Speed sensor problem


Thread Starter



Here we have a turbo-generator with 3 AI-TEK's passive magnetic sensors (2oo3 voting) to over-speed trip function.

There are two logic solvers (relay panel with AI-TEK's tachometers + Invensys Tricon) sharing the 3 sensors in a parallel connection with the same instrument grounding.

Now we are facing a strange behavior for the measures from sensor A. Until 3500 rpm they are equal to others sensors, them they are multiplied by two.

DC impedance is OK. Bad insulation isn't a issue. Wave forms and frequency from a scope meter are normal.

Thanks for any help.
Yes, thank you for asking, sir. The failure goes to new channel of sensor A.

> Have you tried swapping sensor A with one of the others and seeing what happens? <

Plus advances at troubleshooting:

As mentioned, at low frequencies the three sensors match. With 50 Hz/100 rpm as input we have about 500 mV. Thresholds are 50 mV and 450mV for tachometers and Tricon (3511 Pulse Input module) respectively.

Each input (+) and (-) of Tricon and AI-PAK seems differential, because wave forms are equal with GND as reference. However, wave form from (-) to GND has double frequency (for A, B and C sensors).

Inputs are polarized too, DC impedance is lower in reverse way (Tricon and AI-PAK are connected). DC impedance at inputs to GND are the lowest for inputs from sensor A.

In the panel, after of disconnection of field cable and pluggling a function generation to speed simulation, the behavior has been normal, but our func generator has lower impedance than sensors.

In the panel ending, DC impedances of sensor cable are +oo from (+)/(-) to GND and 200 ohms from (+) to (-).

Simulation from junction box in field isn't possible at moment.

Any hints?

bob peterson

I would suggest replacing the sensor since the problem seems to lie there. I would not get real worked up trying to figure out what went wrong with it.
I see, we can try the replacement for lack of other evidences. However, I'm not totally convinced of sensor failure. It would be the 2nd time replacing it (I've forgotten to mention it).
Jim - did you ever solve this problem?

We are having somewhat similar issues with AI Tek passive speed sensors. Above 7000 rpm 3 of our 6 sensors start to drift / experience large swings in speed. Have swapped wires around, etc and it appears to be something with those particular probes, but they have been recently replaced (a few weeks ago during another troubleshooting exercise).

Looking at potential wire issues but not certain that is the problem.
I would ask: Is this a problem which just recently started? Is it a new application of this pick-up?

Or, did it start recently? If so, what changed before the problem started?

When you say a similar problem, does it involve using the same pick-up as the input for two control systems?

Gap distance plays a big role in the output at various speeds. The size of the pick-up (cross-sectional area of the face of the pick-up) versus the width of the teeth being sensed also affects the output. Lastly, the shape of the teeth being sensed is a factor in the output.

In the past, I have contacted AI-Tek for assistance and they have been very helpful. In one case, simply increasing the gap distance solved the problem. In another case, it was determined that the pick-up (3/4-inch diameter face) on a 0.090 inch thick metal toothed wheel of 7 inches diameter was too wide, and using a bushing in the 3/4-inch hole with a 7/16-inch diameter pick-up solved that problem. Another problem was with the teeth; I think they were manufactured using a method called "hogging" which resulted in erratic teeth which, at higher speeds caused intermittent "extra" pulses.

But, a lot of these issues weren't all the "fault" of the speed pick-ups. The control systems which "interpret" the signals from the pick-ups are increasingly sensitive. Increasing the gap distance and changing to a smaller diameter pick-up were necessary because a new, more sensitive digital control system was replacing an older, less sensitive analog control system.

Most importantly, proper wiring (twisted, shielded pairs) with the shield drain wire being properly grounded at one end only is really very critical. It's amazing how many applications I've seen which try to get away with using single conductor wiring for speed pick-ups.

Passive speed pick-ups are pretty simple devices. it's not usually the speed pick-ups that are the problem; it's either the way they are applied (gap; size), the wiring and wiring methods used; and/or the sensitivity accuracy of the control system being used to "read" the pick-up output.

Aliasing of speed signals being read by a digital control system can also be a very bad problem. (I believe wikipedia.org has a good explanation of aliasing.)

So, are you certain the problem is with the output of the speed pick-up? Can you disconnect one of the speed pick-ups and use a scope or frequency meter to check it?

But, do reach out to the manufacturer, or one of their representatives for some application assistance. All speed pick-ups are not created equally. And many vendors and their representatives are very helpful when approached in the right way.

Especially if they think they will sell more product. And, if they're not, there are other passive speed pick-up manufacturers who are interested in your business.
This is not a new application, but we just recently completed an overhaul of the turbine and in the process we removed the gear/probes and reinstalled a new gear with (initially) the same probes.

During the OST testing we identified a speed deviation above 7000 rpm (OST set point is approximately 8900 rpm). We replaced the probes that were reading in error and retested with no issue. Now we are seeing a speed deviation again. It is more siginficant above speeds of 7500 rpm (normal running speed is > 7500 rpm).

We are in contact with AI Tek and they have been helpful - just not so sure they will be able to provide the answer to the problem based on their initial response (Oh - never heard of that happening) so thought we would check other sources.

We do know that the probe tip is less than the width of the gear tooth - but it was the same in the past. Gear tooth is not a true gear.

We cannot check the gap as this turbine is critical to our operation and cannot be shut down. But we've had several probes in and out of this installation for various reasons and our techs are very familiar with this process (have a total of 36 probes in 4 applications that are the same idea).

Yesterday we obtained scope traces of the faulty probes and have forwarded them to AI Tek - awaiting a response. I do not know what they are looking for but I can say that the signal at the probe is very stable and distinct. The voltage is lower than they expect, but the frequency is correct even at 7945 rpm. However, the same signal taken at the control room (several hundred feet of shielded twisted pair) but NOT connected to the governor system - knife switches open - is weak and erratic. So, this would seem to point to the wire. Will check the grounding as you mention in your post.

So, to top it off here are a couple of other bits of information. We have a different turbine that uses the same gear design in the same speed range with the same model probe that performs fine (speed deviation of less than 5 rpm). There are slight differences (probe holder not exact, etc)but one significant difference is the number of feet of wire between the turbine and the control room (working turbine has less).

And . . . on the turbine with the issue there are 6 probes - 3 at 6 o'clock and 3 at 12 o'clock. The 3 at 6 o'clock are working fine and the 3 at 12 o'clock have the issue. All 6 probes are the same model and all look at this same gear. This seems to point to the gap. We know that the turbine shaft rises in the bearings from rest to in service speed approximately .005", but we took that into account when setting the probes (6 o'clock probes were set with a .015" gap and 12 o'clock probes were set with a .025" gap - end result would be a .020" gap). Only other thing I can think of is that probe holder is deforming upwards, but we cannot check that.

My guess is that we have a marginal design based on the gear teeth being unoptimized. This is making us susceptible to other problems (slight gap variation and wire issues). However, since we cannot shut the turbine down to work on it we are stuck with trying to go after it from a wiring standpoint while working to improve the other items and waiting for an opportunity.

The signal at the deck is 8 V rms and in the control room it is 0.5 V rms. What voltage drop would you expect over say 800 feet of 16 gauge wire?

What can be done to minimize this drop?

Can the signal be reliably boosted?
I would not expect to see much voltage drop for the 800 ft. of wire unless there is appreciable impedance between the wires (that is for an AC waveform, capacitance and resistance) to reduce the voltage. Check to see if its either a fault in the wire itself -or- that the rated capacitance of the cable matches the specification for the wire as published by the manufacturer. Especially since the speed impacts the sensing of the generator, that seems a likely source for the problem. The higher the frequency, the lower the impedance between the wires.

A test you could do is to put 8 VDC on the wire at the sensor and see if the same 8 VDC shows up at the other end. If it is near 8 VDC you probably have a capacitance issue, if it also is 0.5 VDC, likely there is a resistive load somewhere on the wire. I’d likely look for a break or kink in the cable that might have happened when the equipment was out of service.

A simple source for the DC would be a 9V transistor radio battery. You would want to put a resistor in series with the battery to limit the current if there is a partially shorted cable causing the problem. Maybe 1000 ohms to limit the current to under 10 mA, however the value is not critical.

There may be some more involved issue causing the problem but I'd check these items first.

Okay; this is a problem which has just started after a mechanical outage of the turbine, and no new control system was installed. The speed pick-ups and brackets were removed and re-installed using (initially) the same probes, and the toothed speed pick-up has not changed.

The probes were changes and the same gap (I presume you are relating the distance between the probe tip and the toothed wheel to the width of the toothed wheel; I've never seen gap expressed like that, but I guess it's one way and probably works, especially if the gap (the distance between the probe tip and the toothed wheel) is the same as it was before, no matter how that's specified or expressed.

And, the shape of the teeth hasn't changed, though one wonders if they were "damaged" during the overhaul (mechanics aren't always the most considerate of things without bolts or gaskets or can't be assembled or disassembled without a cutting torch). But, let's presume they are in no worse condition than before.

Have you used a meggar to measure the wiring between the control system and the speed pick-ups? Are you sure the shield drain wires are properly terminated?

You seem to have multiple speed pick-ups. Are they all behaving similarly? Or just one or two? What happens if you reconnect probes and cables in the junction box closest to the speed pick-ups and then observe the feedbacks? Do the problems follow the pick-ups or remain with the cables?

In situations like this, the vendor may or may not be very helpful. There are a lot of things which they may recognize or be able to recommend, but there's just a lot they can't know and must presume.

I've never heard of adjusting gaps to compensate for shaft movement, but I would imagine for some applications it might be necessary. Do you have any of the original documentation supplied with the turbine and control system to see what the OEM recommended?

I'm leaning more towards a wiring issue because you say the signals are very, very stable at the JB closest to the speed pick-ups, and not very good in the control room.

I'm not familiar with a speed pick-up signal booster.

There are tables for impedance/unit length and resistance/unit length for wire and cables. Most manufacturers nowadays publish them for their products. A lot of it has to do with number of strands, twists, material, etc., but I wouldn't expect that much of a drop over that length with "new" wires or wires in good condition. Are all the pick-ups exhibiting the same voltage drop, or just one or two?

Is the probe holder/bracket firm?

Are there any intermediate junction boxes with terminal boards? What condition are the terminations in those junction boxes in?

Another "simple" check would be to run a temporary cable (of suitable size/type, and with the shield drain wire properly grounded) from the JB closest to the speed pick-ups to the control room and observe the feedback to see if there's any degradation in the new cable.

I've always equated speed pick-up gap to spark plug gaps. Too little gap and too much gap are bad. And there is an optimum gap which produces the best voltage or spark for all operating conditions.

You say there are other turbines which are not experiencing the same problems. What are the voltages at the control room with respect to the voltages at the JB closest to the speed pick-ups?

Let us know what you discover, please!
> Have you found out the cause(s) to this problem Jim?

We did identify what we feel is the problem. The machine in question has continued to run and we have not been able to implement our final solution. We have a reasonable interim solution installed (used low capacitance wire).

For a detailed overview of the problem/solution go to:

http://turbolab.tamu.edu/proc/turbo...s in Electronic Governor and Trip Systems.pdf

<b>Moderator's Note:</b> When copying and pasting long URLs, please make sure you copy the entire URL and remove any spaces placed in the URL by the forum software.