Fanuc Analog Spindle Commissioning: Enable Chain 8133#5/3716#0/3717, 3730/3731 Voltage Calibration, and JA40/JA41 Wiring

Bottom line: when an analog spindle will not behave, the cause is almost always one of three things – 3730 (gain) and 3731 (offset) applied the wrong way round, 8133#5 left at 0 so the control keeps looking for a serial spindle, or the A/B phases on JA41 reversed so the one-rotation direction is inverted. None of these produces a clean alarm. They show up as a speed that is off by a fixed ratio, a slow drift at low speed, or SP742 the moment you try to rigid tap.

A VFD-driven spindle (usually called an analog spindle) is a half-open arrangement on a Fanuc control. The CNC outputs a 0-10V reference and a handful of discrete signals; the actual speed loop, acceleration ramps and orientation all live in the drive. That is why the method is completely different from a serial spindle: with a serial spindle you tune the CNC side; with an analog spindle you align the two sides.

1. Know the capability boundary first

Before choosing this route, look at the table. The line between what an analog spindle can and cannot do is sharp, and most failed projects picked the wrong architecture at the start.

Function	Supported on analog spindle	Precondition
Forward/reverse and S-command speed control	Yes	Basic function, needs only the JA40 voltage output
Spindle speed display	Yes	Position coder on JA41 plus a correct pulse count in 3720
Spindle orientation (M19)	Yes	The drive itself must support orientation and the JA41 coder must be healthy
Rigid tapping (M29)	Yes	Drive must support a rigid tapping mode, G61.0 must be routed through the PMC, and the coder line count must be sufficient
Cs contour control	No	Requires a serial spindle
More than one analog spindle	No	One control has only one analog output channel

The hard limit: Fanuc provides one analog output. A machine can therefore have at most one analog spindle. If there are two spindles, the second must be serial. Separately, the JA41 port has a maximum input frequency of roughly 256 kHz, which caps the product of encoder line count and maximum speed – see the worked figures in section 5.

2. Wiring: three interfaces, do not cross them

Interface	Purpose	Key pins	Common mistake
JA40	0-10V (or +/-10V) reference to the drive frequency input	SWC (pin 7), ES (pin 5), ENB1 (8), ENB2 (9)	Polarity reversed on a unipolar hookup so the drive never changes speed; ENB1/ENB2 miswired
JA41	Spindle position coder input (differential A/B/Z)	PA(5) *PA(6) PB(7) *PB(8) PZ(15) *PZ(17) +5V(9/18/20) 0V(12/14/16)	A and B swapped so the displayed direction is inverted and tapping trips SP742; shield not single-point grounded
JD1A/JD51A	I/O Link for discrete signals (forward, reverse, zero speed, alarm, orientation done)	Per the I/O unit	Zero-speed and speed-arrival not routed back to F45.1 / F45.3, so the control believes the spindle never turns

On polarity, JA40 has two legitimate uses:

• Unipolar: output is 0 to +10V only; direction comes from the drive’s forward/reverse discrete inputs. Wire the polarity backwards and the drive does nothing at all.
• Bipolar: output is -10V to +10V and the sign carries direction. This saves two discrete outputs, but the drive must accept a bipolar reference – many general-purpose VFDs do not and need an interface board.

Confirm which mode the drive expects before wiring JA40. Get this wrong and no amount of parameter work will help.

3. The enable chain: four steps, in this order

The enable parameters must be set in sequence. Skip one and the symptom is “everything is set but the spindle does not turn”.

Step	Parameter	Value	Purpose	Consequence if wrong
1	8133#5	1	Do not use spindle serial output	Left at 0 the control still tries to talk to a serial spindle amplifier, producing communication alarms and no spindle command
2	3716#0 (A/Ss)	0	Spindle motor type: 0 = analog, 1 = serial	Wrong setting raises SP1996 (spindle parameter setting error)
3	3717	1	Spindle amplifier number for each spindle; for an analog spindle, its position in the spindle sequence	Unset or wrong: no alarm at all, and the spindle simply never turns – the worst failure mode here
4	3799#1	see below	Whether to check the analog spindle position coder for disconnection	Left at 0 with no coder fitted raises SP1240 (position coder disconnected)

On 3717, the official wording is to set the analog spindle in the spindle configuration. The rule: with N spindles in the system, enter the analog spindle’s position in the spindle sequence. Two serial spindles plus one analog spindle means the analog one is third, so 3717 = 3. A single analog spindle gets 3717 = 1.

On 3799#1 (0 = check for disconnection, 1 = do not check):

• No position coder fitted (speed control only, no orientation or tapping): must be 1, otherwise SP1240 appears at power-up.
• Coder fitted: set 0 and keep the disconnection check. It is a genuine safety feature that catches a broken coder cable; do not leave it disabled just to silence an alarm.

Then the three basic quantities:

Parameter	Meaning	How to set it
3735	Spindle motor minimum clamp speed	= minimum clamp rpm / motor max rpm x 4095; use 0 unless a minimum is required
3736	Spindle motor maximum clamp speed	= maximum clamp rpm / motor max rpm x 4095; use 4095 unless a lower ceiling is required (M-series systems usually need 4095 entered explicitly)
3741–3743	Spindle speed corresponding to 10V, per gear	With a gearbox, compute per gear ratio; with a single range, enter the spindle maximum speed directly

The classic symptom of a wrong 3741 is a speed that differs from the commanded S value by a constant ratio, because the voltage is linear in S and 3741 defines how many rpm correspond to full-scale voltage. Get that reference wrong and the whole curve is scaled.

3720 is the position coder pulse count = line count x 4 (quadrature). A 1024-line coder gives 4096. If this is wrong the displayed speed is scaled while the actual speed stays correct.

4. Calibrating 3730 and 3731: two formulas, one order

This is the only step that genuinely requires a multimeter, and the one most often applied backwards.

Separate the two jobs clearly:

• 3731 = offset compensation for the spindle analog output. It governs whether the output is 0V at S0 – the zero point.
• 3730 = gain adjustment data for the spindle analog output. It governs whether the output is 10V at maximum speed – the full scale.

The order must be 3731 first, then 3730. The reason is straightforward: the offset is a constant shift across the whole range. Tune the gain first, then change the offset, and the full-scale point moves again and you start over. Zero first, span second, and the two steps do not interfere.

Procedure:

1. Disconnect the drive (pull the analog input terminals or drop the drive enable) so the motor cannot actually spin.
2. Set 3731 = 0 and 3730 = 1000 (standard values) as a starting point.
3. Command M03 S0 and measure the voltage between pins 5 and 7 of JA40 with a multimeter. Call it V_offset.
4. Compute 3731 = -8191 x V_offset / 12.5. For example, measuring +0.03V gives 3731 = -8191 x 0.03 / 12.5, about -20.
5. Write 3731, command M03 S0 again and confirm the output is near 0V.
6. Command M03 S (the value in 3741) and measure the voltage. Call it V_span.
7. Compute 3730 = 10 / V_span x 1000. For example, measuring 9.97V gives 3730, about 1003.
8. Write 3730, command M03 S again and confirm the output is near 10V.
9. Reconnect the drive and verify under real load.

Two common traps:

• Doing only one half. Adjusting 3730 without 3731 leaves a zero error; even 0.2V is 2 percent at low speed, which shows up as a speed that will not settle and drifts left and right.
• Getting the sign wrong. The formula for 3731 carries a minus sign because the compensation acts against the measured offset. A positive measured offset needs a negative value; entering it positive doubles the error instead of removing it.

Note: the standard initial values of 3730 and 3731 differ between models (0i-D, 0i-F, 0i-MF Plus). Some ship with 3730 = 1000 and 3731 = 0; some field references quote 960 for 3730. Use the parameter manual for your control, but the calibration formulas are the same. The only acceptance criterion is the measured 0V and 10V – not the parameter value itself.

5. Coder selection: the 256 kHz ceiling

The JA41 port accepts roughly 256 kHz. The pulse frequency from a coder is:

f = line count x rpm / 60

So for a given line count the displayable maximum speed is fixed:

广告

Coder lines	3720 (lines x 4)	Displayable max speed	Verdict
512	2048	60 x 256000 / 512 = 30000 rpm	Speed is fine, but 512 lines is not enough resolution for rigid tapping
1024	4096	60 x 256000 / 1024 = 15000 rpm	Tapping is fine, but the display misbehaves above 15000 rpm
2048	8192	60 x 256000 / 2048 = 7500 rpm	High-precision tapping, low speed ceiling

This is where the classic conflict comes from: a 24000 rpm high-speed spindle and rigid tapping cannot both be fully satisfied under the 256 kHz limit. Choose 512 lines and the full speed range displays but tapping accuracy is insufficient; choose 1024 lines and tapping works but the display becomes unstable above 15000 rpm.

Practical rule:

• If the display is wrong but the actual speed is correct and rigid tapping executes normally, leave 1024 lines alone and tell the operator that above 15000 rpm the displayed value is not trustworthy and the drive panel should be used instead.
• If the process truly requires precision tapping above 15000 rpm, this route is not viable and a serial spindle is the answer.

One more point that is easily missed: 3716#7 controls whether the spindle position coder pulse count is changed. Set it to 0 when rigid tapping, because setting it to 1 stops the spindle speed being displayed during tapping. That is parameter behaviour, not a fault, but it routinely sends people on a long wild-goose chase.

6. The PMC side: half of an analog spindle lives in the ladder

The CNC supplies a voltage; everything else is the PMC tying the two sides together. Without this, correct parameters produce no function.

Direction	CNC-side signal	Meaning	Connects to drive
CNC to PMC	G70.6 (orientation command)	Set on M19	Drive orientation start input
CNC to PMC	G70.2 / G70.3 (CTH2A / CTH1A)	Gear selection (with a gearbox)	Gear change solenoids
CNC to PMC	G70.0 / G70.1 (SFR / SRV)	Spindle forward / reverse command	Drive forward / reverse terminals
CNC to PMC	G61.0	Rigid tapping signal	Drive rigid tapping mode input
CNC to PMC	G33.7	Spindle override source: 0 = CNC, 1 = PMC	Decides who owns the override
PMC to CNC	F45.1	Spindle zero speed detect	Drive zero-speed output
PMC to CNC	F45.3	Spindle speed arrival	Drive speed-arrival output
PMC to CNC	F45.7	Spindle orientation complete	Drive orientation-done output
PMC to CNC	F34.0 / F34.1	Gear select feedback (GR10 / GR20)	Gear confirm switches

The three most frequently missed:

1. F45.1, F45.3 and F45.7 must be routed back. Without them the control believes the spindle never reaches speed, so the block after M03 waits indefinitely, or auto mode never advances. Check by commanding M03 S500 and watching F45.3 on the PMC signal screen.
2. G29.6 (*SSTP spindle stop) must stay at 1. It is an active-low, emergency-class signal. Omit it or invert the logic in the ladder and the spindle never moves.
3. G61.0 must be synchronous with G28.1 for rigid tapping. G61.0 puts the drive into tapping mode; G28.1 switches to the tapping-specific gear range defined by 3742. If they are not synchronised the tapping voltage curve still uses the normal cutting range and the error cannot be brought down.

On G33.7: this signal decides whether the spindle override (50/100/120 percent) is controlled by the CNC or the PMC. The usual arrangement is CNC in AUTO/MDI (G33.7 = 0) and PMC in JOG/HAND (G33.7 = 1). Leaving it entirely to the CNC also works, but then the operator will find the override has no effect in manual mode.

7. Five high-frequency failure points

Symptom	First suspect	Criterion	Remedy
Parameters set but the spindle does not turn at all, no alarm	3717 unset, or G29.6 not held at 1	Is there any voltage at JA40? Check PMC state of G29.6	Enter 3717; fix the ladder
Commanded S differs from actual speed by a fixed ratio	Wrong 3741 or wrong 3720	Is “S / actual rpm” constant?	Recompute 3741 from full-scale speed; recompute 3720 from lines x 4
Unstable speed at low rpm, drifting left and right	3731 not calibrated, or drive drift compensation off	Is JA40 at 0V on M03 S0?	Recompute 3731 from the formula; enable drift suppression in the drive
Spindle lurches then trips SP742 (LSI overflow) during tapping	A/B phases reversed on JA41; 3742 inconsistent with the drive tapping limit; 5280 and 5291–5294 too high	Rotate the spindle by hand and check the displayed direction against reality	Swap A with B and *A with *B; align the two tapping speed limits; back off the gains
SP1240 (position coder disconnected) at power-up	No coder fitted but 3799#1 = 0; or a broken cable	Inspect the JA41 plug and harness	Set 3799#1 = 1 if no coder; otherwise trace the cable

On SP742: this is the most common and most frustrating alarm when an analog spindle is asked to rigid tap (rigid tapping alarm, LSI overflow). Its essence is that the spindle cannot follow the command, so the error accumulates until it overflows. Work three layers in order:

1. Feedback polarity (wiring layer). Reversed A/B means the control sees the opposite rotation direction and the error explodes immediately. Check this first – rewiring is the cheapest fix.
2. Speed-limit mismatch (parameter layer). 3742 on the CNC side and the drive-side rigid tapping maximum speed must agree. A factor-of-two difference guarantees overflow somewhere in the middle of the range.
3. Excessive gain (tuning layer). Pushing 5280 (rigid tapping position loop gain) and 5291–5294 (spindle loop gain multiplier) high to reduce tapping error actually makes the control overshoot and increases the error. Tapping accuracy should be adequate, not maximal.

Also remember the drive needs its own tapping mode. Most VFDs have a dedicated rigid tapping gain group (speed loop proportional gain, integral time, position loop proportional gain, accel/decel times). Without those set, no amount of CNC-side tuning will hold. On the CNC side, shrinking 5261–5264 (tapping accel/decel time constants) reduces error, but go too far and you trip SP742 directly because the commanded acceleration exceeds what the system can deliver.

8. Acceptance checklist

Walk this list at the end of commissioning. It is done only when every line passes:

1. M03 S0 gives 0V +/- 0.02V at JA40
2. M03 S gives 10V +/- 0.05V at JA40
3. M04 runs reverse and the displayed direction matches the physical direction
4. Turn the spindle by hand: display error under 1 percent
5. Measure five points from minimum to maximum speed: linearity deviation under 2 percent
6. M19 orientation, five repeats, deviation within process tolerance
7. Rigid tapping test cut with DGN 453 close to 1 (no SP0741)
8. All parameters survive a power cycle (3799#1 and friends are not reset)

Item 7 is the core rigid tapping criterion; the full treatment is in Fanuc rigid tapping parameters explained and the lookup table is at Fanuc rigid tapping parameter chart. Note that analog spindle voltage calibration and serial spindle gain tuning are two different bodies of knowledge – do not mix them.

Frequently Asked Questions

Can an analog spindle rigid tap?

Yes, subject to three conditions being true at once: the drive itself supports a rigid tapping mode (with its own tapping parameter group), a position coder is connected to JA41, and the coder line count is sufficient (1024 lines is the practical minimum). Without all three, only floating-holder (synchronous-free) tapping is possible. Note also that G61.0 must be set by the PMC and the gear range switched at the same time.

Why does the spindle not turn at all with no alarm when 3717 is unset?

Because 3717 defines which spindle amplifier this spindle belongs to. At 0 the control concludes the spindle has no amplifier, so it never issues a spindle command and does not regard the configuration as invalid – hence no alarm. This is the single biggest time sink in analog spindle work; check it first whenever “all parameters are set but nothing turns”.

Which one is zero and which is full scale, 3730 or 3731?

3731 is the zero point (whether the output is 0V at S0); 3730 is full scale (whether the output is 10V at maximum speed). A useful mnemonic: the larger number handles the larger voltage. Calibrate 3731 first and 3730 second, because the offset is a constant shift across the range and adjusting it afterwards pulls the full-scale point off again.

Last updated: 2026-10-31