# How to improve HX711 to feed two or more load cells in parallel?

HX711 datasheet

I have typical load cells and want to connect them in parallel directly to the HX711 "A" section of the amplifier.

The load cells have an input resistance of 400 Ω and an output resistance of 350 Ω each.

I have a few questions to solve:

1. What exactly is the analog supply regulator in HX711, and how does it work?
2. Are it the halved resistors in the Wheatstone bridge that create that problem? If so, according to Ohms law, the current flowing through the resistors should increase about two times, right? Is that why parallel loadcells don't work?
3. If my problem connects to the analog supply regulator, how do I manage it? By replacing the PNP transistor (s8550 appeared in the second pic) used in HX711 or by enlarging the current out of the analog supply regulator by the number of load cells linked in parallel times?

• electronics.stackexchange.com/questions/111887/…
– Drew
Apr 21, 2022 at 8:58
• What are the values of R1 and R2, which determine the $V_{AVDD}$ value? Have you measured $V_{AVDD}$ with one and two load cells connected? What are the voltage values? Apr 21, 2022 at 10:49
• @devnull I have not measured it yet, I will tell you when I measure it. is it a voltage divider or a current divider? Apr 21, 2022 at 11:27
• R1 and R2 form a voltage divider which determines the feedback for the "analog supply regulator". While measuring, please also chek $V_{SUP}$. It is also possible that the main power supply it is not capable of providing enough current for the load cells. Apr 21, 2022 at 11:33
• @devnull, OK - part of the load cell assembly.
– Nedd
Apr 21, 2022 at 16:09

After reading your updates with the measurement information, and without knowing exactly which board is in use, I wonder if it is not one of these, sold with a PCB error:

Images source

If this is your board, there is no connection between the AGND pin and the GND connection to the micro-controller. This workaround will bring the board closer to this one, sold by sparkfun:

These resistor values (20k and 8k2) will provide 4.2 V ~ 4.3 V at the load cells.

Finally, regarding you original concern of connecting 4 sensors in parallel, the current requirement would be $$\\approx 42 mA \$$ and the power dissipated by the BJT (according to the above voltage) would be $$\\approx 34 mW \$$. Both, around 1/10 of the device limit, according to this datasheet.

• what should I do now? I am a little confusing Apr 22, 2022 at 13:52
• Is this the board you are using? Have you checked if E- is connected to AGND but not to GND from the microcontroller? Apr 22, 2022 at 13:53
• I did not check yet, I left it at another place and will let you know when I get there Apr 22, 2022 at 14:06
• What if it is a board I am using, then do I have to make the connections above? Apr 22, 2022 at 14:06
• AGND - GND GND - ARDUINO MEGA are connected It worked when I did not use the board regulator and connected directly to the Arduino mega 5V, GND. What I am wondering is, what are the consequences of connecting in that way? Apr 23, 2022 at 11:52

The detailed operation of the analog supply regulator is not fully explained in the data sheet, the equation for determining R1 and R2 is, (see section: Power Supply Options). The regulator would seem to be a simple adjustable (using R1 & R2) low drop out type with an external transistor for increased current. But as with most users you likely have the module that already has the resistors mounted. You should not need to change the resistor values when using two load cells unless there is a specific voltage requirement for them.

Connecting the load cells in parallel should work. The only limit should be the max current of the S8550 PNP transistor, (typical Ic spec is about 700ma, however the spec sheet doesn't claim what the actual maximum transistor drive is).

Your diagram does not show your connections to the load cell outputs, be sure you are connecting the loads cell outputs correctly in parallel, (not series), the two +Outputs go together then the two -Outputs go together.

Obviously in your arrangement (2 parallel load cells) the total current required will be 2 times that of a single load cell. Also, your available full scale measurement range will become 2x the capacity of one load cell. For best accuracy you should try to load both cells equally.

Here is an older StackExchange string with a similar multiple load cell question: Load cells in series

.

• Should I replace the S8550 or not? Can we calculate the transistor value to handle that issue(maximum current limitation) and pick one from the market? Apr 22, 2022 at 5:10
• @Tumenbayar... From what I see it is likely not required to swap out the S8550 when using your 2 load cells. Even if you had used 4 load cells in parallel (giving 100 ohms to be driven), the total load cell current would only be 50 ma, (assuming a 5v drive voltage). This would still be a fraction of the max Ic current. But again as I said above the max transistor drive is not given in the spec sheet (Base pin). Assuming everything else is correctly wired you could assure yourself of the issue by measuring the Vavdd voltage then compare it to the spec sheet's Vavdd calculation.
– Nedd
Apr 22, 2022 at 7:30
• Good info from devnull. If the load cell ground was not correctly routed on the module then this could give some strange results, especially if troubleshooting using the E- pin as a ground reference. From what I have now read some low rate vendors likely copied the printed schematic and overlooked signal names that indicate connected nodes. If that wasn't all, seems the original Vavdd equation in the HX711 datasheet is incorrect, (even some older Sparkfun drawings had it incorrect). The newer Sparkfun schematic posted by devnull has the corrected form, AVDD=VBG(R1+R2)/R2 .
– Nedd
Apr 23, 2022 at 17:24
• Of course there is also the possibility that on the questionable module the analog supply transistor, the associated resistors, or other components are open, shorted, or just defective.
– Nedd
Apr 23, 2022 at 17:24