Jewellery scale noise problem

hi,

i am building a jewellery scale.
i am looking at 600 gm / 10mg machine.

my load cell is 1mv/v.
i am amplifyng the diffeential signal with ina114 with gain 10. this is fed ltc 2410. am reading the data at 8 hz. i am directly connecting the load cell out put to the ina114.

i am facing couple of problems as follows.

1. If i put x gms weight and remove the weight my system doesnt come to zero everytime.
2. if i put x gms weight i expect the system doesnt give same reading again its in the +/- 3 units.

Can anyone help me out in solving this problem ?

Regards

Anil

how is loadcell mounted?

hi,

The load cell is fitted on to a base firmly.

regards

anil

anil_asdaq said:

my load cell is 1mv/v.
i am amplifyng the diffeential signal with ina114 with gain 10. this is fed ltc 2410. am reading the data at 8 hz. i am directly connecting the load cell out put to the ina114.

i am facing couple of problems as follows.

1. If i put x gms weight and remove the weight my system doesnt come to zero everytime.
2. if i put x gms weight i expect the system doesnt give same reading again its in the +/- 3 units.

Click to expand...

Why do you have the ina114 in there? The LTC2410 is a fully differential part and at 24 bits has extremely fine resolution, so there is no need for gain. As long as you have lot capacitance between the load cell and the LTC2410, you're fine. Op amps bring with them a number of additional errors, on a low signal level like this it makes things worse.

Looking at the spec sheet for ina114, it has an offset error that begins at 4uV on powerup and decays with time, disappearing 60 sec after turn-on. I have never encountered this type of error specification before! This sounds like the source of your problem.

What you describe could also be a hysteresis error in the load cell. Are ANY surfaces on the weighing table touching something else in any direction? The load cell should be the only thing touching the table. If it's in the cell itself, I don't see how to fix it unless you throw in a pager vibrator to shake off the hysteresis error. That would be pretty weird.

Hi,

How do you suggest me to connect the load cell to the LTC2410. Do i connect directly or RC filter.

I am trying with a LC filter with a ferrite bead and capacitor.

My power supply is derived from AD3330-5. I tried LM7805 and LM2931. With AD3330 the performance dint change much.

What do you think could be the source of my problem?

Regards

Anil

hi oznog,

i am observing that putting any capacitive filter rc or lc on the circuit (infront of ina or ltc) the performance is going bad.

how can i take care of the drifts or noises on the pcb ?

regards

anil

Apart from low pass filtering the load cell signal, are you doing software averaging in your scale? This can lower down noise effects considerably.

anil_asdaq said:

hi oznog,

i am observing that putting any capacitive filter rc or lc on the circuit (infront of ina or ltc) the performance is going bad.

how can i take care of the drifts or noises on the pcb ?

regards

anil

Click to expand...

Right, you can't do this with a sigma-delta ADC. Capacitance adds error to a high impedance input, and your strain gauge is high impedance. Look at Fig 17-26 of the LTC2410 spec. I also learned this the hard way, RC filters are not suitable.

The LTC2410 was designed with an input stage that filters out high freq input noise. Look at Fig 39-43. The highest freq the part will respond to is like 15 Hz. So it seems unlikely you will see substantial noise.

Do not speed up the conversion clock F0. It speeds up the conversions but only with a decrease in accuracy. The loss is in conversion accuracy is greater than what can be gained by averaging the additional conversions you'll be able to pull off.

Be sure to tie the Vref to the power for your strain gauge. The strain gauge is ratiometric, so any errors cancel out. If the supply voltage dropped by 5%, the differential output voltage of the sensor drops by 5%. But if you tie Vref like this, the converter will read 5% higher due to the lower Vref. So it works perfectly.

Hi,

I am using software filters to show the best result. The display is quite stable. By manupulating the data we can do many things.

But the point is to make the signal stable.

Oznog, I connected the Vref to the source of excitation for the starin gauge as you explained.

i connexted F0 to Vcc i.e. in internal oscillator mode. with 50 hz notch filter.

I am acuiring 7 Hz data. I am using moving average filters to stabilize the data. Till then is fine. What else do you think i must adopt to make the system better ?

I wonder we cant achieve all the 18 stable bits. Keithly is making good nano voltmeters. I dont know how these guys are making such standard nano voltmeters. They must be doing some magic.

What should i do ?

Waiting for ur reply

Regards

Anil

anil_asdaq said:

Hi,

I am using software filters to show the best result. The display is quite stable. By manupulating the data we can do many things.

But the point is to make the signal stable.

Oznog, I connected the Vref to the source of excitation for the starin gauge as you explained.

i connexted F0 to Vcc i.e. in internal oscillator mode. with 50 hz notch filter.

Click to expand...

Are you in a country with 60 hz or 50 hz power? 50 hz notch is for countries with 50 hz power.

Have you checked to see how much accuracy even 18 bits corresponds to? This is still a lot. There are a lot of reasons it's hard to get sub-gram accuracy.

Don't be afraid to call Linear's tech support. They're quite knowledgeable and easy to get ahold of.

hi oznog,

ya u r right i am from india here power is supplied at 50 Hz.

I am taking 18 bits out of the 28 bits. But the 18 bits arent stable.
I wrote mail to the linear ppl. lets what happens ? i will update you with their reply.
what do u think how ppl are making nanovoltmeter ?
If we solve all these problems we will be with another nanovoltmeter.

Mail me back...

Regards

Anil

anil_asdaq said:

hi oznog,

ya u r right i am from india here power is supplied at 50 Hz.

I am taking 18 bits out of the 28 bits. But the 18 bits arent stable.
I wrote mail to the linear ppl. lets what happens ? i will update you with their reply.
what do u think how ppl are making nanovoltmeter ?
If we solve all these problems we will be with another nanovoltmeter.

Mail me back...

Regards

Anil

Click to expand...

Actually I've got a similar problem. I built a high current ammeter with the LTC2411 using a 50mV@200A shunt. There is noise so the readings are up/down by about 200mA, with the 2.5V vref I use, that's 50uV of noise or 1/50,000th of the full scale, which is less than 16 bits of accuracy. If you average the last 10 readings or so the result gets much more accurate.

I cannot figure out what the noise comes from, the shunt is not connected to power. It's not supposed to respond to noise over 60Hz and it's hard to get noise this low in freq. Low freq do not couple readily into wires and there aren't many sources of EM this low to begin with.

You do know how to do a rolling average from a buffer with a part this slow, right? Rather than collecting 10 or 20 samples (which takes awhile) and displaying a result, the part is always collecting samples and putting them in a buffer. When the system wants to update the display, it reads back the last n entries and averages them immediately.

hi oznog,

i have replaced the tedea huntleigh load cell with another load now the system is performing as it should. now the final problem is repeatability.
infact my system is repeatable +/- 3 counts.

i have seen a system which uses 20 pin ic, guess is an ADC,with 4 mhz crystal btw 9 and 10 pins. do u know this ADC?

In this he is capacitive filters. I am gonna implement this and will verify how it is going to help.

I had a talk with one of my friends who opened keitly's nano-voltmeter he says there are about 1500 dicrete components on the pcb. he says there is no microprocessor on the pcb. then they must be really doing some magic.

my averaging algo is sth like this

lets say OUT(n) output of adc at nth instant
FILTER(n-1) is output of filter at n-1 th instant

then my filter out put will be
FILTER(n) = (0.3 * OUT(n)) + ((0.7 * FILTER(n-1)).

I havent worked upon shunts. The shunt on which i worked has got three terminals. This i used in my energymeter project. the accuracy of shunt is much better than CT. We used 500 micro ohm sunt. We used AD7751 ic in the energymeter project. checkout if it helps you.

If you know the 20 pin ADC which works on 4 MHz crystal. please let me know. It takes differential input btw pin 1 and pin 2. crystal is connected btw 9 and 10 pins.

Regards

Anil

Hi Anil,
The 20-pin chip you are talking about may be a PSoC Mixed Signal Array from Cypress. I have seen quite a few of them in 20-pin package which has 14-bit ADC and 8-bit microcontroller in-built.

CY8C27243 seems to be a close match to your description.
Try having a look at its datasheet.
Pin 1 & 2 are analog IOs
Pin 9 and 11 are crystal pins.

I am not 100% sure whether its the same IC that your are talking about but its one option that came to my mind after reading your post.

kinjlap,

thats not th the adc.

i m doing good.

regards

anil

Are you on a stable surface? Could air currents be affecting it? Or could it be a thermal error as temps slowly change? This would be indicated by the readings going up by 3 counts every time you read it for awhile, you know, long time periods in the trends.

Sounds like you have a mechanical friction (maybe it could be called hysteresis) problem. There's little you can do about that, frankly. You may have hit the limits of that sensor. But it could also be the result of anything other than the load cell touching the weighing plate. Even lateral contact with a guide or whatnot will support some of the vertical force through friction.

This is why scales which measure 0.1 grams or below are considerably more expensive.

Burr-Brown (division of Texas Instruments) is the other big maker of ADC chips. They do plenty with external xtals.

I don't think your filtering strategy (that 0.3x + 0.7y) is valid. It does not filter well. That ADC reads something like 10 readings/sec. A 1 or 2 sec period isn't bad, but you want to do continuous updates. So you set this thing up to read the ADC as fast as it converts, put them in an endless buffer of the last 20 readings, and every 0.5 sec update the display with the last 20 readings for a 2 sec averaging period. The display update period and averaging period need not be the same.

What were you planning to do for a capacitive filter? You do remember the prob with sigma-delta ADCs and cap filters, right? What problem are you trying to solve? The only thing you have left is repeatability.

Hi Oznog,

Even after the 0.7x+0.3y, i used lot of mathematics after studying the systems behaviour. I dont have any repetability problems now.

The problem left now is beacuse the vibrations.

When i disturb the table on which my system is put, my system also vibrates as the table vibrates, so the signal varies. so my diplays shows whatever it sees. obviously display oscillates.

i have seen in many instruments, theie display doesnt oscillate at all even when hit i the table.

i realized by putting capacitor, the systems perfomance decreases. so i cant put a capacitive filter.

I am meditating on how to solve this problem.

What do u say ?

Regards

Anil