Hi everyone, I am busy with building a milligram scale with the end goal of being accurate to 0.005g. The components I bought came from my local electronic hobby shop. To thus far those components did ok with the desired accuracy however the HX711 module has a fair amount of noise and the excitation voltage of the load cell keeps varying under no load.
This got me thinking of the quality of the products that I am using. How reliable are brands like Sparkfun, Polulu, DF Robot etc. to use to go from prototype to an actual product? Where do you start at sourcing quality components and what do you look for?
That is not the component's fault. You need to provide a quiet excitation voltage. Batteries are the lowest noise.
How much noise are your reading. What readings do you get if you disconnect your source and short circuit the inputs together?
Do you use a low pass filter and have a capacitive element across the A/D input?
I do not know.
Having worked in many industries for many decades, I know there are teams in companies who work solely on this issue. A recent satellite design of mine was found to be giving a problem due to a short circuit resistor. I was atonished that after all the quality assurance work, pre-assembly inspect etc that a faulty resistor could be fitted to a space standard board!
Firstly read the datasheet so you know whether the component is precise enough for your application - precision is not the same as quality (quality is about meeting the datasheet reliably and being robust and long-lived, nothing to do with performance at all).
Those brands should be reliable, as with the major electronics suppliers - they source components carefully and many suppliers have pro-active policies about detecting counterfeits - their reputations are important to them.
However note that "hobby" components are exactly that.
Which Arduino, and which HX711 board, and how do you power them.
All HX711 boards I know have a 4.3volt regulated excitation supply that only can do it's job if you power it with (clean) 5volt.
Read the 'how to post" guidelines.
We need more info to properly advise (diagram, picture, code).
Post a link to the load cells. I think they shouldn't be more than 100gram FS for your required resolution.
Leo..
Companies like Sparkfun, Pololu and Adafruit buy genuine parts from certified distributors and stand by their products.
On the other hand, anything you buy from Alibaba, DFRobot, Aliexpress, etc. is likely to contain counterfeit, fake, subquality or reject parts. The boards are cheap for a reason, but they might work to your satisfaction.
I picked up a few issues with my setup in the mean time. The load cell that is was using measured different resistance value across the leads. I used a 9V battery as suggested and the excitation voltage of the loadcell was never constant under no load conditions. I replaced the load cell with new one and then measured the same resistance across the leads and when a voltage was supplied to the leads the excitation voltage remained constant under no load conditions.
With the new load cell there was still some fluctuations in the readings. This could be the power supply not being constant. I have power supply unit but I am not sure if this is a reliable constant source of supply voltage that I require?
I currently have a 100uF capacitor across the A/D input. I still need to add a low pass filter. I am reading up on this and trying to figure out the size of the capacitor and resistor to be used. From my understanding the component sizes are driven by the frequency that need to be filtered. How does one go about determining this?
Who said 9volt.
That could have given 4.5volt on the load cell output terminals,
and that would have been outside the common mode range of the HX711 inputs (1.2 to 3volt).
The HX711 couldn't have worked with that.
Ouch. Hopefully not to ground.
You still didn't post the required information to properly help you.
Leo..
My intention with the original post was more to get information with regards to sourcing components that are reliable and not so much to dive into the psychical project itself, apologies for the vague answers in the previous replies.
Here is the current circuit layout of my project with regards to the load cell and HX711. I am using a Arduino Nano which is powered by a step-down converter supplying 7.5V to the Vin pin. The converter is supplied with a 12V power supply which is required to run a DC motor not illustrated on the schematic. The weight from the load cell is displayed onto a Nextion touch screen which in turn also controls the DC motor as well. Up to now I have used this circuit layout as a starting point to be able to develop the rest of the project.
The inaccuracies of the load cell was not an issue up to now, however the project has come to a point where I now need to get the accuracy and repeatability of the load cell to the desired range. Thus the initial post about the quality/ reliability of components. Doesn't make sense to try and force an accuracy out of components that falls out of the specifications of the component.
Some points that I now need to look into is:
Determine if noise is indeed a problem in the system and see if I must implement a low pass filter.
Provide a separate stable 5V supply voltage to the HX711 to not power the HX711 from the arduino itself.
stevemj suggested to use a battery as it has a quiet excitation voltage. I decided to use a 9V battery and only supplied this voltage to the loadcell itself, the HX711 was disconnected at the time. I did this to confirm that the load cell itself was not damaged.
I made a mistake in stating the location of the 100uF capacitor in my previous reply. It is actually placed over the input of the 7.5V supply to the arduino to try and create a more stable supply voltage.
Why the 12volt supply and buck converter.
A 5volt cellphone charger connected to the USB socket is more practical.
HX711 Aref and excitation voltage must be exactly the same, to keep ratiometric behaviour.
Therefore it's not wise to power a load cell externally. It will introduce instability.
Excitation voltage provided by the HX711 is already bypassed with a 10uF cap.
You can of course try if adding a 100uF there improves things.
You can also add one to the 5volt supply, close to the board.
Shielding the module and load cell wiring could also help.
Pin 0,1 of the Nano are already used by the USB<>Serial chip, so don't use them (display).
Use other pins, and SoftwareSerial.
Also make sure the HX711 board you have has a groundplane on the bottom.
There are boards out there where without it, where grounding of the load cell and is missing.
You can see it by holding the board against a strong light.
Leo..
Note that C4 is 0.1uF (100nF) and on the layout it is close to the input, if ceramic these respond to high frequencies. The 100uF you mentioned may be aluminium electrolytic, in which case they are only capacitive a tens of kilohertz. They are great for bulk power supply decoupling and their lossy aspect can help quash ringing on a power supply - I think that is a digretion.
You also mentioned a motor. This is likley to produce more conducted and airborne noise than a noisy thing . Test your bridge and A/D without the motor running.
A series impedance in the supply to the A/D and load cell will also possibly help reduce the power supply switching noise. Something like a low value resistor (10 ohms maybe - I don't know what the current draw is, so check the volt drop) or an inductor, typically called a choke in this application. It depends on what you have handy.
Note also that the data sheet gives single sided layout example, observe that there is 'ground fill' or 'minimum etch', i.e. all the board is at ground referenced to the A/D except the signal traces and some separation of them from the ground. I would assume that if the manufacturer gives this in the data sheet it is a good way to go. If you have bought a module, Iwould expect this is a copy of the data sheet layout; I've some somewhere but can't be looking now.
The sellers missed out, IMO.
Thanks for the heads up. I've some (eBay I guess) so far unused and stored away. When they may not work, I will try to recall your information.
. Test your bridge and A/D without the motor running.
