Sign up Calendar Latest Topics Donate
 
 
 


Reply
  Author   Comment  
vogel

Sr. Member
Registered:
Posts: 76
Reply with quote  #1 

So this is a continous monitoring application with ICP/IEPE accelerometers. How far would you run the field cables to the monitoring system?

I've searched the net and I've found different answers, I'm not too convinced by any of them.

  • This article from IMI/PCB suggests that the maximium cable length of the is limited by the capacitance of the cable, which affects the frequency response and it relates it to the amplifier slew rate (I can't see the reasoning here, I think that the slew rate is a characteristic of the amplifier not related to the capacitance of the cable): If the charging of the electric field is restricted by a low sensor constant excitation current, the field cannot be charged fast enough in order to provide adequate voltage to the sensor amplifier in order to maintain its slew rate (ie. the ability of the amplifier to effectively maintain output).
  • Then it uses a formula to calculate the maximum frequency that can be achieved based on the cable capacitance. However it doesn't explain how this formula is derived. I understand that it assumest that the maximum current flowing through the capacitor is 3 mA supplied by the system - 1 mA going through the sensor, this doesn't make too much sense to me.
  • This article from Wilcoxon uses the same formula but the conclusion is somehow different: when using too long wires (thus too high capacitance) one will get a triangular waveform and consequently harmonics, while the PCB article states that the result is an LP filter. 
  • Other articles from MMFNI or Rockwell describe the problem on similar terms. The article from MMF shows a plot where the frequency response doesn't change too much until 5 kHz with different cable capacitances.
  •  
  • On the other side, this article from Hansford Sensors seems to limit the length only due to hazardous area specifications. 
What do you think? Can i run long wires without a horrible frequency response?

John from PA

Sr. Member
Registered:
Posts: 963
Reply with quote  #2 
Instead of looking at what several vendors are recommending, why don't you check with the supplier of the exact accelerometer you will be using?
fburgos

Sr. Member
Registered:
Posts: 670
Reply with quote  #3 

according to ni 300ft or 90 meters, its is a very large distance.

rockwell 152 meters 6db atenuation... is it something between 60 and 70% of the signal at 10khz.

The question is how many information your signal have above 5khz, and how accurate you want it to be, trending with same atenuation its ok if you have a good set of data.

I think its more important to avoid ground loops

OLi

Sr. Member
Registered:
Posts: 1,915
Reply with quote  #4 
You should note that you have a phantom feed system IEPE and the only thing I would consider is that using the constant current feed capability of your system you get voltage enough at the transducer end of the cable to feed the inside electronics. There are tables of that too to be found but depends of the cable resistance. So check how many mA your system can produce and adapt accordingly 2,4,8 mA or more. I would if possible double screened coax like RG58 with double screen if possible as it is a RF freq cable and with 2 screens correctly connected at signal ground only at instrument end and no other place you at least get 1 working screen and that makes a difference. The screen you use to run supply to the sensor does not count as it does not work as a screen being connected at both ends and carrying supply current. Pair screened cable would work but may give more signal damping not that I noticed it here it is called FKARPG. So powering the transducer and reducing noise is what I would make sure work for long cables, specific power, w/o power you get no good signal.
__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 599
Reply with quote  #5 
Quote:
Originally Posted by OLi
You should note that you have a phantom feed system IEPE and the only thing I would consider is that using the constant current feed capability of your system you get voltage enough at the transducer end of the cable to feed the inside electronics.


Wouldn't the limitation be the maximum power deliverable from your IEPE supply? If the impedance is too high for your IEPE supply to maintain the 2mA, does it (generally) just supply less than 2mA or does it stop supplying the current? Most IEPE sensors have an acceptable current range, regardless of the voltage at the supply end. So as long as the power limit of the supply is sufficient, the leads of the cable shouldn't affect the voltage at the sensor whatsoever. Right?
Walt Strong

Sr. Member
Registered:
Posts: 889
Reply with quote  #6 
"What do you think? Can i run long wires without a horrible frequency response?"

What is your length requirement?
What is your frequency response requirement, especially F-max?
What is providing accelerometer power (voltage and current)?
What is the monitor/analyzer input impedance requirement?

I have used single CAT-5 cable for 4-accelerometers (also microphones and optical tachometer) up to 500-ft! CAT-5 cable has four twisted-shielded wire pairs.

Walt
OLi

Sr. Member
Registered:
Posts: 1,915
Reply with quote  #7 
If you put like 2mA in to the cable and you have specific voltage drop in the cable due to the resistance due to cable length going to the sensor it is the voltage left in the sensor end that will tell if you can use it or if you need to increase the current if you can. Current is constant, that is the whole idea with constant current supply but in the end the transducer need a minimal voltage to perform the charge amplification and if it is not enough it will clip the signal or not perform properly.
__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
vogel

Sr. Member
Registered:
Posts: 76
Reply with quote  #8 
Quote:
Originally Posted by John from PA
Instead of looking at what several vendors are recommending, why don't you check with the supplier of the exact accelerometer you will be using?


Because I have different accelerometers models to choose. Moreover,  should I trust the supplier when there are so many differences in their answers for a single phsysics problem? I'm trying to understand how it works, why it may fail and what are the possible problems that I may encounter. I am somehow flexible on accelerometer and DAQ selection if I know what I need.

Quote:
Originally Posted by Curran919

Wouldn't the limitation be the maximum power deliverable from your IEPE supply? If the impedance is too high for your IEPE supply to maintain the 2mA, does it (generally) just supply less than 2mA or does it stop supplying the current? Most IEPE sensors have an acceptable current range, regardless of the voltage at the supply end. So as long as the power limit of the supply is sufficient, the leads of the cable shouldn't affect the voltage at the sensor whatsoever. Right?


If the impedance is too high the voltage will drop but the IEPE supply (usually a current limiting diode) will keep providing the 2mA no matter what the impedance is.

Quote:
Originally Posted by OLi
If you put like 2mA in to the cable and you have specific voltage drop in the cable due to the resistance due to cable length going to the sensor it is the voltage left in the sensor end that will tell if you can use it or if you need to increase the current if you can. Current is constant, that is the whole idea with constant current supply but in the end the transducer need a minimal voltage to perform the charge amplification and if it is not enough it will clip the signal or not perform properly.


I would say that the voltage drop is mostly due to the cable capacitance, not resistance. It's for sure the case with the specific cable I'm using (capacitance is higher than resistance). In this scenario, the IEPE current would not be able to provide the current needed to charge the cable capacitance, I = 2*pi()*C*V, and thus the accelerometer waveform on the capacitor would be clipped ... this is what I've understood from the PCB article, but I doubt that the circuit is actually working this way.

Quote:
Originally Posted by Walt Strong
"What do you think? Can i run long wires without a horrible frequency response?"

What is your length requirement?
What is your frequency response requirement, especially F-max?
What is providing accelerometer power (voltage and current)?
What is the monitor/analyzer input impedance requirement?

I have used single CAT-5 cable for 4-accelerometers (also microphones and optical tachometer) up to 500-ft! CAT-5 cable has four twisted-shielded wire pairs.

Walt


Length: I'd like to arrive to ca. 250-300 m (800-900ft) 
F max: I would be happy with 5 kHz, but 2,5 or 3 kHz is also ok.
Excitation current: 2,1 mA.
Input Impedance: 305 kΩ. (I could change DAQ if needed)




OLi

Sr. Member
Registered:
Posts: 1,915
Reply with quote  #9 
Cable resistance I discuss is the DC resistance for the supply (not the AC signal property) and if you don't have a lo resistance cable the voltage split at 2mA according to Ohm's law will not leave enough voltage at the sensor end to supply the sensor. In the end it is the DC supply voltage at the sensor that define the signal swing and if the amplifier inside the sensor works. It is in the end supplied by a voltage not the current so if the current is not enough and for a few 100m, 2mA would be on the low side if you apply Ohms law. This is different from the frequency transfer properties, I basically don't care about that I never had any problems from that but if your sensor don't get power enough you get no or clipped signal and that is worse. On the other hand if you just can crank it up with another CC diode or setting it is no problem. Sensor OEM normally got a table for this DC supply cable length property too or a basic number like "for cables more than 200m increase CC supply to 8mA", just an example. So select the sensor with the best properties that you have budget for, buy 3x100m TP cable or if you find any other cheaper TP cable that you can buy 100m at the time and start testing. Sensors are not identical or at least not the electronics inside (and for sure not the piezo either) so there are no reason they all work the same in this respect so that the OEM tables will differ does not surprise me. So as I remember IEPE started out suggesting 4mA CC and allowing as low as 2mA battery operated instrument makers soon adopted 2 or 2.1 mA that is the closest standard CC diode as the norm, some things for continuous monitoring can be set for even 8 or 12mA I hardly ever seen mentioning for 16mA but who knows. So long cables should be possible.
__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 599
Reply with quote  #10 
Quote:
Originally Posted by OLi
Cable resistance I discuss is the DC resistance for the supply (not the AC signal property) and if you don't have a lo resistance cable the voltage split at 2mA according to Ohm's law will not leave enough voltage at the sensor end to supply the sensor. In the end it is the DC supply voltage at the sensor that define the signal swing and if the amplifier inside the sensor works. It is in the end supplied by a voltage not the current so if the current is not enough and for a few 100m, 2mA would be on the low side if you apply Ohms law. This is different from the frequency transfer properties, I basically don't care about that I never had any problems from that but if your sensor don't get power enough you get no or clipped signal and that is worse. On the other hand if you just can crank it up with another CC diode or setting it is no problem. Sensor OEM normally got a table for this DC supply cable length property too or a basic number like "for cables more than 200m increase CC supply to 8mA", just an example. So select the sensor with the best properties that you have budget for, buy 3x100m TP cable or if you find any other cheaper TP cable that you can buy 100m at the time and start testing. Sensors are not identical or at least not the electronics inside (and for sure not the piezo either) so there are no reason they all work the same in this respect so that the OEM tables will differ does not surprise me. So as I remember IEPE started out suggesting 4mA CC and allowing as low as 2mA battery operated instrument makers soon adopted 2 or 2.1 mA that is the closest standard CC diode as the norm, some things for continuous monitoring can be set for even 8 or 12mA I hardly ever seen mentioning for 16mA but who knows. So long cables should be possible.


I am certain you know more about this than I do, but it does contradict what I've always thought.

I've always considered an IEPE sensor to act as an impedance proportional to acceleration, so discounting lead impedance, the CC will yield a voltage proportional to the acceleration. However, I now realize that this causes two problems. First, the CC should always result in a positive voltage at the DAQ terminal, which isn't true, as the sensors are designed to oscillate around 0V. Second, when there is (nearly) no vibration, the impedance of the sensor would be zero, and therefore there would be no voltage over the sensor for its circuits, even if none is needed for the amplifier. Is there something simple I am missing.
OLi

Sr. Member
Registered:
Posts: 1,915
Reply with quote  #11 
Yes, the Bias voltage that is roughly supply voltage/2 or from 8-12 VDC normally accepted range vary with the sensor design? AC signal rides on that but it also define and limits max signal levels to be transferred w/o clipping/cutting upper or lower side......
__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
Walt Strong

Sr. Member
Registered:
Posts: 889
Reply with quote  #12 
"I've always considered an IEPE sensor to act as an impedance proportional to acceleration, so discounting lead impedance, the CC will yield a voltage proportional to the acceleration."

The CC does not convert acceleration to voltage, but only supplies voltage to the amplifier. The amplifier converts the charge on the piezo crystal (proportional to acceleration) into a proportional voltage with low impedance output. The voltage output from an ICP or IEPE accelerometer is an AC signal that coexists on the same two wires as the DC voltage to the accelerometer amplifier as controlled by the constant current diode (CC). The vibration analyzer or DAC is AC coupled, so the DC "bias voltage" is not present, and the acceleration signal goes +/- about 0-volts.

As Olov stated, the bias voltage (power to amplifier) affects the peak acceleration voltage (primarily negative value). For example, if there is 18-volts DC bias voltage at accelerometer, there would be 9-volts (half of 18) for maximum negative acceleration voltage. An accelerometer with 100 mv/g sensitivity (0.1 volts per g-acceleration) could produce a maximum negative acceleration of 90-g (or 9/0.1), but the practical limit often specified is about 80-g to keep the amplitude linear and not clipped. A less sensitive accelerometer, such as 10 mv/g, can deliver a higher g-value for the same bias voltage.

Walt
P Hine

Member
Registered:
Posts: 47
Reply with quote  #13 
We just finished installing a CSI 6500 online system. We were told by CSI and CTC that we were good up to 1000 ft. for the cables. The cables are shielded and have a drain wire. The longest cables we needed were less than 200 ft. That seems to work well. In the past that has seemed to be the max length. I am having some problems with a 12 pair multi conductor cable (~100 ft.) that acts like a ground loop is present. Need to put an o-scope on it and look at the waveform.

Phil
OLi

Sr. Member
Registered:
Posts: 1,915
Reply with quote  #14 
Even if the supply voltage is 18 to 24 VDC w/o sensor connection it drops to the bias voltage btw. 8-12VDC with a good sensor connected due to the behavior with constant current supply. So a swing of max +-12V for the signal is about what you can have and some sensors only give about 8. It is not always a 50/50 split of the supply voltage it is a function of a sensor electronics sometimes called "low bias" sensors. So at the other end it influence what you can have space for signal voltage wise, using a 500mV sensor, it will be a lower number of g's.
__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 599
Reply with quote  #15 
Ah yes, Bias Voltage... thank you gentlemen.
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.