Load Cell Accuracy | Capacity >200Kg | 10g or Better Accuracy

[Imperitor]

Hi Everyone,

I've been trying to find a load cell for a report and I need one with a maximum capacity of >200Kg and and accuracy of 10 grams or better (used in compression). I've taken a look on the web and the highest accuracy I can find is about 0.03% of the maximum capacity; which with a maximum capacity of 200Kg would be 60 grams.

This is for a report so I don't think my professor would send me on a wild goose chase to find something that didn't exist. Can someone who knows anything about load cells help me?

 

[FredGarvin]

.02% is the best you'll ever see. Here is one that will accomplish that for your load range:

http://content.honeywell.com/sensing/sensotec/pdf_catalog06/IC48.pdf

 

[davidtphillip]

My company - Zwick (http://www.zwick.com/en/products/xforce-load-cells.html?tx_z7treedependingdownloads_pi1[file]=80) can supply load cells accurate down to 0.001%. PLease contact us if you would like more information.

 

[Mech_Engineer]

My company - Zwick (http://www.zwick.com/en/products/xforce-load-cells.html?tx_z7treedependingdownloads_pi1[file]=80) can supply load cells accurate down to 0.001%. PLease contact us if you would like more information.

The spec sheet you gave a link to quotes the load cell accuracy to be:

The accuracy of the Xforce HP and Xforce K load cells is typically 0.001% of the load cell capacity or 0.25% of the measured value from 0.4% of the capacity onwards (whichever is greater).

It's a little hard to decipher, but it sounds to me like the load cell is really only accurate to 0.001% for the first 0.4% of it's capacity, and after that it is .25% accurate (I hope you know what load to expect pretty well). The Honeywell load cell on the other hand is 0.02% accurate across its full scale.

 

[FredGarvin]

Normally I wouldn't say anything, but since this was a spam hit and run, I'll throw this in:

Not only is the statement that M.E. pointed out ambiguous, it is the only sentence in that data sheet that comes close to supporting the accuracy claim. It doesn't go over well when it begins with "our products are typically..." No where in your specs do you state anything better than the .2% full scale accuracy or guarantee the stated accuracy, nor do you state that to get the accuracy you claim, that you need a special calibration.

 

[davidtphillip]

MEch Engineer

The 0.25% relates to the measured value. I understand from your information that the Honeywell value is related to full scale. Therefore please consider the following

We are saying that at 0.4% of the cell capacity (0.4% of 200kg = 0.8kg) we can measure this value to an accuracy of +-0.25% (0.25% of 0.8kg = +-0.002kg).

This represents 0.001% of the call capacity as stated in a previous post.

Also if I understand it correctly the honeywell data sheet you refer to specifies calibration from 20% of the load cell capacity up to maximum, and the accuracy is stated as % of load cell capacity. I think that it is important to consider both accuracy and its associated range.

I hope this helps the explanation - but am happy to involve our technical people for more clarification.

 

[davidtphillip]

FredGarvin,

Our load cells are used on materials testing machines, and the calibration procedure is governed by the ISO 7500-1 Standard. The highest class in this Standard is Class 0.5 which means that the cell must be accuracte to +-0.5% of reading. The standard also specifies a lower reading, which as the honeywell data states, is 20% of full scale. In our example this is 0.1% of load cell capacity. All testing machine manufacturers have to achieve this accuracy to sell a class 0.5 machine.

Many of our customers need a wider range - for example for the measurement of Young' Modulus, and so we developed these load cells with a much lower measurement range.

I hope this explains the situation a little more clearly. Sorry if our document is not 100% clear. With your feedback we can improve it.

 

[dr dodge]

many times in the measurement industry, specs can be very confusing. (we techs here, laughingly call this spec-manship). Many people don't understand that % accuracies, can vary drasticly depending upon if its % of reading, %FS, the "below X" limitations, long and short term drift, uncertainty in calibration, and 4:1 ratios misconceptions. if you don't get "apples to apples" in the units you plan to use, and compare those with the tolerence needed, thruout the whole range, the chance of buying the wrong product increases pretty quickly.

I work our tech support call center and it makes me feel real bad for the guy who just dropped $$$$$$ only to find out someone spec'd something wrong, and all I can tell the guy is sorry, buy this instead. (many times the individual that made the wrong decision, shortens his career, too)

dr

 

[davidtphillip]

FeedGarvin,
Why do you refer to my post as spam hit and run?
What do you mean with your reference to special calibration?

 

[Imperitor]

Thanks for the replies everyone,

We are saying that at 0.4% of the cell capacity (0.4% of 200kg = 0.8kg) we can measure this value to an accuracy of +-0.25% (0.25% of 0.8kg = +-0.002kg).

This represents 0.001% of the call capacity as stated in a previous post.

So as I understand it this accuracy is relative to the load applied? So with a 16kg load the accuracy would be 0.25% of 16kg = +-0.04kg or 0.02% of the full scale (200kg) load. Therefore with anything weighing more than 16kg this load cell is less accurate then the Honeywell load cell? Please correct me if I am wrong.

 

[FredGarvin]

FeedGarvin,
Why do you refer to my post as spam hit and run?
What do you mean with your reference to special calibration?

Because you had a zero post count and then you show up selling something. That is the definition of spam IMO. I am surprised that you stuck around as long as you have. Most others would make the single post and then never come back except to spam some more.

In response to your second question, your data sheet says absolutely nothing about reaching the accuracies you are stating that you reach. If you are capable, why not make a point of saying such? The quote I referenced before is the only place you say anything close to it and it is extremely open ended. It appears that one would need a special calibration to achieve that accuracy since such accuracy is not explicitly mentioned.

 

[davidtphillip]

Fred Garvin,

1) First I never showed up trying to sell something.

Your first post ".02% is the best you'll ever see. Here is one that will accomplish that for your load range" is more sales orientated than mine. This is actually what prompted me to reply.

2) My second and third post already answered with your second paragraph.

 

[davidtphillip]

So as I understand it this accuracy is relative to the load applied? So with a 16kg load the accuracy would be 0.25% of 16kg = +-0.04kg or 0.02% of the full scale (200kg) load. Therefore with anything weighing more than 16kg this load cell is less accurate then the Honeywell load cell? Please correct me if I am wrong.

So as I understand it this accuracy is relative to the load applied? So with a 16kg load the accuracy would be 0.25% of 16kg = +-0.04kg or 0.02% of the full scale (200kg) load. Therefore with anything weighing more than 16kg this load cell is less accurate then the Honeywell load cell? Please correct me if I am wrong.

You make a good point - but let me try to explain. If you want to use a load cell to measure a load then you need to know the accuracy it can read that particular load.

If we look at the example from Fred Garvin you can see that the the load accuracy was stated as 0.02% of load cell capacity.

Lets suppose that we are talking about your required 200kg load cell, and let's make a measurement of load at 40kg (20% of the load cell capacity). The stated accuracy at this point is 0.02% of 200kg (40g). So your actual measurement has an accuracy of 40kg +-40g = 0.1%

Now let's compare the data at 0.4% of capacity (as in my example which you quoted):

For Fred Garvins proposed load cell : 0.4% of 200kg is 800g. The accuracy remains the same at 0.02% of load cell capacity (40g). So the load reading is only accurate to 5%.

Now you see the difference between the two load cells.

So to answer your specific point: You should obtain from load cell manufacturers the accuracy at the lowest point which you want to measure. Above that point the accuracy (related to the measured load) will be better. In my example the accuray at full scale (200kg) of our cell would still be 0.001% of full scale (=2g)

I hope this explains the point. Please come back if not.

 

[dr dodge]

ok, I can not stand quiet.
The specs listed on the x force cells is exactly what I was talking about. Accuracy is stated in the paper for a grade 1 as 1%. Repeatability is 1%, Reversability (hysterisis??) 1.5%, Zero error o.1% and Resolution 0.5%. if we use simple sum total error is 4.1%. that's is 4.1% down to 0.04% of total cell capacity. The Honeywell cell states in note 1 that their total accuracy stated is 0.02-0.05 depending upon load cell. Now, granted you could use RSS to get the accuracy of the xforce below 4.1, but it would never get better than 1%. The x force has a resolution of 1/2 its stated accuracy, which is like saying 2 counts. Honeywell infinite Neither seem to list any kind of sensitivity, unless this is included in the resolution. No long tern drift, so the accuracy is correct with a fresh cal out of the box, but what about 1 year later when it goes for re-cal

spec-manship makes the marketers look good and makes techs and engineers pull their hair out (note the ratio of bald tech/eng to marketers/sales force...lol)
((yes I am bald))

dr

 

[davidtphillip]

Dr Dodge:

Thanks for the input - any chance to improve our documentation is good.

First the easy point:
Resolution : As you must know as a technical person, this is not dependent on the load cell - but on the processing system (conditioner and display). As I have said many times our loadcells are normally sold as part of a Testing Machine and we show data for the complete system including processing electronics. The "infinite" Honeywell data by itself is totally meaningless because it does not include the equipment necessary to display the values from the load cell. So you cannot compare these two points.

Second: Our load cells have to meet the ISO 7500 standard which specifies the maximum limits you mentioned (1% repeatability etc.) We have to do that in order to comply with the standards in our industry. However, in my examples above I am merely saying that we can do much better, and I have tried to illustrate this with examples.

The reason I joined this forum was because I did not agree with the first post from Fred Garvin who appeared to be trying to sell us a Honeywell product. It was unfortunate that he then accused me of being a spammer!

I would rather stay with the objective arguments and I appreciate your comments as to where we could make our documentation much clearer.

Once again thanks for the comments

 

[dr dodge]

The point I was trying to make was the gross lack of consistency in many specs. I was not trying to endorse either cell or make anyone mad. The resolution is indeed based upon the readout in the case of your cells, being as they are tied to a read out, but then this creates other situations of determining specs. If the same cell is connected to a DVM with infinite digits (I know, no such thing), as opposed to a 2,000 count readout both units could be no more accurate than a single count, so cell 1 would apear more accurate. Honeywells cell is rated vs straight line fit. Many units (not just force) that use a "tied together readout" super characterize the readout to match the non linear properties of the sensing element. This (potentially) makes the long term drift much worse, as readings are tied to the electronics, vs a primary sensing element. It also limits the ability to "plug and play" various readouts and sensing elements without re-cal needed. Neither sheets spec drift, nor the uncertainty of the standard to cal it. In the case of some of our products, the accuracy of available standards, due to state of the art, almost doubles its "possible performace accuracy". When purchasing any product for measurement lots of time should be spent to determine all the facts before spending a single buck as specs that appear poor may just be conservitive, and others founded on no emperical data are hopelessly skewed

(thanks to all for the good discussion, I love metrology)

dr

 

[Imperitor]

Thanks guys, this discussion was very informative. I've got all the info I need now.