0

Where I'm working they use an electronic device to calculate the exact wind speed. One of the parameters besides humidity and temperature they use is the pressure. And they measure it by using the following device. They both acquire/use RS232 and analogue outputs. But they use analogue output for calculations. Is that because RS232 output will have some quantization error?

Here is the device specs: http://www.furness-controls.com/assets/library/documents/PPC500_English.pdf

You might wonder why I don't ask them. It is because some other engineers set it up for them before they also don't know.

JYelton
  • 34,119
  • 33
  • 145
  • 265
user16307
  • 12,049
  • 52
  • 184
  • 320
  • 2
    If there is a microcontroller processing the analog signals, it will convert them to digital too. You get quantization errors either way. Usually you design the system so that quantization errors are less than other noise. It may simply be more convenient to use the internal A/D than to use a UART and level translator. It might also be less power. – Olin Lathrop Jul 25 '13 at 18:10
  • The analog output is connected to a A/D converter by a BNC cable. The voltage output then translated to a corresponding pressure by a software in PC. Do you mean that if they use RS232 there will be 2 times quantization error? Is that the reason about the accuracy thing? – user16307 Jul 25 '13 at 18:12
  • I have no idea why they use the analogue output. – user16307 Jul 25 '13 at 18:15
  • Another thing I forgot to mention is that we use 16 bit A/D card. Maybe this devise RS232 is resolution is not that high? But couldn't find in specs – user16307 Jul 25 '13 at 18:18
  • 1
    I have no idea what your question is, or how that is related to the product you linked to. What I do know is that you are comparing apples to oranges. RS-232 is a communications standard, and does not have anything you would consider to be a "resolution". "Analog" in this case is used to send some sort of "sensor data". They are completely different, and have different uses. I strongly suspect that you are confused on the proper operation and application of this device. Your confusion on this is exactly why I cannot answer your question. –  Jul 25 '13 at 18:53
  • ok i ll explain it again. they send the pressure data to PC by 2 ways: via alanlog output of this device and rs232 output of this device. data i mean the voltage values of both. these data is saved in PC. then in PC by a software the voltage values from both rs232 and analogue output is converted to corresponding pressure units by using a slope and offset. rs232 has different calibration line eq than analog. but the point is they use analog one. and results from rs232 and analog ways they are slightly different . my quest. is why analog output? – user16307 Jul 25 '13 at 19:09

2 Answers2

2

In the general case, there is no answer to your question. Most likely, the RS232 output is digitizing the same analog output, so the question is really this: is the built-in D/A conversion less noisy than whatever external measurement equipment would be used for an analog signal? The answer will depend on both the sensor and the specific measurement equipment being considered.

In this particular environment, I bet the reason the analog output was chosen over RS232 was that it was easier or cheaper, and good enough. Like good programmers, good engineers are lazy.

Phil Frost
  • 57,310
  • 18
  • 145
  • 267
2

I'm not really sure you understand what the RS-232 does. It's only a transport mechanism that sends data using some format that is hopefully documented. It isn't actually related to quantization at all or in any way related to accuracy.

So what we really need to find out is how the device actually performs measurements and how it generates data output. From what I can see by reading the promotional flyer you linked, I'd assume that the device does internal analog to digital conversion at some stage and outputs that information on its display. Don't assume that its analog to digital conversion system as a whole is worse than the PC's and that it's only the number of bits on the ADC that matters. Unless explicitly stated otherwise in the manual (I sure hope that you have it or can at least obtain it!), I'd expect that the calibration and resolution/uncertainty data to relate directly to the data shown on the screen. At that point, you already have one source of quantization noise, but I wouldn't worry too much about it. It's up to manufacturer to provide proper calibration data and make sure that the initial analog to digital conversion quantization noise isn't a problem.

Next we have the output. I'd expect the device to send data which is at least as precise as shown on screen using RS-232, because it already must have that data in a digital format in order to display it in the first place.

I'd also suspect that the analog output is created using digital to analog conversion and that it is based on digital data. That suspicion arose from mention of digits in the accuracy section of the linked document. Unless it's explicitly stated in the manual that analog output is derived from initial analog signal, I'd expect it to come from digital data displayed on the screen. As I mentioned that few times before, do check that in the manual.

Also note that even if the analog output is based on actual analog source, there's no guarantee (unless it's stated explicitly in the manual) that it will be of better quality than the data sent over RS-232.

If the output is based on what I suspect, then you'd have two more sources of quantization noise: Digital to analog conversion at the output and then again analog to digital conversion at the input at the computer.

You said that the computer captures both analog data and RS-232 information. Why don't you do some checking on expected values and see which one seems to be more accurate in your use case? Also note that if you use analog output and it is set up as I suspect it is, it could be creating false precision from the extra noise which comes from the DAC, coaxial cable and ADC on the computer.

In the end, remember that you have the manufacturer of the device too! Why don't you ask them which output type is more precise? The worst that can happen is that they don't answer your question.

Short version: Find a real manual for the device and read it! It should have data you need and if it doesn't, ask the manufacturer. If you can't do that either, then don't touch is already working! Do try to understand it though.

AndrejaKo
  • 23,431
  • 27
  • 112
  • 189
  • thanks for the answer. but thats the reason i wrote here. this crap doesnt have a real info how it works or calibrate itself or other issues. just like coca cola formula. – user16307 Jul 25 '13 at 19:23
  • i think i know rs232 thing as a big picture. its not about that my point was this: they send the pressure data to PC by 2 ways: via alalog output of this device and rs232 output of this device. data i mean the voltage values of both. these data is saved in PC. then in PC by a software the voltage values from both rs232 and analogue output is converted to corresponding pressure units by using a slope and offset. rs232 has different calibration line eq than analog. but the point is they use analog one. and results from rs232 and analog ways they are slightly differen – user16307 Jul 25 '13 at 19:23
  • @user16307 Well this is great way to gain experience: For example, now you know that buying devices with bad documentation is a bad idea. Now about what can be done: It's mentioned that offset of analog voltage can be changed. Could that be affecting the readings? Look for any systematic differences that are affecting large number of readings. – AndrejaKo Jul 25 '13 at 19:35
  • @user16307 Another thing to think about is what's happening later during data processing. For example, let's say that the correct value the device should output is A and that RS-232 outputs B and analog outputs C. Also let's say that B is for example very close to A and that C is not so close to A.If the later data processing after getting the C result produces correct output for A, then it doesn't matter if B is closer to A than C is. So don't disturb a working system. – AndrejaKo Jul 25 '13 at 19:41