A1S64AD output problems

[pirate 0]

I have a –1 to 20bar pressure transmitter which sends a 4-20mA signal to a Mitsubishi A1S64AD 4 channel Analog to Digital conversion Module. I have managed to write all the “correct” script to initialise the module (at a 1/4000 resolution which gives me a raw data figure of 2000 at 20mA) and the math that converts this to milibar data which can be read from the SCADA. I have checked the accuracy of the set-up using a 4-20mA source across the whole range and that is all good. Although I am not using the other 3 channels at the moment I have set them all up identically to see if the problem exists when I switch channels and it seems like it does. When I go live with the system it seems that the signal is all over the place and every once in a while the correct readings are produced. I initially attributed this to fast fluctuations in the pressure (pulses) which could be possible in our system. So I fitted a little mechanical damping device. Did not make too much difference. Then I checked the movements using a meter in series across the 4-20mA, although it does fluctuate it does not seem to fluctuate that much. I have tried changing the resolution but with no great success. I have also changed the sampling rate and averaged the signal over a set amount of readings however the average seems to be way too low when I get to a setting which produces fairly steady result. I have to admit that the example given in the A1S64AD programming manual leaves a lot to the imagination and I may well have gotten the last line of code wrong. One thing I have not tried is to average the signal across a period of time. Now although this signal is OK for display purposes I will not be able to use it for actual control. Maybe I should give up and use the A1S68AD? Does anyone now of a physical device that I can fit in the 4-20mA signal that will buffer (smooth)the output? Maybe I have the code wrong can anyone help check this for me? Is there something I’m missing? Anyone HELP! Many Thanks

[panic mode 246]

there can be several things: - wiring - noise - sampling etc. the first two are obvious and can be easily tested - just bring known value of current into input (can be set by potentiometer for example or little analog checker circuit or multimeters designed for use in instrumentation). noise can be filtered out by use of capacitors accross input for example. if the reading is still not stable, it's time to look into cards itself. analog card converts analog signal into digital. as mentioned in manual, this conversion takes some time (depending on used AD method). when you read 'result' from analog card, it may not represent correct value. you should only consider readings when 'conversion complete' flag is set (this flag comes on and off) or results of unfinished conversions will look 'bumpy'. normally you would use FROM instruction to read all analog values and the conversion status into temporary strage (set of D registers for example). then you evaluate conversion status bits and if conversion for particular channel is complete, copy the reading from temporary to another D register. this will be your actual analog input that you will scale into engineering units etc. manual also says that single conversion is about 0.5ms so if you enable three channels (out of four that your card has), you will wait about 3x0.5=1.5ms. this is still pretty fast but as mentioned, if you get the reading in wrong moment it may not be correct value since conversion may be incomplete. also this is maximum conversion speed, if the CPU keeps asking card for results, actual conversions will be a bit slower. manual for this card is IB(NA)66576 and current version is "F" or IB(NA)66576-F. i find it very straight forward and the whole code (including init, reading alarms, and reading analog inputs) is on just one page (three rungs). the last rung does just what i've explained - take into account good readings and discard incomplete ones. there is no point to replace your card, specially for same one that only has more channels. good luck

[pirate 0]

OK so I finally figured it out after much fiddling. Seems the 24vdc supply was not man enough and was only producing around 20v. Hours and Hours Go figure Anyway thanks for your help and valued input.