Hi, I am using 8-channel 4051 multiplexer,the o/p of this is connected to CS5532(24 bit sigma delta A/D converter). when only a single channel is selected and converted by the adc it works properly. when more than 1 channel is selected(sequentially ch1-ch2-ch1-..)the effect of 1 signal change is reflected on the other channel o/p as well and vice versa. To avoid this as per the application note i am already rejecting the first five conversions to flush out the residues of previous channel,but this does not solve the problem. please help me with your ideas /suggestions to understand/solve this problem. sujit
OK. So the Hardware is wrong. Thanks for solving part of my problem. But can you please suggest an alternative design or A/D converter. Here's what my problem is. I need to sample 4 analog channels sequentially (ch1-ch2-ch3-ch4-ch1-ch2...) all of which are weigh scale signals and based on this i need to do some set-point based output control.(channels and thier control are independent of each other) I had till this point tested the complete application for a single channel and it worked perfectly but the multiplexing part didn't work. I Know there is a 4 channel A/D conerter from the same family(CS5534), but i was not able to get it out there. Can you suggest any other A/D converter which can be used in this application(with internal mux or with compatible external mux). a 16-bit ADC might also be sufficent. thanks, sujit
I had till this point tested the complete application for a single channel and it worked perfectly but the multiplexing part didn't work. So if you were only sampling a single channel, signal changes in the other three channels had no effect on the one channel you were sampling ? Can you suggest any other A/D converter which can be used in this application(with internal mux or with compatible external mux). a 16-bit ADC might also be sufficent. Have a look at Analog Devices' ADuC8xx series.
Can you suggest any other A/D converter which can be used in this application(with internal mux or with compatible external mux). a 16-bit ADC might also be sufficent. SILabs has what you want inside the micro http://www.silabs.com/tgwWebApp/appmanager/tgw/tgwHome?_nfpb=true&_pageLabel=interactiveGuide&col=adc1 There are other '51s as well with mux'ed A/D built in check Keil device database http://www.keil.com/dd/parm_search.asp For external A/D and multiplexers, check e.g. Maxim, Analog Devices, Linear Technology. Erik PS, If you are new at this, be ready for some real work. If you, with the best first 4 or 6 layer layout you can do, get to 11 bits of precision (not resolution) within one week consider yourself extremely lucky. When I did weigh scales the engineer that did nothing but analog front ends for 10+ years considered himself lucky when reaching 15 bits in the high priced models. PPS: if you have any hope of getting above 12 bits (again precision, not resolution) with 'economy' (no 4 layer board etc) I am sorry, but I have to crush that hope.
Thanks for the replies. So if you were only sampling a single channel, signal changes in the other three channels had no effect on the one channel you were sampling ? Yes.That's right. When i was sampling only a single channel fixing the multiplexer inputs for a particular channel, there was no effect of the other channels on the sampled channel. I will check out the ADuC8xx series suggested by you. Although , it will be really helpfull (and time saver) , if i am able to find any suitable front-end multiplexer for the present configuration replacing the 4051. The reason why i am using(reluctant to change) the CS5532 is that i have worked with the chip previously also and have already written the spi interface and the register configration part. I googled extensively , but was not able to find any example / application note showing the chip used in externally multiplexed configuration. SILabs has what you want inside the micro I am presently using the atmel AT89S52 microcontroller.I am not sure what all changes i will have to make to accomodate for the microcontroller change.that seems a little difficult to me. As for the precision /resolution part , i don't think it will be a problem for the present application as i am already able to get a stable count for the single channel.(all are identical channels.The Sampling rate may cause / introduce problems though when i start switching channels) Thanks again for your replies. I will try out the ideas/suggestions/links and post any progress i am able to make on this. PS: Erik, i am really a newbie to this. could you please explain (with an example if possible) , the difference between precision and resolution.Thanks.
As for the precision /resolution part , i don't think it will be a problem for the present application as i am already able to get a stable count for the single channel if the stable count is off, where is the precision could you please explain (with an example if possible), the difference between precision and resolution.Thanks. Ok, to make it easy we make scale that is full scale at 65.536 kg. You put a weight of 30,270 kg on the scale and get a count of 30127, when you put a weight of 30.271 kg on the scale you get a count of 30128. The difference between 30,127 and 30,128 is the resolution (1g), the difference between 30.270 and 30.127 (143g) is the precision (error) the resolution is 'automatic' the precision is a ***. A scale that is precise to 5 decimal digits will cost you a minimum of $10.000 a scale with a resolution of 6 decimal digits can be had for $100 Erik
Eric, I am new to development using the CS5532 for a weighing scales and would be greatful if you could guide on how about going with it. I am using a 1mv/V tedea huntleigh loadcell with a total capacity of 600gms and looking to measure 10mg. ie a total of 60K external counts.
thnks Singh
if you could guide on how about going with it. I am using a 1mv/V tedea huntleigh loadcell with a total capacity of 600gms and looking to measure 10mg. ie a total of 60K external counts. that is 16 bits of precision, it IS achievable, the cheapest industrial scale I know of that is that precise cost about $8000 and the manual states "calibrate before each weighing".
To get to 16 bits will probably require a 6 or 8 layer board, an autocalibration wiight and circuit. a thermal chanber for some of the components, 4 to 7 ultra stable power supplies, a lot of isolation between analog and digital .....
as I said in an earlier post PPS: if you have any hope of getting above 12 bits (again precision, not resolution) with 'economy' (no 4 layer board etc) I am sorry, but I have to crush that hope.
Erik