Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
Zeeshan Ashraf
11/01/11 02:20
Read: 637 times

#184499 - RE
Responding to: Kai Klaas's previous message
Kai Klaas said:
As Stefan and I pointed out earlier a standard load cell isn't offering a higher precision than about 0.05% to 0.1%, which corresponds to 10...11bit precision. So, no more than 13...14bit precision of front end electronics is needed. There's no need to use an ADC providing 24bit resolution and 18bit precision.

That is the reason why most PLCs with generic (though industrial grade) 16-bit ADCs can connect to a standard 0-10V (or 4-20mA) output of a Summing amp (Passive, semi-Active OR Active) and yet give quite a good result. I have repaired quite a few Siemens S7-300 & -200 series CPUs and Analog modules, not to mention Allen Bradely, and i have seen quite dumb parts that one would not expect in such widely used systems.

My point on all this was that if a PLC can use generic ADCs and an analog front-end, primarily designed with the highest concern on input protection against electrical shorts, over-voltages and such, Then why cant an embedded system be built in the same fashion, though, not SO GENERIC. ?

I know that it is dumb to HAVE the opportunity of developing something from scratch and yet not doing the best possible way as indicated by all (one compact unit with tight integration and possibly no IAs in-between the Loadcell and ADC to reduce errors of those devices added on top of the load-cell internal error and line losses etc.). However, in a less than perfect world, we sometimes need flexibility as a trade-off.

Right now, what i am considering is to develop a Batching control system that is not involved or concerned with the load-cell correction and calculations etc.. Rather, a generic system which should expect a real-world quantity to be scaled to a 0-5V or 4-20mA (low-impedance) signal by a separate device (Summing amp) and just measure it precisely with a good quality ADC and scale the digital reading back to the real-world units for measurement and reporting to the master computer. WHy you may ask ?

1) Well, even though the present application is for weighing a batch of granular material and delivering that to production machine, i do not intend to let my efforts go to waste in the sense maybe a next project would involve batching of liquids where volumetric quantities are more desirable. If i develop a completely integrated control system with AFE designed specially for weighing with load-cells, i will have to do considerable changes in the analog design, AND get the PCBs fabbed again for the other applications. All this translates into re-work and a lot of time.

2) Moreover, not every batching & Weighing system needs such high resolutions as this one. There are other factors that count more in some systems. if i develop a high precision tightly integrated system, it might be too costly and overly sized for simpler applications. For example, A plant producing bricks with each batch of around 2000 KGs per 2 hours, can easily tollerate a precision of 11 to 12 bits. However, a Pharmaceutical plant producing quality dry-syrups or suspensions of mixed powders with batch sizes of 100 KGs would need atleast 16 bits or higher for controlled dosage of final product to a consumer.

So i am of the view to tackle problems piece-wise and generally break a whole system into small re-useable blocks, just like you develop smaller reuseable pieces of code for a larger application.

Now with all the valuable suggestions and feedbacks, there can be a few possibilities that i am open to and considering.

A) I go with the 2-part system, develop the control system that is part-1, and use a good quality Summing amp OR Indicator (as a signal conditioner) from the markets.

B) Again 2-part system, but i design a summing amp with an on-board uC and AD7730 that connects directly to the load-cells. The ADC can compensate for a lot of errors due to very low self-noise and AC excitation that too ratiometric. For transferring the conditioned data to the Control system, i either use a DAC for 0-5V output OR a serial (RS-485) link.

C) A single-unit system, but that will be tooo specific for this particular application. However, i am considering this as well though as a last resort.

Kai Klaas said:
In fact, in industrial weighing scale applications other issues are way more important: Is the weighing scale really empty after clearing? Are the load cells working properly or did they become stuck due to dust and dirt? And and and...

Kai Klaas

For the question on whether the scale (Hopper) is really empty after discharging or not, well there are 2 main types - 1) which actually keep checking the weight going down in the hopper and keep the doors open till then. 2) most Mass or Bulk material Batching systems are based on time. They keep the hopper door open for a (programmable) time and then close it. Once the doors are closed, a new (stable) reading is taken and considered as new empty weight and tared to give ZERO. process resumes from there.

This last option, based on time, is especially used in places where the material(s) are sticky or of high-viscosity in nature, for example Cement / concrete weighing Batching system for brick making plant. If it is purely based on hopper discharging completely, then due to sticky or viscous material, the hopper may never close automatically and process will be put in a waiting state, unless someone manually scrapes off the material from the hopper for every batch load.

Some new systems use both weight and time. Once the weights have gone down to a certain tollerance, they start a timer, the expiry of which closes the doors and re-starts the cycle.

Load cells deteriorate over tie, specially in open / outdoor environments. carbon-steel ones become rusty and hence develop a lot of fatigue and shear. Such applications require Stainless steel cells, particularly of a SS-308 or SS-316 material.

Even slight changes in the components make a world of difference. I have seen changing simple 4-wire load-cells to 6-wire types of relatively the same precision grade giving quite better performance, specially in open atmosphere or environments. The effect of heat from Sun, RH, etc. play a very critical role. Wires elongate / expand under high temperatures and contract in lower temperatures... The elongation adds to the Effective-Series-Resistance, while contraction lowers it. the extraction and contraction also forms cracks and hence ghost resistance in the wires.

So thee are a lot of factors. Not all can be catered for. There is no such thing as an IDEAL system ! having said that, it would be foolish to ignore the ones we CAN correct or compensate for.


List of 58 messages in thread
Load-cell conditioning / summing      Zeeshan Ashraf      10/29/11 10:29      
   what is that?      Erik Malund      10/29/11 15:43      
      RE: #184430 - what is that?      Zeeshan Ashraf      10/29/11 15:45      
         then say what you mean      Erik Malund      10/29/11 15:53      
            your description leaves something unclear      Erik Malund      10/29/11 16:02      
               RE: #184435 - your description leaves something unclear      Zeeshan Ashraf      10/29/11 16:27      
                  where are the loadcells???      Erik Malund      10/30/11 06:08      
                     RE:      Zeeshan Ashraf      10/30/11 09:01      
      RE: #184430 - what is that?      Zeeshan Ashraf      10/29/11 15:56      
         some Q/A      Stefan KAnev      10/30/11 14:14      
            RE      Zeeshan Ashraf      10/30/11 15:27      
               Q/A 2      Stefan KAnev      10/31/11 03:00      
                  RE      Zeeshan Ashraf      10/31/11 04:18      
                     re re      Stefan KAnev      10/31/11 05:00      
                     get rid of that summing amp      Erik Malund      10/31/11 07:09      
               you will be lucky to get 16 bits      Erik Malund      10/31/11 07:04      
                  Precision again...      Kai Klaas      10/31/11 16:58      
                     Comparison      Per Westermark      11/01/11 01:12      
                     RE      Zeeshan Ashraf      11/01/11 02:20      
                        why do you keep babbling about a summing amp      Erik Malund      11/01/11 06:41      
                        Priorities...        Kai Klaas      11/01/11 07:07      
                           RE      Zeeshan Ashraf      11/01/11 07:45      
                              cirrus, ti      Stefan KAnev      11/01/11 08:42      
                                 maxim, linear tech, Analog Devices      Erik Malund      11/01/11 10:09      
                                 RE      Zeeshan Ashraf      11/01/11 17:36      
                              it takes the same time ...      Erik Malund      11/01/11 10:14      
               500 running foot of cable      Stefan KAnev      11/01/11 22:25      
                  RE      Zeeshan Ashraf      11/01/11 22:57      
                     Go for ADC with multiplexed inputs instead of multiple ADC        Per Westermark      11/02/11 07:14      
                        Good Advice      Michael Karas      11/02/11 08:00      
                           It depends...      Kai Klaas      11/02/11 08:48      
                              This is a MODULAR concept !      Zeeshan Ashraf      11/02/11 19:57      
                                 it depends      Stefan KAnev      11/03/11 00:50      
                                 don't be ridiculous      Erik Malund      11/03/11 07:17      
                                    Consider other parameters too.      Zeeshan Ashraf      11/03/11 08:16      
                                       calibration      Stefan KAnev      11/03/11 08:46      
                                       if you do not buy bootleg parts ...      Erik Malund      11/03/11 08:50      
                                          Voltage reference      Per Westermark      11/03/11 09:08      
            Accuracy...      Kai Klaas      10/30/11 16:40      
               RE      Zeeshan Ashraf      10/30/11 16:55      
                  Keep the signal paths as simple as possible.      Michael Karas      10/30/11 20:34      
                  sun umbrella      Stefan KAnev      10/31/11 03:10      
   Separate Gain Paths - Not The Best      Michael Karas      10/29/11 23:33      
      RE: #184439 - Separate Gain Paths - Not The Best      Zeeshan Ashraf      10/30/11 00:11      
         Error Budget Calculations?      Michael Karas      10/30/11 09:50      
            Hysteresis      Kai Klaas      10/30/11 10:14      
         This should alarm you!      Kai Klaas      10/30/11 09:58      
            RE      Zeeshan Ashraf      10/30/11 10:14      
               No contradiction...      Kai Klaas      10/30/11 10:22      
                  RE      Zeeshan Ashraf      10/30/11 10:40      
                     Digital Summing        Michael Karas      10/30/11 12:55      
               with DC exitacion ...      Erik Malund      10/31/11 07:20      
            welcome on board      Erik Malund      10/31/11 07:14      
      RE:      Zeeshan Ashraf      10/30/11 00:16      
   Look at ADuC845      Oliver Sedlacek      10/31/11 01:44      
      ADuC845      Russell Bull      11/16/11 20:54      
         Noise V good      Oliver Sedlacek      11/17/11 04:39      
            Ok, then      Erik Malund      11/17/11 06:26      

Back to Subject List