PCB 393B04 IEPE Accelerometer + NI

[hyprfrco]

Dear Community,

I mainly want to ask if somebody have worked with the PCB ICP (IEPE) 393B04 Accelerometer with NI equipment. 

I am requiring the following performance specs: 

- 1000 ft max, 100ft avg distance from accelerometer to the module,

- 200 Samples/s/ch.

I am still undecided on which equipment to use for monitoring these units, and mainly i am considering the following hardware setup:

- DAQ Modules: One of the NI 9230 (Discont.) / NI 9232 / NI 9234 (which of these should be ok?)

- DAQ: a CompacRIO unit (can a CompacDAQ be used?, which processor should i use?)

- Cable Belden 1829A 16.2 pF/ft, 0.097uH/ft, 18 AWG copper, 0.27 in dia, RG-6 Coaxial cable (should i use another cable? could i use a CAT6 cable?)

Do i have any problem with this selection?. 

 Thanks in advance,

JFA

[niarena]

I have worked with accelerometers before although not with that particular one.

Any of those CDAQ modules can acquire at 200 Samples/s/ch.

The DAQ modules NI 9230 and NI 9232 should be ok, they measure +-30V. The voltage output of the sensor is 7-12 VDC, the NI 9234 range is +-5V, so the NI 9234 will not allow you to get the full range of measurements from the sensor.

Any chassis that supports those modules can be used if it is only for the acquisition, about the processor, it is a matter of the task that you want to implement and features you would like to have, how do you want to process the data? or, what do you want the processor to do? do you have any more specifications for the project?

The RG-6 seems ok to me, what one would need to take into consideration for the cable selection are the voltage drop on the cable, attenuation of the signal, and a proper insulation so the cable does not work as an antenna, it probably needs some signal conditioning to calibrate the signal.

[hyprfrco]

Dear NIArena,

The tasks are:

• Detect peaks on the accelerometers,
• Acquire a queue of data, for storing data "before" the peaks,
• Store the data "after" the peaks,
• Saving the data on disk when finished.

I dont think that should be a major processor consuming task, but it should be done in a reliable way, so i am most decided for the CompacRIO platform. 

Have you worked with IEPE / ICP Accelerometers?. Most IEPE literature available from accelerometer manufacturers (PCB, Endevco, BKSV, MMF) indicates that there is not a need for a signal conditioner. Given the relatively low sampling frequency, and the low coaxial cable capacitance, i "think" the design is ok. 

Instead voltage-based protocols, the IEPE standard should be a current based protocol, hence the voltage drop and the EMI should not be an issue "with virtually no loss in signal quality", PCB Ref. The main issue should the signal attenuation due to the cable & connectors impedance and RF losses.

The PCB 393B04 Constant Current Excitation is between 2-10mA. So for feeding the proper current to the IEPE accelerometer, both the NI 9230 and NI 9232 can supply 4.25mA typ. which i think it should be acceptable, and the NI 9234 can supply 2.1mA typ. which could be insufficient?. 

The NI 9230 / NI 9232 are presented as including IEPE signal conditioning. And because the standard signal conditioning are actually amplifying the constant current excitation -which you can see fall in the range with 4mA, i fail to see how an additional SC module should be adding resolution to the system. On the contrary, i would prefer a lower capacitance cable -maybe a CAT6 cable, which reach 12pf/ft normally. 

Most of these are assumptions and still questions I am looking to confirm, for proceeding with the final setup. 

Best regards.

[etuzuner]

Hi hyprfrco,

NI 9234 is very popular in DSA modules or systems. It has a CCS of 2 mA and it has software selectable AC/DC coupling options. It has built-in anti-alliasing filter which is very important while you're acquiring data. For an application like this, you can use cDAQ chassis. If you want to control a plant or sth, then you can think of using cRIO.

Now, here you can find a white paper. You're right to be concerned when 2 mA is used in high distances. If you're sure that you won't need high frequency response then it seems like you can use NI 9234 due to the white paper. Most of the time, 2 mA supply is enough for me so I always use NI 9234. I'm a fan of that module. It helped me a lot in the past projects. 

If I were you, I would be concerned of the cable first, not the conditioner. You can also find the cables' capacitance values in the white paper. My advice would be to ask PCB for the cable. 

P.S., PCB uses NI 9234 in their video tutorials in Youtube. 

Emre

[hyprfrco]

Emre, thank you for your reply. Due to the quantity of questions i have regarding this application, i will take your answer as antecedents for replying them. I could open new topics and redact independent questions if that subdivision is better for the forum purposes. Again, i appreciate your valuable feedback:

NI 9234 is very popular in DSA modules or systems. It has a CCS of 2 mA and it has software selectable AC/DC coupling options. It has built-in anti-alliasing filter which is very important while you're acquiring data. For an application like this, you can use cDAQ chassis. If you want to control a plant or sth, then you can think of using cRIO.

1. I am confirming with you that a cDAQ is enough for the application, because i dont need anything more complex than:
1. acquiring the data,
2. saving the data (pre + post event) when a vibration is detected, and
3. maintaining a data time series in memory for building the pre-event.

Now, here you can find a white paper. You're right to be concerned when 2 mA is used in high distances.

If you're sure that you won't need high frequency response then it seems like you can use NI 9234 due to the white paper.

1. The paper is confirming me these references (PCB, Endevco, BKSV, MMF), which impose a limit enough superior to the 200Hz i am currently requiring,
2. Also, the paper is confirming me that it is indeed feasible -though not common- to employ a 1000 ft distance, through the proper cable.

Most of the time, 2 mA supply is enough for me so I always use NI 9234. I'm a fan of that module. It helped me a lot in the past projects. 

In here, i am having these fact/concerns:

1. The accelerometer Exc. (IEPE) Current Range is 2-10 mA,
2. I will repeat this here: The Cable Length L, Maximum Frequency f, Cable Capacitance C and  Voltage Range V, are expressed through a similar relationship provided by most IEPE Manufacturers (PCB, Endevco, BKSV, MMF) as L=(I-1) / (2 pi f C V),
3. Based on the previous relationship, and also in the white paper, i shoud select the maximum Exc. Current possible, in this case, 4 mA. Hence, the module choice between the NI 9234 (12.8 kS/s/ch, 5V, 2 mA) and the NI 9230 (12.8 kS/s/ch, 30V, 4 mA) or the NI 9232 (102.4 kS/s/ch, 30V, 4 mA) should be favouring the NI 9230 (I know that you are suggesting me the NI 9234, though 🙂 ),
4. The NI 9234 is 5 V Voltage Range, and the NI 9230 / NI 9232 are both 30 V Voltage Range. Which is the interpretation of this voltage?. I have the following figures:
1. Accelerometer V Range, Sensitivity & g Range): +-5 V, 1000 mV/g, +-5 g,
2. Accelerometer Exc. (IEPE) Voltage: 18-30 V,
3. Accelerometer Bias Voltage: 7-12 V, Hence, the Accelerometer Voltage should vary between 2-17 V (Is this expression Vaccel=Vbias+Vrange+Vnoise correct here?),
4. The accelerometer Voltage of 2-17V is less than the DAQ IEPE Compliant Voltage of 19V for NI 9234 and 22 V for NI 9239 / NI 9232 (It is ok to compare Vaccel and Vcompliant in that way?).

If I were you, I would be concerned of the cable first, not the conditioner. You can also find the cables' capacitance values in the white paper. My advice would be to ask PCB for the cable.

Thank for your confirmation. At this moment I am considering these cable:

1. Belden 1829A, RG-6 Coaxial, 75 Ohm, 16.2 pF/ft, 0.097uH/ft, 18 AWG Solid Copper, PVC Jacket, 270 in Overall dia, 28 Ohm/1000ft Conductor, 9.0 Ohm/1000ft Shield, 27 lb/1000ft, 0.151 US$/ft,
2. Belden 9258, RG-8X Coaxial, 50 Ohm, 24.8 pF/ft, 0.065 uH/ft, 16 AWG Stranded Copper, PVC Jacket, 242 in Overall dia, 4.3 Ohm/1000ft Conductor, 3.3 Ohm/1000ft Shield, 37 lb/1000ft, 0.988 US$/ft,

PCB is recommending me their manufactured cable:

1. PCB 002, General Purpose Coaxial Cable, 50 Ohm, 29 pF/ft, 30 AWG Stranded Silver Plated Copper Covered Steel, FEP Jacket, 0.075 in dia, 6.25 lb/1000ft US$1.95/ft

My main facts/concerns are the following:

1. The PCB cable is several times expensive than the others (Is there any parameter that is strictly required for selecting the PCB cable and not the Belden 9258?)
2. The white papers uses the following three cables: RG 62 (93 Ohm), RG 58 (50 Ohm) and RG 174 (50 Ohm). The RG-6 (75 Ohm) is the cheapest, and the recommended equivalent RG-8X (50 Ohm) is 6.6 times expensive.
3. The Accelerometer Maximum Output Impedance is 500 Ohm. Considering the IEPE sensors are low impedance voltage output, How much important is the impedance matching for the IEPE protocol, for considering 50 ohm and 75 ohm or another cable impedance?.

P.S., PCB uses NI 9234 in their video tutorials in Youtube.

I will check them soon. That will take me some additional time.

Thanks in advance.

[etuzuner]

Hi hyprfrco, 

I didn't use NI 9230 or NI 9232, they can be said to be newer products compared to NI 9234. They have good dynamic range which is very important for environmental or signal noise. They have screw terminal option, NI 9234 does not. I don't know why the analog input range is +-30V but I'm sure that NI produced for a reason (I haven't used any but there may be other IEPE sensors having an output range like that). I also develop analog PCBs and I've never seen an ADC IC with having analog input range +-30V.

If a module is said to be an IEPE conditioner, you don't need to think about the compliance voltage so much. Probably it is produced to comply with almost all of the sensors but of course you should be careful anyway. For example, take a look at the this datasheet. You can find the IEPE compliance voltage at page 10. Also you may want to take a look at the comparison table for C series DSA modules at page 2.

The reason why I told you to ask PCB is that long distance for a cable like that means an antenna behavior. If we're talking about mVs to be acquired, then a trusted cable should be selected. If you trust that producer, then of course you can use. I always used the producer's cable. When I had problems on the field, I never suffered from a noise caused by the cable. That made me comfortable all the time. Again, it is a matter of choice. 

I've never faced a situation that was caused by impedance matching. I think you're right, it should match for maximum power transfer but I don't have an answer for a question of how much. I guess a person who is doing analysis for tranmission lines would answer better than me.

Emre

[Henrik_Volkers]

I work a lot with accelerometers 🙂

Long cable (Do you REALLY NEED the >300m cable?)  but not much bandwidth ...

How many channels do you want to read?

Both the 9232 and ..34 will do the job.   but:  Long cable want more current (capacitive load) and the source impedance is a function of the sourcing current.  (URI 🙂  about const. bias voltage/ bias current -> source resistance 🙂

The ranges of both cards will fit, and you will not destroy your sensor (not with the bias current sources of you cards, a thunder strike nearby with long cables is much more effective :>)

 So I would vote for the ..32 .   However you can always ask your local NI sales rep to test both in direct comparison.

For the cable: At 100ft a good shielding and a low cap/ft is important.   A cat6 cable would allow a double shielding (one side connection!)  for the cost of a higher cap load. For a IEPE you don't really need a strict coax cable...  but usually you can get them with a better shielding. )

EDIT: If you need more supply current you can add a signal conditioner or just add another 10+mA current source with a Umax 19V (below your original source Umax .....)    it's a hack and I guess it's not needed, however IEPE allow such hacks 😄

[hyprfrco]

Hello Henrik 

Long cable (Do you REALLY NEED the >300m cable?)  but not much bandwidth ...

Yes. This is a big structural building. Perhaps i could distribute the acquisition task between two DAQs, thus reducing the lengths to a maximum of....... 250m.

How many channels do you want to read?

32 channels . Or 16+16.

Both the 9232 and ..34 will do the job.   but:  Long cable want more current (capacitive load) and the source impedance is a function of the sourcing current.  (URI   about const. bias voltage/ bias current -> source resistance

NI 9230/9232 have 4mA IEPE constant exc. current and the accelerometer IEPE current range is 2-10mA.

I am correctly assuming that the cable will not consume more than 2mA in capacitive effects (that would be a huge!)? and thus, i am correctly infering that the current then should be enough for any of these modules?

(BTW, what is URI ?)

The ranges of both cards will fit, and you will not destroy your sensor (not with the bias current sources of you cards, a thunder strike nearby with long cables is much more effective )

Good advice. Everything is indoors, so i will not have so many thunders around. I guess.

I've been talking (I was. They never replied) with my NI office and i am still unclear of what is the meaning of +-30V | +-5V  on the NI 9230/9232 | NI 9234 respectively, considering that both of them have IEPE support, and a IEPE compliance voltage of 22V | 19V respectively. Is there somebody from NI to answer this?.

So I would vote for the ..32 .   However you can always ask your local NI sales rep to test both in direct comparison.

Waiting for the accelerometers to be manufactured in order to make those tests.

For the cable: At 100ft a good shielding and a low cap/ft is important.   A cat6 cable would allow a double shielding (one side connection!)  for the cost of a higher cap load. For a IEPE you don't really need a strict coax cable...  but usually you can get them with a better shielding. )

Can i use a twisted pair cable instead a coax?. Both them have the same capacitance, but with obvious impedance and ground capacitances differences. Which at this moment are not too clear (reserved for a next weeks research). The 70 ohm coaxial is really cheap, and with good capacitance value, so, can i disregard the 50 ohm cable and use a 70 ohm coaxial?

EDIT: If you need more supply current you can add a signal conditioner or just add another 10+mA current source with a Umax 19V (below your original source Umax .....)    it's a hack and I guess it's not needed, however IEPE allow such hacks

[Henrik_Volkers · Accepted solution]

The cable isn't consuming  the current 🙂 (well, at low frequencies..)  but it is a capacitive load to the output of the IEPE sensor.

URI means U=R*I .. the Kirchhoffs law.

In a simple approach the output impedance of the IEPE circuit is R_output=U_bias (~12V)/I_iepe(2mA) = 6kOhm.

For 10mA Biascurrent  it's only 1200 Ohm ...

More on the effects of cable length to IEPE current can be found in application papers from PCB, Endevco, B&K ...

A simple simulation yields:

The cable capacity is asumed to be 5nF, 2mA IEPE current ... at 100Hz about 1° phase error...

(with 10mA,1200Ohm 0.2° phase deviation)  due to the long cable....

I have no experience with long cable, multisensor  setups... I 'just' calibrate accelerometers in a lab 😉

[obenozkan]

I had the opportunity to compare the NI-9234 and NI-9230 modules at the same sampling frequency. I received data from BK 8339 accelerometer at 12.8 khz then I applied SRS(shock response spectrum) transform. NI-9234 performed better across a wide frequency range. And I decided continue with 9234.