 RPM was the minimum achievable even when setting a lower RPM which is indicative of the PWM bottoming out at about 0.027v.

I am still unsure why with the 7i80 there should be a 4-5 second lag when commanding an S30 M3 (for a tool change), whereas the 5i24 is immediate rotation.

The PWM base clock is different on a Ethernet card vs a PCI card (200 MHz vs 100 Mhz)
but the driver takes this into account.

Can the above difference in frequency have an effect on PID tuning? Or the voltage scale value (per the .ini?)

It also still doesn't explain the smart serial timeout messages.

Rather stumped....

I would not expect the spindle speed command to have much repeatability
because its in the offset range of the digital potentiometer on the 7I47S
(that is the minimum output is in the range of you commanded speed = 30 mV for 30 RPM)

What PID tuning changed?

I did some tests with the bitfile I sent you and did not see any delay in
setting PWM outputs so I'm still not sure whats going on here

Smart serial timeouts usual mean a bad noise problem or real time issues with the host (or power supply issues)

I'll have to keep trying to nut this out. The spindle is definitely 4-5 seconds slower in getting going with the Ethernet than with the PCI.

I'll try a different power source to power the 7i80 and see what impact that has on things. I believe I'll take 5V power from the PC's power supply, which is where the PCI board is getting its power from.

I'm not sure whether this will have any effect whatsoever, but at this point I'm not sure why there should be a difference with all else being equal (aside from the PCI running Wheezy RTAI, vs the ethernet running Stretch and its current preempt kernel, in fact I'm using the same PC and swapping HDD's to test)...

Are you saying normal high speed spindle command are delayed?

I would expect trouble commanding 30 RPM because its down in the offset/noise
region of a 7I47S