Cabling issues in conduit

T Chap

I am experiencing a new problem I have never seen. There is a master controller that connects to a BLDC motor that is a 35' cable run away. The 3 conductor 18G cable that goes to the Motor is in the same 1" ID metal flex conduit as the CAT6 cable that runs the hall sensor lines, encoder lines, 5V and GND. Some systems with about 20' cable runs are fine. Two systems with the 35' cable runs are acting strange, both identical conditions, swapping the master controllers does not solve it. But, taking a master to the motor and using test cables that are 10' work fine on the motor. The problem is that on deceleration, the motors start doing what feels a jerky motion versus the smooth decel under normal circumstances. When I run a new CAT6 direct across the floor separate from the 3 conductor 18 that drives the motor (27V typical), the motors run perfect. So I have tried both a shielded CAT6 with the drain to the master chassis. The master chassis is to the ground plug on the AC, and is also tied to VSS on the main board. I have tried a non shielded CAT. No change difference. I have tried take the shield on the 3 conductor 18G cable to chassis, still the same problem. I cannot get back with a scope until next week, but I am assuming that there is something to do with the motor logic lines and motor 3 power lines in the same conduit, since separation solves the problem. Shielding does solve it.

The power to the systems is with a temporary source of a box that is ran into the house from the pole. I was told that this box has ground on it, but they also said the ground may not be identical to how it would be done when the real power is wired to the outlets. So that makes me wonder if there is an earth ground at all, since taking the shields to chassis doesn't solve anything.

I would appreciate any suggestions on what might be going on here. I am told that running the separate conduit will be a big project for the electricians, but they require conduit on all wiring so I am looking for a solution. Separate conduit was requested but never done.

Mark_T

The BLDC controllers really should be near the motors, but you presumably know this.

Sounds like you have electromagnetic induction, no just capacitive coupling here, and nothing
short of a Faraday cage approach is really adequate (separate conduits for motor and signals
being one implementation). Are all the logic signals differential? This is normally how things are
done in a noisy industrial environment. Opto-coupling is another way to deal with things.

Are the logic signal correctly terminated (CAT5 is 120 ohms I believe)?

T Chap

Thanks for the info. The controllers cannot be at the motors in this case because once installed, the motors are not accessible. The controller is most likely to need adjustment or repair so it is accessible and separate. On the motors that work fine, the logic and 3cond motor wires are in separate conduit. I will request the changes. I was just more curious as to why the wires can be fine if wired tied together for a distance outside the metal conduit, but inside there are big problems.

kwinn

I agree with Mark_T, most likely EMI, and shielded cable will not help much, if at all. Try using a twisted pair with differential line drivers/receivers and terminating resistors for each logic signal, and a twisted pair for GND/+5V.

skylight

Data shouldn't be in the same conduit as power, in the UK this is against the IEE regs is there a way to run two conduits to seperate them?

T Chap

My apologies, I had a misunderstanding with the installer. The 3 cond 18 for the BLDC motor windings is in separate conduit from the CAT6 for the Hall sensors, encoders. Yet, the system is very jerky on decel when the CAT6 is inside it's own conduit. When it is outside it's conduit, it is fine. I have tried a shielded CAT6 and unshielded. Both behave the same. Inside the conduit, jerky. Outside, fine. I have hundreds of these motors installed in all kinds of environments, and never have a problem. In most cases, there is no conduit. This is a first. I have swapped out master controllers, the problem is definitely in the wiring on two identical systems with the longer cable runs. Single ended on all, no way to change to differential at this point.

There are 4 systems in room. 2 have 25' cables, and run fine. 2 with 35' cables are identical problems. I have run much longer than 35' without conduit before with no issues.

whicker

did the cable have any kinks and tight bends in it when inside the conduit?

or rather, with the cable outside of the conduit, maybe induce a possible failure mode by putting a knot in the cable?

can you see if it is the metal conduit acting as some sort of capacitor by using plastic flexible conduit of the same length instead?

skylight

Are both conduits properly earthed?

T Chap

I understand that they are not using the actual power from the wall outlets, but are using a box that is pulled into the building from the pole. It looks like a large box about 2' long with a massive cable on it to the pole. From that, they are pulling an extension cord for the 110VAC to my master controllers. As far as the conduit, I have no idea what they have done but will inquire if they conduit is tied to earth. It is hard to get a straight answer on whether the temporary power is tied to earth.

skylight

Reason I ask if the conduit is tied to earth is that the normal general earth is usually very noisy especially in an industrial area(nearby factories, industrial plants...) and usually for the type of application you are dealing with a new "technical earth" is usually supplied to sensitive electronic equipment in noisy environments.
One way this can be derived is by supplying a three phase supply to a delta/star wound transformer the star output produces a new neutral/earth star point the earth derived from this is usually deemed clean or a "Technical Earth" this was a popular way to supply "computer suites" laboratories or other sensitive equipment environments etc.
Though of late equipment tends to come with their own filtered psu's so the method is not used so much these days.

kwinn

If the conduit is tied to ground it should be tied at one point only. If it is tied at both ends it is possible to have substantial ground current loops through the conduit. Another possibility is magnetically induced interference if the conduits are close and parallel to each other.

At one site an RS232 signal running through about 40 feet of conduit was totally garbled due to ground loop currents. Replacing all the metal clips with plastic (except at one end) solved the problem.

T Chap

Are the logic signal correctly terminated (CAT5 is 120 ohms I believe)?

I am told that the power in the house for the AC is grounded correctly. The metal conduit is not grounded at all.

I will be headed back out to the project on Monday and am putting together a list of things to try when I get there. Since this is my own proprietary wiring scheme and not a typical CAT wiring, I am not sure how this would be "correctly terminated". On another thread, I stated that the wiring scheme is this:


Pair 1
1 VSS
2 ENC A > 1KR > Prop input Quadrature Encoder Obj (5V signal)
Pair 2
3 VCC (5V)
4 ENC B > 1KR > Prop input Quadrature Encoder Obj (5V signal)
Pair 3
5 Hall1 6.25V signal to BLDC driver IC
6 Home Switch Sits at 5V pulled up via a 10k
Pair 4
7 Hall2 6.25V signal to BLDC driver IC
8 Hall3 6.25V signal to BLDC driver IC


Kwinn suggested this as the termination:

For the Vss pair it's Vss to capacitor to 120 ohm to signal. For the Vdd pair it's Vdd to capacitor to 120 ohm to signal.

I am believing that the jerky motion is a result of noise on the 3 hall sensor signals, not on the encoders lines. The hall output lines are pulled up to 6.25 V from the motor driver, so I am not sure I can easily run these through the same dual 74LVC2G14 Schmitt-triggered inverter Phil suggested to clean up the single ended lines.

In the mean time, in the next day or so I will have them test with connecting the master controllers to the actual outlets versus the temporary construction supply. Also they will try connecting one end of both metal conduits to the chassis of the master controllers.

I will work on a board tomorrow that will insert in the path of the CAT6 at the master controller, which will include the Schmitt inverter on the encoder lines. I will draw a schematic to show the idea for adding the resistors and caps at the motor suggested by Kwinn. Apparently the two metal conduits will not be able to be separated from each other, and that their preference is for both the lines to be in conduit. Running new conduit to separate the cables is not being received well.

It is a little late to try to design a single to differential signal scheme, primarily because I am not wanting to add any untested concepts and active electronics to the motors which will very soon become permanently inaccessible. The preference is to solve it at the master( main Prop board, power, BLDC motor driver), since this is accessible.

I appreciate the ideas.

kwinn

Would it be possible to pull another cat cable and move the hall signals to separate hall/ground pairs in the new cable?

T Chap

There is an RJ45 on the motor PCB, I would have to use a single CAT5 plug at the motor, hack the cable to split it off. But what is this gaining though? I think the 27V from the 3 conductor lines are affecting the logic lines, so wouldn't the proximity still be an issue with single ended?

kwinn

T Chap wrote: »

There is an RJ45 on the motor PCB, I would have to use a single CAT5 plug at the motor, hack the cable to split it off. But what is this gaining though? I think the 27V from the 3 conductor lines are affecting the logic lines, so wouldn't the proximity still be an issue with single ended?

Yes, proximity would probably still be an issue. It was a bit of a "grasping at straws" suggestion at this point. I thought perhaps pairing the signal with a ground might help. The only additional thought I can come up with (other than adding diff drivers) is to add terminating capacitor/resistors between ground and the Hall signals. Simple enough to do that it's worth trying.

T Chap

Would it make a difference for the termination at the master or motor? At the master is preferred.

T Chap

Here is the progress of the filter module. This will allow for adding the protection resistors mentioned in my other thread to avoid damage to the hall sensor outputs if plugged into the wrong port. It also allows for the optional use of the Schmitt Trigger Inverter Phil Linked(which I have already breadboarded and tested). This also has the termination resistor/capacitor options added based on my interpretation from Kwinn for the 2 encoder outputs, plus the 3 hall sensor outputs, and the home switch line. I will send this out to be made in an hour, any suggestions welcome. The ground plane will get added when done.

T Chap

I have been looking online at info on termination and see a youtube video that shows a square wave generator on one end, a scope on the same connection point, and a 1megohm pot at the far end of the cable. He shows the signal changing depending on where the wiper is, and demonstrates that distorted signal can be nearly corrected with the right resistance on the opposite end that reduces the reflections.

My question is, that in the case of the hall sensors, they are being pulled up at the motor driver on one end, and taken low at the motor when active. The encoder is different, it is powered by 5V from a source at the master controller, and the encoder output signal originates from the encoder on the motor. I can see how the encoder would in this case need to be terminated at the master controller, which is the opposite end of the encoder. But, the hall sensor seems confusing since there is no real point of the high signal, since it is only held to 6.25V with a resister, I have planned for the filter board to sit at the master controller. Hopefully using the termination at the master makes the most sense. I am still not clear on the use of the resistor and cap in series. The design is all through hole so it is easy to make changes to values. Since the hall sensor output signals are being pulled via resistors, obviously adding the 120ohm from the signal to ground is going to pull the output low, so maybe this is the reason for the r/c termination versus the resistor alone?

If both power consumption and signal quality are major concerns, AC Termination offers a compromise
between the parallel, series and unterminated schemes. By adding a capacitor in series to the termina-
tion resistor the DC current draw is significantly reduced while still keeping signal reflections low.

kwinn

T Chap wrote: »

Would it make a difference for the termination at the master or motor? At the master is preferred.

The termination should be at the receiving end of the circuit.

kwinn

T Chap wrote: »

I have been looking online at info on termination and see a youtube video that shows a square wave generator on one end, a scope on the same connection point, and a 1megohm pot at the far end of the cable. He shows the signal changing depending on where the wiper is, and demonstrates that distorted signal can be nearly corrected with the right resistance on the opposite end that reduces the reflections.

My question is, that in the case of the hall sensors, they are being pulled up at the motor driver on one end, and taken low at the motor when active. The encoder is different, it is powered by 5V from a source at the master controller, and the encoder output signal originates from the encoder on the motor. I can see how the encoder would in this case need to be terminated at the master controller, which is the opposite end of the encoder. But, the hall sensor seems confusing since there is no real point of the high signal, since it is only held to 6.25V with a resister, I have planned for the filter board to sit at the master controller. Hopefully using the termination at the master makes the most sense. I am still not clear on the use of the resistor and cap in series. The design is all through hole so it is easy to make changes to values. Since the hall sensor output signals are being pulled via resistors, obviously adding the 120ohm from the signal to ground is going to pull the output low, so maybe this is the reason for the r/c termination versus the resistor alone?

It really does not matter if the signal is produced by a pullup resistor and an open collector transistor pulling the voltage low. Same physics apply when the signal is sent over a transmission line.

T Chap

I wanted to thank you guys for the suggestions on this. The results on the hall sensors are improved on my bench with a 20' unshielded CAT5. At the site, they have taken each metal conduit to earth and reported less of the jerky motion. I will be adding in this filter/termination board next week. The board serves several purposes, it will be in a box so that any patching to the wrong ports will not damage the sensors at the motor. It contains the Schitt trigger that Phil recommended.on the encoder outputs. It uses the scheme Kwinn suggested for the resistor/cap on all signal lines. On most cases the cap is to the signal, the resistor is to GND. Only on the encoder signal that is in the pair containing VCC(5V) does it use the resistor to signal, cap to GND based on Kwinns notes. On several scopes the signal is noticeable improved.

As a test I included resistor options in the path of VSS and VCC. I used for the test 10ohm. You see more noise with the resistor. Both versions are shown.

No Filter is with the board removed.
Filter 10ohm is with the 10ohm on VSS/VCC
Filter direct power is no res in the power path

There is a little curve to the leading edge on the filtered versions. This is using a .01uf. Smaller caps are on the way Monday to see if I can get this a little straighter.

T Chap

Encoder Signals

kwinn

Good to hear. Hope this solves your problems.