Hi all,

I'm new to AAC, but I just couldn't find an answer to my question elsewhere (or at least one I was confident with).

Short version: I'm looking into load cells with a capability to measure 15000 steps and no less than 100 grams capacity (0.1 grain resolution on a 1500 grain capacity scale), but I'm not finding any load cells with that high of a resolution. How do commercial scales do this? Is it software compensation?

Long version: I'm looking into building my own home powder measuring scale in conjunction with another guy that has some more circuits knowledge, and we're having trouble determining what kind of load cell we should be using. I want the scale to have a max capacity of 1500 grains (1 gram = 15.4324 grains), with a resolution of 0.1 grains. This seems to be the industry standard for commercial powder scales for what I'm looking to do.

I've found several 0.3 Kg load cells with a 0.02%F.S. resolution (5000 steps), but this seems to be the highest I can find. If my understanding is correct, that means these load cells are capable of measuring in .06 gram intervals (about 1 grain). I also took apart a commercial scale, and it seems to be using a very similar 0.3 Kg single-point load cell.

Do I really need to find a load cell with a 0.002%F.S. resolution? Or is there a way to compensate during the A/D conversion to allow my scale to effectively measure the 15000 steps between 0 and 1500 grains?

Any input or knowledge would be appreciated.

Thanks!

 

You are going to find the transducers with full scale loads of numbers like 100 grams or here is an old popular example:



This scale is an early 90s vintage RCBS actually made by PACT. The strain Gauge used is a BCL A Series and this particular strain gauge is a BCL-300 GN meaning it is a 300 Gram (4629.71 Grain) load cell. Here is the data sheet for one such strain gauge. Powder charges are weighed in grains so we figure 1.0 Gram = 15.4324 Grains.

Looking at the linked data sheet:
Rated capacity 2.942 N {300 gf} ~ 44.13 N {4.5 kgf}
Rated output 1.6 mV/V±0.2 mV/V
Excitation, recommended 10 V or less
Excitation, maximum 15 V

What that means is that at full scale (300 Grams) the output will be about 1.6 mV per volt of Excitation. With a typical 10.0 Volt Excitation the output becomes 1.6 * 10 = 16 mV. This is where we need an IA (Instrumentation Amplifier) to amplify that small 16 mV (0.016 Volt) signal making something usable.

Here is an Amazon example of the same type used in the RCBS scale image.
Electronic Balance Weighing Load Cell Sensor 0-300g

Here is an example of a nice 100 gram unit:
Electronic Balance Four-wire Connecting Weighing Load Cell Sensor 100g

Using the 100 gram unit as an example, the units I had were an output of 2 mV/V so at 10 Volt excitation full scale output was 20 mV. Make a X250 amplifier and we get 5 volts so now 0 to 100 grams = 0 to 5.0 volts. Now comes the steps. This voltage is passed along to an AD (Analog to Digital) converter. For example a 10 bit AD will take 0 to 100 grams ( 0 to 1543.24 grains) will give 2^10 = 1024 quantification levels so that won't work. A 12 bit AD will give 2^12 = 4096 quantification levels'

The best way to understand the concept of resolution is by comparison with a yardstick. Divide a 1 meter yardstick into millimeters. What is the resolution?

The smallest “tick” on the yardstick is the resolution. Yes, you might be able to interpolate between these, but in the absence of this sophisticated guessing process the resolution is 1 part out of 1000.

The resolution of a n-bit analog-to-digital Converter (ADC) is a function of how many parts the maximum signal can be divided into. The formula to calculate resolution is 2^n. For example, a 12 bit ADC has a resolution of 2^12 = 4,096. Therefore, our best resolution is 1 part out of 4,096, or 0.0244% of the full scale.

That gets you about .376 grain resolution. You are looking at a 16 bit AD which gets pricey. The transition from grams to grains is done in the software. Don't confuse or mix resolution with accuracy or uncertainty.

<EDIT> I forgot to add that if you buy from Amazon or similar these things are off the boat from China and while they do work and generally come close to specifications the Data Sheets prove to be amusing written in what we refer to as Chinglish. You can always get a high end precision unit which will cost far more than most reloading scales. </EDIT>

Happy Loading & Shooting

Ron

 

Thinking about this a little further I'll share a little more of that RCBS Powder Scale made by PACT. Something interesting I noticed was in the following picture:



They come off the transducer (upper left) and amplify the signal using a few op amps. That done rather than feed an AD converter take note of the AD 652 which is actually a V/F converter. They convert the voltage which is proportional to the weight into a frequency and the display is actually a complete modular frequency counter. Likely because it was simpler and much less expensive, especially early 90s when the scale was made.

On another note to better define Resolution and Accuracy take a look at the following temperature indicator.


If we define resolution as the ability to read an instrument or of the instrument to be read the placed displayed to the right of the decimal are resolution. However, let's think about something. Let's say my range is 0 to 100 Degrees F. and my accuracy is +/- 1.0% of Full Scale. We now get 0.01 * 100 = 1.0 so we have +/- 1.0 Degree F. Really doesn't matter how far out we display digits does it? In the example 76 Degrees F. is all the resolution we need. A reading of 76 Degrees could actually be 76 +/= 1.0 or between 75 and 77 degrees F. Lacking the accuracy to support it the resolution means very little.

Generally as you are aware in most hand loading powder weight charges we seem to like +/- 0.1 grain for resolution as well as uncertainty so we can say the accuracy is +/- 0.1 grain without a reference to full scale value. A good powder throw will maintain that charge tolerance and a good scale will get us +/- 0.1 grain Accuracy as well as resolution. The problem is getting a transducer with the uncertainty (accuracy) combined with a good ADC (Analog Digital Converter) which affords enough quantification levels for good resolution and also affords good uncertainty.

The rest is accomplished in the software. There are also methods such as over sampling to get around the bit count limitations but you really do not want to do that.

Ron

 

Thanks so much! Exactly the info I was looking for and aligns with what I was thinking myself. I had read some information about classification of verified intervals in load cells, and thought that might make a resolution that small kinda worthless (even if the industry standard). For example, there's info that a C6 load cell (higher end) can only verify 6000 steps between zero load and max load, so I thought that it might be pointless to use a 16-bit AD if the load cell was really only good for 6000 steps.

I guess my next step is finding some 16-bit AD converters.

Thanks!

 

More than welcome. If you need help just return with questions. The links to transducers should get you started.

Ron

 

I guess my next step is finding some 16-bit AD converters.

I've been successfully using the AD7680 chip for the last few years, and it has proven itself reliable and easy to use. Its only drawback is that it's SMT, and has to be soldered to the PCB using low-temp soldering paste.

 

This is where we need an IA (Instrumentation Amplifier) to amplify that small 16 mV (0.016 Volt) signal making something usable.

Or you could just run the load cell mVs into a delta-sigma with a built in PGA and save yourself a boat-load of trouble.

 

Or you could just run the load cell mVs into a delta-sigma with a built in PGA and save yourself a boat-load of trouble.

That is another option. When I was toying with automated powder measures I tried several approaches. The gun forums have some really slick ideas guys have come up with and incorporated. While I have a number of good scales I started using a RCBS 1500 ChargeMaste Powder Scale and Dispenser Combo which I have come to really like. Program it and sit back. As the unit trickles close to the programmed charge it slows down dispensing powder. On average if when throwing powder charges (for most rifle) if we can hold +/- 0.1 grain we are good. While some look for better or tighter I have never seen much difference. Powder volume metered density can vary lot to lot as can even how well the primers ignite the charges.

Ron

 

While I have a number of good scales I started using a RCBS 1500 ChargeMaste Powder Scale and Dispenser Combo which I have come to really like. Program it and sit back.

That's the same model I had. At first I was encountering huge weighing errors with it, which turned out to be the forces induced by electrostatic charges on the plastic case and especially on the transparent draft cover over the weighing platform. I gave the unit a good healthy spritz of anti-static spray all over, and that solved the problem.

 

That's the same model I had. At first I was encountering huge weighing errors with it, which turned out to be the forces induced by electrostatic charges on the plastic case and especially on the transparent draft cover over the weighing platform. I gave the unit a good healthy spritz of anti-static spray all over, and that solved the problem.

Overall I have been pleased with mine. It was actually a gift at Christmas a few years ago. There are some tweaks and peaks out for the firmware but I have not screwed with mine. Primarily loading .308 Win and .223 Rem and it does fine. Plus, something about looking a gift horse in the mouth?

Thanks
Ron