Load cell analog requirements

[Lazza]

Hi everyone,

Been reading a number of threads regarding load cell amplification, and not yet at a point to ask for specifics just yet. I think I need to ask a broader question first.

If I have a 20mV/V load cell, 3.3v for excitation, and want to use it to replace a pot in a gaming controller (so analog output, not digital) is it feasible to do this with something like an AD623 and some components on, say, a protoboard, or for interference reasons do I really need a dedicated amplifier board?

I don't yet have specifics on the existing pot output range, getting a meter is part of the project

Cheers!

 

[gophert]

A 20mV/V loadcell also has a capacity. 20kg, 60kg or 500kg or almost anything. The 20mV only occurs at the maximum rating. So, when you're putting 3.3v through it, and assuming the capacity is 20kg, and you apply a 20kg load, you'll see a difference of 20mV/V * 3.3V = 66mV.
I day "difference" because the loadcell is usually two voltage dividers (a Wheatstone bridge), and the output is connected at the midpoint of each branch of the bridge.

 

[Lazza]

Yes, understood. That's why I didn't specify the load at this stage, as whether it's 5kg or 500kg makes little difference - if I choose a suitable range I can cover then the issue is somewhat replicating what the pot does now. When I work out how I'm going to do it I'll delve into specifics.

 

[Lazza]

A 20mV/V loadcell also has a capacity. 20kg, 60kg or 500kg or almost anything. The 20mV only occurs at the maximum rating. So, when you're putting 3.3v through it, and assuming the capacity is 20kg, and you apply a 20kg load, you'll see a difference of 20mV/V * 3.3V = 66mV.
I day "difference" because the loadcell is usually two voltage dividers (a Wheatstone bridge), and the output is connected at the midpoint of each branch of the bridge.

I realised I should say that the controller has a resolution of 10 bits for this axis, as that should probably be taken into account. So my simple, unknowledgeable plan is to use the existing 3.3v and earth to connect the load cell and have it output a range somewhere close to what the pot does.

That should be a low point, A, where 0<=A<3.3, and high point B where A<B<=3.3.

Whether that range is 3v or 0.5v I don't yet know.

My first problem, due to lack of knowledge generally and especially regarding load cell amplification, is whether I can get good enough results with an IA IC and some components, or need a ready-made amplifier board. Once i know that I can provide more info, but will also have a direction I can research first so I'm not asking for every single answer.

Hope that explains a little better.

 

[gophert]

To do what a pot does, you'll need a load cell that has an output of 1V/V so, when you excite it with 3.3V, you can get an output of 1V/V * 3.3V = 3.3V range (assuming the range of the pot is 0-3.3V). I also doubt this type of load cell exists unless it has in integrated amplifier (and I don't know if that exist.

If you want to use a load cell with much smaller response per volt, an AI will work but it needs a very good and stable power supply. If, for example, you are running on batteries, you'll need to recalibrate constantly as the battery voltage sags during discharge, or, you'll need some (very good) voltage regulation.

Note, you are trying to use a 20mV output load cell and need (I assume), 3.3V for your signal. That means 3.3V/0.020V = gain of 165. You'll need more gain if the working range is much smaller than max load of the load cell.

Issues of concern,
- noise from RF or other electrical interference associate with cables and board design.
- temperature drift
- voltage regulation
- finding an instrumentati9n amp that has outputs that match your circuit's input. I haven't looked but the old-school IA model numbers do not have rail-to-rail output and need more voltage than 0-3.3v.

Good luck.

 

[Lazza]

gophert I think we're going a little sideways here.

I've found (but don't have yet, obviously) a load cell that should be fine with 3.3V, a rail to rail IA that starts at 3V, gain I can set with a suitable resistor at what I think should be minimum 50 (3.3 / 0.066V) but may need to be higher when I get to working it out. The specific figures aren't my problem (yet), and I know I need an amplifier.

I'm selecting a load cell that I will use the entire range of, and the power supply is provided by the game controller. I know people have made, and sell, kits for this type of thing (same controller, even) that don't use any external power, just the 3.3V the controller provides, so if I can assume from that it should be fairly well regulated. (if you're curious, the load cell in those kits or examples is not the type or range I want to use, so I can't do a straight copy)

My question at this stage is simply: with that load cell, and that voltage, and an IA like the AD623, can I expect to get clean enough results from something like a protoboard setup (with the IA plus various components to provide the gain and some interference and input protection, as detailed in the datasheet but still to be looked into by me) to give 1024 basically clear steps across an output voltage range of something like 1V, or perhaps 3V (or whatever it happens to be, in that ballpark)? It won't be buried in/next to any other equipment (power cables no closer than 20cm or so, computer at least 30cm away).

I can work out the figures with some research and time, I just don't know if amplifying voltages in the range of 0.3mV to 66mV is going to be reliable enough with a bit of a hack job. The IA datasheet talks a bit about RF interference, and ground return impedance and using PCB ground planes to reduce it, but I don't know how much of that is really more related to 'busier' installations and smaller voltage steps than what I'm considering. All the numbers don't tell you what experience probably does.

Things like temperature drift, detailed in the load cell datasheet, I can safely ignore for my application.

 

[Lazza]

Just revisiting to answer my own question in the unlikely event someone googles the same thing I did, and of course in the meantime I think I've basically read the answer in other load cell threads (and I'm sure the same ones I read initially, prior to asking this question... ).

Having got the necessary parts and doing a test run on a breadboard, the load cell is working fine with a rudimentary circuit on the AD623 with stable output. As expected the controller is taking care of the voltage requirements nicely, though I'll have a decoupling capacitor (or two) on the power pin just to make sure. The load cell I got has a 200 lb limit and a (rated) sensitivity of 36mV/V, for info. My calculated gains appear to be about right, with a pot included in the circuit so I can reduce the effective max to around 90 lb as required - from what I've read in the datasheet there shouldn't be any output issues if I max it out with gain...?


As an aside - is it common practice, or indeed recommended, to test all jumper wires for continuity after purchase? I didn't, and spent 2+ hours trying to fix a problem that wasn't there

 

[gophert]

You can always cancel noise from free hanging wires by twisting them together to avoid big open areas (circles or polygons) that can catch radio interference. Also, you can add decoupling capacitors across your dc load cell signal wires (size of caps can be calculated based on how frequently you are taking measurements - the cap will add a lag time between actual (instantaneous) resistance of the loadcell. A few 10s or 100 pF should get rid of RF noise and still allow samples every few milliseconds without error.

As an aside - is it common practice, or indeed recommended, to test all jumper wires for continuity after purchase? I didn't, and spent 2+ hours trying to fix a problem that wasn't there

It is not common practice for me, but I have also suffered a wasted afternoon troubleshooting to find that a purchased jumper wire with micro grabber clips had a bad connection. It might be a good idea to measure them all when you get them but also assume they can be intermittent (position dependent) which can cause even more problems.