Load Cell Wiring Length Question

[Freixas]

I would like to weigh our pet parakeets. They are rather skittish, so the idea I came up with was to hack a digital scale so as to put the controls and display some distance from the scale. Then I would add a perch to the scale. When one of our birds lands on the scale, I could turn on the scale, wait for it to fly off, and read the weight (displayed as negative number since most scales tare when they are turned on).

I know only about enough electronics to be dangerous. The scale I picked (like most) uses a load cell. I managed to extend the wiring from the load cell to the controls and displays by about 8 feet. I assume the circuit board does the signal amplification and digitization required to make the load cell useful.

The problem is that the display reading is always 0. It's possible I zapped a component, but there aren't that many (there's just one IC chip) and if things were zapped, I would expect the display to malfunction.

One thought I had was that the added resistance from 8 feet of wire affects the signals from the load cell, which I understand to be fairly weak. Ideally, one would amplify the signal before sending it down an 8 foot wire, but the amplification is probably done on the circuit board that drives the display, so that option isn't possible with this scale.

Before I try anything else, I'd like to hear whether the wiring length is the likely source of the problem from someone who actually understands electronics and load cells. Other ideas about things to check are welcome. Since my soldering skills are questionable, I used a continuity checker to make sure the connections from the load cell (red, black, white, and green wires) connect with the circuit board, and that no wires short with any others, which is about the extent of my troubleshooting.

As I noted, the display works. I can turn on the scale, and I can change the mode to grams, kilograms, ounces, or pounds. Whatever I pick, the reading is 0 and remains 0 no matter how I push on the end of the load cell.

One other change I made was to go from 2 AAA batteries to 2 AA batteries.

 

[jrmichler]

We need more information.

What is the full scale load rating of the scale?
Approximately how heavy is a parakeet?
The number and colors of the wires indicate that it is a strain gauge load cell. A strain gauge load consist of four strain gauges connected as shown below:

The input voltage is typically 3 to 5 volts, but can be 1 to 10 volts or so. Normally, all four strain gauges (2 or 3 could be resistors) are all the same resistance. Typical resistances are 120, 350, or 1000 ohms. If the load cell is disconnected from the signal conditioner/amplifier, you can measure the resistance between any pair of wires. For 120 ohm gauges, that resistance will be either 120 or 90 ohms, depending on which pair you are measuring.

If the load cell is properly connected, you can measure the $V_{in}$ and the $V_{out}$ with any DMM. The $V_{out}$ will normally be zero to a few millivolts, and will change slightly if you push on the load cell. Disconnect the wires to measure the resistances, measure the voltages with all wires connected.

Lead resistance is typically less than one ohm, and should be roughly equal in all four leads. That being the case, an ohm or so of lead resistance will not significantly affect the operation of this type of load cell. All four leads should be in one cable, or at least twisted together, to reduce temperature effects.

 

[berkeman]

I would like to weigh our pet parakeets.

How heavy is their cage compared to their weight?

 

[Freixas]

How heavy is their cage compared to their weight?

Very. The cage by itself overloads the kitchen scales we have. The scales that can handle the weight of the cage don't have enough precision.

There are some postal scales with extensions for the controls/display. These have the same problems--they are designed for big packages, so they handle more weight but with less precision. Postal scales for smaller packages don't have the extensions.

If you're considering alternate solutions, I also thought of a kitchen scale with bluetooth. Generally, though, I think you have to manually turn on these scales; you can't turn them on with bluetooth. A lot of scales (all of them) also have a time out, even when connected to an external power source. If these bluetooth scales do, too, then they won't work. A bluetooth solution would require either a remote on button or no timeout.

There may be other solutions that I haven't thought of. One that I have thought of is an Arduino-based scale that I could design however I wanted. But that's a lot of work.

 

[Freixas]

If the load cell is properly connected, you can measure the and the with any DMM. The will normally be zero to a few millivolts, and will change slightly if you push on the load cell. Disconnect the wires to measure the resistances, measure the voltages with all wires connected.

Thanks. I think the red/black wire voltage was around 1.5 volts (not millivolts), half of the total voltage available from the two AA batteries. I tried other color combinations, looking to see if any voltages changed as I pressed on the load cell. I never got anything that reacted in any consistent way.

I can try this again more methodically. I have no idea which wires are Vin and Vout, though. Red and black would have been my guess, but the readings there aren't in millivolts and they don't change when I press.

 

[BillTre]

What is the birds' behavior like at night?
Could they be fetched out of their cage and weighed then?
I am guessing the anesthesia of birds is kind of tricky.

Or have a vet do it for you. They will have their methods.

Or bundle the bird up in a sock or something and weigh it and then the sock separately.

 

[berkeman]

Maybe consider buying a small cage that is light that would make it easier to weigh them.

 

[Freixas]

What is the birds' behavior like at night?

They sleep at night.

Could they be fetched out of their cage and weighed then?

Not really a good idea.

I am guessing the anesthesia of birds is kind of tricky.

Correct.

Or have a vet do it for you. They will have their methods.

Yes, our vet weighs the birds when they come in for a physical once a year. I was thinking of monitoring their weight once a week.

Or bundle the bird up in a sock or something and weigh it and then the sock separately.

While anything is possible, this would be rather traumatic, especially on any sort of regular schedule. The birds would soon learn to fly away at the sight of a sock.

Maybe consider buying a small cage that is light that would make it easier to weigh them.

We already own all sorts of cages. If we put a small cage nearby, they are unlikely to be motivated to go into it much. Waiting for them to do so might take a long time (hours). The birds are semi-wild; we can't catch them if they are outside their cages and are trying to avoid catching them when they are in their cage. Catching a bird that doesn't want to be caught is something we know how to do, but always has a risk on injuring the bird, so we do it only when absolutely necessary.

The goal is to weight the birds without creating any trauma. They like to perch. Putting up a nice perch that sits on a scale would work for them. But if we start walking up to the perch to push an ON button or to read a display, they are going to fly away.

The ideal system might be an Arduino scale (with a perch) that is externally powered and with a continuous reading that appears on a large wall display. Even better is if it permanently tares the weight of the perch. But that's a lot of work.

Right now, my question centers on ways of getting my hacked scale running.

 

[BillTre]

Yes, our vet weighs the birds when they come in for a physical once a year.

Do you know how they do it?You might want to try:

Training your bird(s). This is how they get animals to do things in labs. Probably takes a while. On the other hand, animals can be trained to do a lot of things for food. I have trained a fish to jump out of the water and take food from my hand. A friend trained some lab fish to make visual decisions under controlled situations.

Some cameras have remove button pushers for cameras. Or you could make a mechanical one, like a reverse bicycle break cable, or a hydraulic pushing thing.

The only birds I have really had experience with are parrots or duck, geese and chickens. All pretty robust.
Delicate birds are collected in the wild in mist nets (very light but entangling). Doesn't hurt the birds, but would probably traumatize them.

Training them to perch on a scale of some kind, maybe in a continuous read mode might work.

 

[DaveE]

Have you considered training them to sit on the perch, for a reward, while you look at the display. Parakeets shouldn't be difficult to train. There are undoubtedly many other benefits from teaching them to come to your finger and transfer to a perch. Not everything needs to be an engineering problem.

I went to YouTube to look for bird training videos, but, as an experienced dog trainer, I was really disappointed in the number of crappy trainers posting stuff; kind of embarrassing IMO, a real Dunning-Kruger gallery. It's just basic animal behavior after all. So be careful, maybe look for chicken training videos, they are often used to teach training skills in general.

Anyway, I thought Elle in this video was really good, she gets it. Look at her YouTube page for other training examples.

 

[BillTre]

Wow, @DaveE, that was a good video!

 

[Tom.G]

Hmm... sounds like time for some serious troubleshooting.

Please start by telling us the resistance between the various wires and their colors. (color coding varies by manufacturer) I would expect the resistance RATIO of different wire combinations to be 4/3, as @jrmichler pointed out:

Typical resistances are 120, 350, or 1000 ohms. If the load cell is disconnected from the signal conditioner/amplifier, you can measure the resistance between any pair of wires. For 120 ohm gauges, that resistance will be either 120 or 90 ohms, depending on which pair you are measuring.

The simplest way to check for damage or a mis-wire is to eliminate the 8 foot extension cord. Re-connect the load cell directly to the circuit board (you did document the original connections, didn't you?) and see if the scale still works.

If no success there, either the re-connection is wrong (four wires have 24 different ways to connect, 6 of them may work) or something is damaged.

If it works you might want to identify which leads are the supply and which are the signal output. It can make later troubleshooting easier if you know what the correct voltages are.

​

Measure from each lead to the other(s).​

The combination with the highest reading are the supply voltage.​

The other lead(s) is/are the output.​

​

Most often, the voltage from an output lead to a supply lead is half of the supply voltage; AND the voltage between the output leads is zero or perhaps a few millivolts.

Now, re-connect that 8 foot extension cord one wire at a time, testing to see if it still works after each extension wire is connected.

It Works: You are Home Free!

It DOESN'T work:

Extension is bad​

Solder joint bad​

Load Cell or circuit board damaged from all that soldering (easy to do! )​

​

Cheers,
Tom

 

[Freixas]

Do you know how they do it?

Yes. We have caught and weighed one of our birds many times. They don't like being caught and there's a potential that they will get hurt.

Training them to perch on a scale of some kind, maybe in a continuous read mode might work.

Yes, it would work. It doesn't really require much "training". They like perches, although they get a little nervous with anything new.

 

[Freixas]

Please start by telling us the resistance between the various wires and their colors. (color coding varies by manufacturer) I would expect the resistance RATIO of different wire combinations to be 4/3, as @jrmichler pointed out:

Ok, this may take a while. As I understand it, I have to disconnect the wires from the signal conditioner/amplifier, which I think means that I have to unsolder the connections and then re-solder them when I want to measure the voltage. This is a pain.

The simplest way to check for damage or a mis-wire is to eliminate the 8 foot extension cord. Re-connect the load cell directly to the circuit board (you did document the original connections, didn't you?) and see if the scale still works.

Reconnecting is also a pain. There is no problem with documenting the original connections. The load cell has fine wires that were connected to the circuit board. All I did was cut them and then spliced 8 feet of wiring in between the cut ends. The splicing was done by soldering them to some small chunks of breadboard. To undo do this would take a while. The wires from the load cell are so fine that I don't even have a wire stripper that works.

I'd have to cut them from the breadboards, strip the ends very carefully, and re-solder them using a new chunk of breadboard. If the scale worked, then I could conclude the extra length of wiring was the problem. If it didn't work, then I hosed something. Neither of the two possible results would lead to my goal.

If no success there, either the re-connection is wrong (four wires have 24 different ways to connect, 6 of them may work) or something is damaged.

The four wires are colored. Connecting them properly is not a problem.

Most often, the voltage from an output lead to a supply lead is half of the supply voltage; AND the voltage between the output leads is zero or perhaps a few millivolts.

It DOESN'T work:
Extension is bad
Solder joint bad
Load Cell or circuit board damaged from all that soldering (easy to do! )

To verify the solder joints, I tested the continuity from the source load cell leads to the circuit board. I think we can eliminate "solder joint is bad".

I never soldered near the load cell, just the wires, so it's unlikely any soldering damaged the load cell. It might have been damaged by something else I did.

Same at the other end.

I think @jrmichler had the best suggestion. With power enabled and everything connected check the voltage and see if I can view a few millivolts change when pressing on the load cell. I think I've tried this, but it sounds like I should first identify Vout and check to see if it changes when pressing on the load cell.

By the way, the four wires lead to four circuit board connectors labeled Gnd, E+, S+, and one unlabeled. Which one is likely to be Vout?

Of course, if I see a voltage change, that means the circuit board is damaged. If I don't see a voltage change, then the load cell is damaged. (@jrmichler didn't think the 8-feet of wiring would be a problem). So again, it looks like no matter what I find, I'm done, so I could just not bother.

It seems there is no path that leads to a working system with the system I have. The question is whether this approach would work with another scale. If @jrmichler is wrong, and the 8-foot extra length is a problem, then there is probably little point in hacking another scale. Removing the 8-feet of added wiring would at least answer that question.

 

[Freixas]

OK, I captured some voltages. I captured them at the source (output of load cell) and at the circuit board.

The first number is the source voltage, the second is the circuit board voltage.

B (Gnd) - W: 0 mV, 0 mV
B (Gnd) - R (E+): 3.447 V, 3.445 V
B (Gnd) - G (S+): unreliable, 1.493 V
W - R (E+): 3.437 V, 3.441 V
W - G (S+): 1.487 V, 1.485 V
R (E+) - G (S+): 3.44 V, 3.439V

Black to Green yielded varying values, so I'll need to work on that solder connection.

The only reliable mV reading was Black to White. This reading did not vary when I pushed on the load cell.

The display shows numbers (always 0 for the weight), but sometimes "Err", so that also points at a flaky connection. I'll look into this before giving up.

 

[jrmichler]

Based on your last two posts, this is what I think you have:

When the excitation voltage is 3.4 volts, there should be 1.7 volts from either W or G to B, and 1.7 volts from either W or G to R. There should be 0 or a few millivolts between W and G, and that voltage should change slightly when you press on the load cell.

Keep in mind that a typical load cell will tolerate a 50% overload with no or minimal damage, while a significant overload will normally destroy the load cell.

 

[Freixas]

Based on your last two posts, this is what I think you have:
View attachment 321758
When the excitation voltage is 3.4 volts, there should be 1.7 volts from either W or G to B, and 1.7 volts from either W or G to R. There should be 0 or a few millivolts between W and G, and that voltage should change slightly when you press on the load cell.

Keep in mind that a typical load cell will tolerate a 50% overload with no or minimal damage, while a significant overload will normally destroy the load cell.

Thanks! After I found the screwy Gnd to S+ connection (black to green), I re-soldered it and it fixed it! Then I screwed the scale's case back together and it broke. So I bit the bullet and I redid all the soldering using a fresh bit of breadboard and being a lot more careful about the soldering. That fixed it much more reliably. I tested it with some pennies (after 1983, they are 2.5 g), and the scale was still accurate +/- 1g.

I'm finishing the remote case right now. It was helpful to know that the 8 feet of extra wiring wasn't likely the problem. That led me to keep looking at the wiring. The continuity checks misled me; the voltage checks, done methodically, were more useful.

Thanks for your help!

 

[.Scott]

I would use a calibrated cantilever.
1) Install a cantilever able to bend as a small branch of a tree would.
2) Hang a perch from the end of the cantilever.
3) Calibrate: Hang several weights from the perch, one at a time, and mark the amount of bend you get from each one.
4) When the bird visits the perch, read the amount of bend.

 

[Freixas]

I would use a calibrated cantilever.
1) Install a cantilever able to bend as a small branch of a tree would.
2) Hang a perch from the end of the cantilever.
3) Calibrate: Hang several weights from the perch, one at a time, and mark the amount of bend you get from each one.
4) When the bird visits the perch, read the amount of bend.

A clever idea.

There are, in fact, scales one can buy pre-made. They are a little hard to read from a distance, though. If I made one myself I might be able to create a readout that is easy to read from a distance.

A parakeet weighs around 32-40 grams. To make a cantilever that can measure this weight precisely and is viewable from a distance would require something that is probably going to bend a lot. I'm picturing an indicator that shows the reading against some fixed scale. As the bird lands, the perch they land on sinks and the indicator sweeps a large arc. The net result is that the bird is likely to say "What the ---!" and avoid this perch in the future.

It's also possible they would decide this was their favorite toy. You never know, but I would vote for the former reaction with my birds. :-)

 

[.Scott]

A parakeet weighs around 32-40 grams. To make a cantilever that can measure this weight precisely and is viewable from a distance would require something that is probably going to bend a lot. I'm picturing an indicator that shows the reading against some fixed scale. As the bird lands, the perch they land on sinks and the indicator sweeps a large arc. The net result is that the bird is likely to say "What the ---!" and avoid this perch in the future.

It's also possible they would decide this was their favorite toy. You never know, but I would vote for the former reaction with my birds. :-)

I have never had a bird. But I understand that parakeets like mirrors. So if you hung a mirror in front of the perch, that should do it.

In the wild, birds seem to expect small branches to bend and sway - and they adapt.

You can set the cantilever up so that the perch hangs 12cm from the cantilever attachment point and the scale is 36cm away. So the pointer will move 3 times the distance that the perch sinks.

You can also have the pointer "cage" below 25grams. So the swing will only occur between 25 and 40.

 

[sophiecentaur]

Or bundle the bird up in a sock or something and weigh it and then the sock separately

The bird needs to be immobile so something like a sock is essential. It's what the professionals do with 'active' birds of all sizes. And your birds will forgive your brutal treatment.