We've discussed several times what will give you "ski slope" problems with data, and I've run into something I can't figure out.

My data collector setup is a 2120, a CTC coiled cable with a male BNC connector, a BNC female-female union, a 'whip' with a male BNC and a military 2-pin connector, and a variety of standard 2-pin accels. This arrangement allows me to quickly change accels in the field since I have a 'whip' attached to each of my accels (I use 3). The BNC union is 'sacrificial' -- when I wear it out from connecting/disconnecting the whips, I just insert a new union. This arrangement works really well, but.....

Lately I've started getting ski slope on a number of points. I thought it was a bad batch of 'whips' that I bought recently, but I had the problem with all of them. Then I noticed that when I was taking data, if the 'union' that connects the BNC cables was in contact the machine, I'd get the ski slope. Isn't the accel supposed to be "case isolated" so you don't get a 'loop' ? I switched accels, and it's better, but it still happens more than it used to. How can I check to see what is going on here?

May be your cable or such, but I have had the adapter that connects to the meter to get "worn" or loose (the screws that hold the spring loaded retainer to the adapter) or whatever and the slightest movement of it, such as when reaching to another point on a machine and "stretching" the cable or "swinging the cable", would cause a bad ski slope. I am not sure if you are using this type adapter (Example #625, etc).

As far as the coupling touching the machine, that seems strange since the coupling should be connected to the "ground" for the coax cable's outside "wire mesh" wire.

But anyway, just a thought.

Rusty,

When I was installing on line systems, I found that we had changed manufacturers of the "whip" as you call it, that connects to the transducer through the mil connector. The cable they started to use actually had the shield connected through the mil connector to the transducer shell. That was a no-no. I ended up taking a bunch of connectors apart, and cutting the shield loose from the mil connector.

You might want to take an ohm-meter and check continuity from the male end of the whip to the screw threads of the female mil connector (part that connects to the transducer) and see if the shield is attached to the transducer end also. If so, then cut it. You only want the shield connected at the analyzer end.

Dave

Rusty,

I'll looking into the connection at the 2-pin connector to see if the shield is connnected to the shell.

One concern is the quality of the BNC unions. I saw some from Radio Shack the other day that were really bad. Make sure you're using a quality part here even if it is sacrificial.

The BNC shell is of course part of the signal path. Randomly shorting it to machinery may induce other noise than a ground loop. A piece of sleeving would be a good idea to insulate it.

Jon
http://www.spintelligentlabs.com

Rusty,

In a small sample of CTC 2-pin - BNC cables, the shell of the 2-pin connector does not connect to the shield side of the cable. As far as I know, this is always the case (I'm waiting for an answer from CTC to verify).

An intermittent ground between a floating system like the CSI 2120 & accelerometer and a system tied to earth ground sounds like trouble to me. There's a potential on the floating system and a different potential at the ground point. When these get connected, charge flows for at least a short period of time (if there is a DC potential difference) or continuously if the point of connection has an AC potential.

My suggestions are to use a quality BNC adapter with gold-plated center contact and insulate the BNC connections with a sleeve or other material.

Jon
http://www.spintelligentlabs.com

Are you also integrating to in/sec? If so you are likely to get the ski slope at low frequencies unless you apply a suitable high pass filter.

John

Rusty,

I concur with Jon on the ground loop issue. I'm hope that you make sure that your coiled cable isn't moving much while collecting data. I've seen that induce ski slopes. Only skiers like ski slopes.

The reason is most probably the poor cable shield and the BNC connectors.

The coiled cable may be RG 58 type.
So it may work like an radio antenna.

If not, the BNC connector has to be shield also.

quote:

Are you also integrating to in/sec? If so you are likely to get the ski slope at low frequencies unless you apply a suitable high pass filter.

If you have a good cable and hold the accel reasonably still while taking data you will not get ski slope when using analog integration (which is the typical setup). Digital integration is prone to ski slope, but I don't know anyone who uses digital integration for routine route work.

Rusty
If attaching a pickup instead of using a stinger or Magnet, I've seeen Ski slope if the point is hotter than the previous point and you don't wait for it to setttle. Also if you tighten a center holding screw too tight.

"you will not get ski slope when using analog integration (which is the typical setup). Digital integration is prone to ski slope, but I don't know anyone who uses digital integration for routine route work."

Rusty,

It does my heart good to hear that you use analog integration for routine data collection.
There are those around this area who use digital for "speed", even on new equipment that they are just setting up. I "conversed" with one gentleman who had set up a 3600 rpm pump with an F-max of 40K. He wondered why he missed some things going on. Even after pocketing his money and leaving, I don't think he believed what I was telling him and had showed him.

Keep up the good work. By the way, I agree with your statement on the ski slope.
I think John was right, shield the BNC coupling with something.

Dave

Rusty,
I have experienced this same phenomina with Datapac equipment. Most cheaper BNC "barrel" union connectors use the outside case as a transmission path for the cable shield... thus when the union contacts metal at a different potential than the floating data collector, a DC voltage is induced, thus the ski slope. I isolated mine with black electrical tape and it solved the problem. For me, this was a major problem when using my 8 channel zonic with long cable runs using the barrel connectors. If you do a google search for "BNC to BNC in line connector isolated" you will find that bulkhead BNC connectors often offer isolation due to the induced voltage problem.

Rusty,

I have found the CTC cables to be very high quality, and I would be very confident that they are very aware of the correct way to hook up the wires.

If you are using a CSI adapter such as a model 625 or a model 611 these are prone to having problems, I would try another adapter to see if the problem goes away. The adapters go bad internally and they also loosen up at the 25 pin connector. You can take your hand and apply some downward force and a lot of times the data will improve.

Rick brings up a very good point. DB-25 connectors are rated for a limited number of make-break cycles.

A quick search on the web found the following:

"DRAFT Standard for Medical Device Communications - Recommended practice for high cycle life connectors

"1.Overview

"This draft proposes a recommended practice describing the function and performance of high cycle life DB9 and DB25 style compatible connectors. This recommended practice defines a variation of the industry standard DB type connector described in IEEE 1073.3.2. The typical design of DB type connectors limits the cycle life to approximately 500 mate and demate cycles, with some variation due to the quality of the manufacturers . Due to the variances in gold plating, and the wiping action inherent in the standard pin and socket; at or around 500 cycles intermittent contact can occur."

The expected life of these connectors is only 500 mate/demate cycles! If you change cables a lot, or disconnect the sensor when putting the data collector away for the day, intermittent problems can be expected.

Jon
http://www.spintelligentlabs.com

Rusty,
I have used audio XLR connectors for years without problem. It's a holdover from using an IRD 350 which I still use on occasion. I've made a short adapter with 2 female XLR connectors for the analyzer end, and a couple of short cables with male XLR to the 2 pin connectors at the pickups. When I need to extend the cable run, I just add in a stock 25 foot length of microphone cable bought with XLR connectors on each end. The microphone cable is very flexible, reasonably durable, and noise immune. Connection and disconnection is click lock / push-button quick.

vger,
I didn't know anyone was still using an IRD 350. That is the first analyzer I learned to balance with. I used it in the shop on our stand and in the field for balancing fans. It was a good instrument but I don't miss balancing with a strobe.
Sorry, didn't mean to interrupt the thread but that brought back some old memories.

Thanks,
Ronnie

Rusty, have had ski slopes develop from well worn 625 type adaptors as well.

If you read the original thread, I was actually asking about getting a "loop" if an accel lost it's case isolation. I assume that "case isolated" mean the ground is isolated from the case (sort of goes without saying that the signal is isolated from the ground and case, right?).

Have any of you ever had an accel that lost it's case isolation?