Keysignt 3458A connections for long distance measurements.

[vini_i]

The 3458A is an amazing meter with the worst manual ever. It's almost as though the manual was written by a lawyer.

I want to make ACV, DCV and resistance (4 and 2 wire) measurements from about 25ft away. (No Current measurements)

What would be the best way to make the connections?

The 4 connections obviously must go to the device being measured, HI, LO 2 wire and HI, LO 4 wire. After that is were things start to break down for me. The cable is going to be in a noisy environment and I would like to shield it. Where do I connect the shield?  Do I connect the shield at the meter or at the device being measured?
 Is the shield separate from the Guard or is the shield the Guard?

[dl1640]

it depends on whether your device under test is grounded or floating?

[HighVoltage]

How noisy is noisy?
Where does the noise come from.

Tell us more about the signal and DUT that you want to measure.
Is there a wall (steel?) between the 3458A and the DUT

I had a similar challenge a few years back with a few 34401A and used shielded
cables, the shield on both sides to a good ground. If one side was not grounded
the problem was huge if both sides got grounded at the shield, it was usable at
50ft distance.
 
 

[zhtoor]

hello,

is there an isolated GPIB i/f out there, and could a GPIB i/f be extended to run 25+ feet?

regards.

[vini_i]

it depends on whether your device under test is grounded or floating?

It could be either, floating or grounded.

[vini_i]

How noisy is noisy?
Where does the noise come from.

Tell us more about the signal and DUT that you want to measure.
Is there a wall (steel?) between the 3458A and the DUT

I had a similar challenge a few years back with a few 34401A and used shielded
cables, the shield on both sides to a good ground. If one side was not grounded
the problem was huge if both sides got grounded at the shield, it was usable at
50ft distance.

The signals nearby are 60Hz to 1200Hz at 4A and up to 260V. There are also RS232, RS485, CAN, and a few other comms protocols.

The existing setup has a shield but whoever designed it tied the shield to the LO 2 wire pin somewhere in the middle. As it sits right now any ACV measurement gets 30 to 50 mV of RMS noise on it.  I want to upgrade the system just not sure how to redo the wiring.

[vini_i]

hello,

is there an isolated GPIB i/f out there, and could a GPIB i/f be extended to run 25+ feet?

regards.

While there are isolated GPIB setups, my setup is already built. It needs an upgrade so to speak.

[zhtoor]

hello,

is there an isolated GPIB i/f out there, and could a GPIB i/f be extended to run 25+ feet?

regards.

While there are isolated GPIB setups, my setup is already built. It needs an upgrade so to speak.

hello,

would having an "upgrade" to GPIB and extending GPIB instead of the input terminals solve your problem?
or do you still *have* to extend the input lines?

regards.

[dl1640]

try to connect the shield to 3458 guard/ground if uut is floating

try to connect the shield to uut ground if uut is grounded

also minimize the loop area of your cabling or maybe double shield your cables

have fun...

[MisterDiodes]

In addition to other suggestions:

Normally you don't want the cable shield carrying current.  Normally the system guard circuit is connected to ground at one point only, if the guard is to be grounded.  If your UUT isn't grounded, then it's going to try to be anyway thru a cable shield connected at both ends of the cable, and that may cause current flow in the shield.

Since you're meter is so far away one thing to try is adding an extra dedicated ground conductor from the UUT to the same ground your meter uses - try to eliminate / reduce shield current flow.  Also keep the signal cable away from any steel parts that are in close proximity to your power cables (i.e. don't run the delicate signal cables in the same raceway as the power cables) - the shielding isn't doing much for near-field magnetic pickup.  The combination of a steel cable raceway + power cable + low level signal cable makes a great set of multiple random one-turn transformers and can be a real problem even with shielded cable. 

If your signal cable has to cross over a power cable, always route your signal cable perpendicular to any power cables - never run signal + power cables parallel to each other.

Try what works in your situation for best effect.

[vini_i]

I think I need to clarify some things.

The 3458A has 6 jacks. Two are used for 2 wire measurements. Two are used for 4 wire measurements. One is used for current (i'm not using that one). The last one is the guard. The manual does little to explain what the guard does besides to connect it to the negative side of the UUT along with the LO jack and that it reduces common mode noise. The guard is not grounded. Even the graphics on the meter go out of their way to say that the guard terminal has a max voltage of 500V to GND and 200V to LO jack. Not much is said about the guard beside that.

Clarified questions.

1. Should I use twisted pairs of wires for the 2 and 4 wire jacks?
2. If I wrap them in a shield should I connect the shield to the guard terminal or the GND lug on the back of the meter or both or none?
3. If the shield is not connected to guard, should the guard get its own wire in the bungle out to the UUT where it would get tied to LO?
4. Should the 2 and 4 wire pairs get shielded separately running them together in the same shield acceptable?
5. Are there any scenarios that I didn't cover that I should be aware of?

[dl1640]

i think *guard* is the shield of analog section
normally guard is not connected to chassis the metal case
guard could be tied to Lo if you push the switch on the front panel

i think you can try to leave the guard and make a one point ground connection in your system

and what kind of cabling you use now?

[vini_i]

i think *guard* is the shield of analog section
normally guard is not connected to chassis the metal case
guard could be tied to Lo if you push the switch on the front panel

i think you can try to leave the guard and make a one point ground connection in your system

and what kind of cabling you use now?

Currently, the wiring is two pairs (2 wire and 4 wire) that go to the UUT. There is a shield that covers all 4 wires. The shield is tied to the 2 wire LO line about 3 feet from the UUT.

[HighVoltage]

I would tie both ends of the shield (beginning and end of cable) to a really good ground and start there.

[RandallMcRee]

Ott warns against grounding at both ends...I can't seem to find that on the internet--its in his book on Noise...
AD has this tutorial which says the same:
http://www.analog.com/media/en/training-seminars/tutorials/MT-095.pdf

Also, note that shielded twisted pair is necessary in almost all real-life cases.

I did not read that as carefully as I would if it were *my* measurement, but it seems to be recommending grounding at far end and capacitive coupling at other.

[dl1640]

i think *guard* is the shield of analog section
normally guard is not connected to chassis the metal case
guard could be tied to Lo if you push the switch on the front panel

i think you can try to leave the guard and make a one point ground connection in your system

and what kind of cabling you use now?

[IanJ]

I would tie both ends of the shield (beginning and end of cable) to a really good ground and start there.

By grounding both ends you create a loop effectively, and on long runs this can be detrimental. It's a no-no in the world of instrument cabling offshore etc. However, with a short run you might get away with it and you might get mixed results.......worth trying (experiment) if you've nothing else to lose.

Ian.

[HighVoltage]

I would tie both ends of the shield (beginning and end of cable) to a really good ground and start there.

By grounding both ends you create a loop effectively, and on long runs this can be detrimental. It's a no-no in the world of instrument cabling offshore etc. However, with a short run you might get away with it and you might get mixed results.......worth trying (experiment) if you've nothing else to lose.

Ian.

I agree under normal circumstances that this can be bad. But, I had a project last year where I could not get rid of the noise until both ends were grounded.

This was about a 10 m long cable and we had to deal with a motor controller noise. Once we grounded both ends to a real heavy metal ground, we had solved the problem and could start taking measurements. In this case, the ground loop had no influence.