20140720 Information Sharing: - What can be Inside a Program Part II
July 20th 2014 Sunday Taipei
Dear Fidelity Measurement Friend,
Information Sharing: - What can be Inside a Program Part II
This is the part II of our information sharing on July 213h 2014. If for any reasons you missed that e-mail, please check https://www.fi-measurement.com/news/hnMUeFIVqeLxvbBWOoN219DnkdT5Ghbl4zg9Mltj
…… After completing the above tests, and no matter what the results will be, it is still too early to call it a day and go home. It is because 2 important tests are still missing.
Yes, they are: - creeping and stabilization.
- Creeping happens to all kinds of strain gauge load cell weighing scales, heavy or light capacity, low or high resolution.
- Load cell manufacturers usually apply creep compensation components inside load cell unless someone wants to save some money and work hours.
- Creeping is mostly positive. Negative creep is usually because of excessive compensation is applied or effect brought and contributed by the raw material.
Stabilization (here we only talk about non-high speed dynamic weighing equipment)
- Stabilization is mostly related to usage environment (let’s ignore effects caused by a weak structure here). For example, some desktop scales are used next / close to conveyor belts and moving machineries… make it worse, on a not so stable table / not strong enough movable cart.
- All scales have stabilization programming steps inside to give as stable results as possible. These steps are common referred as filtering. If without filtering, it will be extremely difficult (if not impossible) to read the true and accurate weighing result.
- Filtering can be by any averaging method of the last xx (xx = times of AD readings) values detected. A big xx will give more stable and accurate results, but this slow down the response / stable speed; a small xx does weighing faster but more likely to be less stable. While some programs use more complicated methods / formulas for their filtering steps.
- Balancing speed and stability can be a nightmare for programmers especially when the speed of MCU is slow and when the memory (to hold temporary calculated values) is limited.
One common understanding about creeping and stabilization is: they are 2 separated performances. But it seems that more and more scale program uses one simple method to solve these 2 issues at the same time. Let’s call this V tracking (V stands for variation).
V tracking function behaves similar to auto zero tracking. But instead of functioning only when a scale is unloaded, V tracking comes into effective once a stable weight has been detected and shown on display. Let’s use 2 simple examples below to as illustration,
Case 1: for a weighing /retail scale which wants to give an impression to customers that it is real fast and stable result
- Scale capacity x division = 6kg x 2g
- AD conversion speed = 15 times per second
- Average = 2 times, then reference is formed and result is displayed
- V tracking range = = +,- 4d
After a load is put on scale, then the vision effect will be very like the below
- Weight result will come out about 0.15~0.2 second unless the reference differs more than 8g than the first displayed value.
- Scale will maintain the same displayed value unless the weight change will be more than plus or minus 8g.
- There should not be any weight even you shake the scale horizontally.
Case 2: for a balance which wants to give an impression to customers that it is stable and no creep
- Scale capacity x division = 30000g x 1g
- AD conversion speed = 15 times per second
- Average = 12 times, then reference is formed and result is displayed
- V tracking range = = +,- 2d
After a load is put on balance, then the vision effect will be very like the below
- Weight result will come out about 1 second unless the reference differs more than 2g than the first displayed value. But I would say this first result is very likely to be the accurate result by nowadays technology and it is just a 30000d balance.
- Balance will maintain the same displayed value unless the weight change will be more than plus or minus 2g.
- The same displayed weight probably will remain not affected by load cell creep unless it uses a bad load cell.
It seems that by V tracking, both creeping and stabilization issues are solved at the same time. But there are risks, we do not know until customers will claim.
Risks for scale in case 1 above
- If it is used to weight light products like pepper / flower / mushroom and only small amount (say 5g) of needs to be added or removed from scale?
- several results are accumulated to give a total weight? Can we trust the totalized result?
Risks for balance in case 2 above
- What will happen if the balance will be used for recipe, or
- What will happen if the balance will be used for ink / paint mixing applications?
If you do not want V tracking inside your scales / balance, then start checking now by adding 1.4d to each testing step. May be you will agree that 1.4 can be your lucky number.
We are not going to comment V tracking is good or bad or else. FM just stays away from this kind of V tracking function. It is because there are still other better methods in solving both the creeping and stabilization problem separately and at the same time.
Again, we appreciate your participation in our information sharing process and sharing with us your valued experience. If you find anything mentioned above needs to be corrected or you want to add your comments, please e-mail us.
Next time, I would like to share with you something about metal. Since next 2 weeks will be quite bad for me because of heavy workloads and trips. If I cannot make it on July 27th, then I will do this on August 2nd.
Have a nice day!