Need help designing and choosing an ADC!

[AmeliaBuns]

Hi, I started using an hx711 as they're common for my projects, but now i'm looking at something faster and better. I also have some noise issues no amount of shielding is solving! my power supply had 3mv~ noise on the internal regulator.

I assume a delta-sigma is best for my application? I Don't want to go past 8$! (I get my parts from LCSC)
I want a sample rate of at least 1KSPS and a resolution of 24 bit (I'm just saying that off the top of my head as my hx711 is like that)
I'm basically measuring a load cell that's just two resistors on a PCB (They work surprsingly well! who knew simple 2512 resistors can be used as strain gauges!)
the hx711 is great so far, but it's so noisy and very slow (80hz)

I came down to two options, the ads1220 and ads1255/1256. but reading the datasheet, I realized i'm not experianced enough or knowledgable to choose one properly. any ideas/options?! should I grab one with an integrated PGA or just amplify the wheatstone bridge myself with an OP-AMP?

this is the board i'm using atm, https://universal-solder.ca/product/hx711-scale-amplifier-module-shielded/ . I noticed no difference with or without the shield, and even grabbing some copper sleeving for the cables and grounding it had no effect (Simply conneceted it to my hx711's ground input. I also need to compensate for temprature, but for now I figured ICs can't help with that, and i just have to put a sensor next to my gauge, or maybe see how hard and expensive it is to build an eddie current gauge? (tho that might also be subject to temp drift)

Thanks

[moffy]

If you could specify the voltage range you are trying to measure and the impedance of the bridge and tolerances, as well as your required resolution, it might be better to go for a low noise instrument amp/amp with gain, and a lower number of bits. With ADCs you normally trade of resolution for speed i.e. enob.

[AmeliaBuns]

If you could specify the voltage range you are trying to measure and the impedance of the bridge and tolerances, as well as your required resolution, it might be better to go for a low noise instrument amp/amp with gain, and a lower number of bits. With ADCs you normally trade of resolution for speed i.e. enob.

That's the embaressing confession, I'm just a hobbyist trying to make a sensor I came up with that'd be super cool if it works, but yeah. I'm fine with 24v-3.3v as my wheatstone bridge is just 4 1w 2512 resistors, and the power supply can be both 24v in or 5v in. 1KSPS is good for me. I could go for 2ksps but that'd not be a huge change. But i honestly am not sure!
I figured more bits are better if cheap enough? and 24 bit technically has enough resolution, as long as it's not noisy.

Oh and i Forgot to mention, the resistors are 1kOhm each but I'm more than fine enough with switching them for something else!

[moffy]

If you could specify the voltage range you are trying to measure and the impedance of the bridge and tolerances, as well as your required resolution, it might be better to go for a low noise instrument amp/amp with gain, and a lower number of bits. With ADCs you normally trade of resolution for speed i.e. enob.

That's the embaressing confession, I'm just a hobbyist trying to make a sensor I came up with that'd be super cool if it works, but yeah. I'm fine with 24v-3.3v as my wheatstone bridge is just 4 1w 2512 resistors, and the power supply can be both 24v in or 5v in. 1KSPS is good for me. I could go for 2ksps but that'd not be a huge change. But i honestly am not sure!
I figured more bits are better if cheap enough? and 24 bit technically has enough resolution, as long as it's not noisy.

Oh and i Forgot to mention, the resistors are 1kOhm each but I'm more than fine enough with switching them for something else!

What are the values and tolerances of the four 2512 resistors, I assume you are interested in the difference voltage between the two legs of the bridge? What is the voltage range of this difference you are trying to measure, you don't have to be precise but is it of the order of +/- 1mV, +/- 10mV etc, here I am talking about the signal level not the offset due to tolerances.

[selcuk]

I can recommend NAU7802 as an affordable alternative for load cell applications. It is very reliable, has 24 bit resolution but isn't that fast. You may even need to decrease the speed to lower the noise.

https://www.nuvoton.com/resource-files/NAU7802%20Data%20Sheet%20V1.7.pdf
https://www.digikey.com/en/products/detail/nuvoton-technology-corporation/NAU7802SGI/2769782

Load cells need some stabilization time. When you drop a load on to the cell, the output fluctuates for a while. Maybe this is due to the mechanical setup. You normally wait for it and don't require high sampling speeds. I think your resistive bridge is more responsive than this so there may be a difference than load cell applications.

[AmeliaBuns]

If you could specify the voltage range you are trying to measure and the impedance of the bridge and tolerances, as well as your required resolution, it might be better to go for a low noise instrument amp/amp with gain, and a lower number of bits. With ADCs you normally trade of resolution for speed i.e. enob.

That's the embaressing confession, I'm just a hobbyist trying to make a sensor I came up with that'd be super cool if it works, but yeah. I'm fine with 24v-3.3v as my wheatstone bridge is just 4 1w 2512 resistors, and the power supply can be both 24v in or 5v in. 1KSPS is good for me. I could go for 2ksps but that'd not be a huge change. But i honestly am not sure!
I figured more bits are better if cheap enough? and 24 bit technically has enough resolution, as long as it's not noisy.


wait, doens't the tolerance not matter? I thought it only measures the initial resistance, so worst case scanario is an initially different scale or different change ratio? It's tared eitherway.
Oh and i Forgot to mention, the resistors are 1kOhm each but I'm more than fine enough with switching them for something else!

What are the values and tolerances of the four 2512 resistors, I assume you are interested in the difference voltage between the two legs of the bridge? What is the voltage range of this difference you are trying to measure, you don't have to be precise but is it of the order of +/- 1mV, +/- 10mV etc, here I am talking about the signal level not the offset due to tolerances.

[AmeliaBuns]

I can recommend NAU7802 as an affordable alternative for load cell applications. It is very reliable, has 24 bit resolution but isn't that fast. You may even need to decrease the speed to lower the noise.

https://www.nuvoton.com/resource-files/NAU7802%20Data%20Sheet%20V1.7.pdf
https://www.digikey.com/en/products/detail/nuvoton-technology-corporation/NAU7802SGI/2769782

Load cells need some stabilization time. When you drop a load on to the cell, the output fluctuates for a while. Maybe this is due to the mechanical setup. You normally wait for it and don't require high sampling speeds. I think your resistive bridge is more responsive than this so there may be a difference than load cell applications.

Oh I did take a look at that! However 320SPS is too low for me. I need at least 1KSPS!

[zapta]

Is this for a loadcell?

Last year I designed an interface for high rate high accuracy load cell.  The schematic is below. It was a plugin board with 3.3V SPI interface and 5V supply rail.  It also has three thermistor/PT1000 inputs which you can drop.  For low noise you want to excite the loadcell and the recommended voltage (e.g. 10V) since its output is linear with the excitation voltage. The board also has an optional Kelvin topology to reduce the error due to the excitation current voltage drop.

https://github.com/zapta/daq/blob/main/adc/kicad/adc.pdf