Millivolt signal distance?

We have some Emery Winslow hydrostatic load cell systems which measure weights of large blenders and send a millivolt signal (16mv, 6 wires) to a metler toledo indicator that is remotely located.

Of the three scales in my current application, one of them produces a rock solid reading, and the other two show a lot of "bounce" in the signal. They will vary plus or minus 50pounds on a scale range of 0-8000 pounds.

I have not found any specs in the manuals that describe the maximum recommended distance for the millivolt signals.

Do any of you have any rules of thumb or recommendations?

I haven't started tracing conduit yet, but the wiring would have to be at least 150' from the load cell totalizer to the indicator...

I need to verify I have done everything possible to ensure the accuracy of these load cells to prove to mgmt that I can't guarantee the detecion of a 20 pound bag of spice (manually added) using the existing equipment with only software changes (read inexpensive).

TIA
Paul

Just curious, what kind of cable is being used?

I am thinking along the lines of telecommunications. Maybe you need something like a telephone "twisted pair" to null out an induced currents along the way.

Could you setup a meter logging data right at the sensor and see if you see the same variations? Maybe there is some unknown force changing the reading? Like a large ventilation fan above or next to the vat.

Wish I had a good rule-of-thumb, but maybe I can pick one up too!

OkiePC said:

We have some Emery Winslow hydrostatic load cell systems which measure weights of large blenders and send a millivolt signal (16mv, 6 wires) to a metler toledo indicator that is remotely located.

Of the three scales in my current application, one of them produces a rock solid reading, and the other two show a lot of "bounce" in the signal. They will vary plus or minus 50pounds on a scale range of 0-8000 pounds.

I have not found any specs in the manuals that describe the maximum recommended distance for the millivolt signals.

Do any of you have any rules of thumb or recommendations?

I haven't started tracing conduit yet, but the wiring would have to be at least 150' from the load cell totalizer to the indicator...

I need to verify I have done everything possible to ensure the accuracy of these load cells to prove to mgmt that I can't guarantee the detecion of a 20 pound bag of spice (manually added) using the existing equipment with only software changes (read inexpensive). TIA Paul

Click to expand...

Paul
Worked with Mettler local vendor here in Seattle they supplied load cells and readouts similar to your setup. They also did calibration and setup and gave us the purty certification. Generally these were potted units at the cell end so nothing to do there electrically and the lead length was about 25 feet. In essance I did not do setup engineering but did the install and trouble shoot.
We did not cut the leads but coiled excess.

Cells were often located near floor and this was food so washdown was required and you know the rest of the problems.

We had problems with water leaking thru the cable seal at the cell. We finally reduced the washdown problem by sliding a length of shrink tube over the cable and butted it against the cell then injected 3M (?) 6100 RTV to seal the cable end. Then shrank the tube until some RTV oozed out between tube and cell. Cell and first 18" of cable were splinted to a stick to hold stationary and then set aside 48 hours to let RTV setup.

Make sure your cells are properly mounted with no slop. Check each cell with known weight to ensure proper response. If you are going to check voltage output (weight signal) on cells you will need meter to get into uV range.

Make sure your cells will give you the precision (resolution?) you need. When ours worked well we could easily get half pound or so accuracy on 2 to 3000 pounds in a mixer.

The only other thing I can suggest is to make dang double sure the mixer is competely free floating ie no pipes conduit hanging it up. Cell leads were ran thru non metallic liquatite to protect from abuse this did not seem to bind the machine -- was flexible enough.

I thought our Mettler support guys very knowledgable honest would repair when they could instead of trying to sell more. Assuming your guys are as good I would recommend relying on them - yes it is gonna cost a few bucks but well worth the cost.

Let me cogitate further on this and see if I get any brilliant inspirations.

Dan Bentler

What I found

Here is what I found. It doesn't tell you maxium lenght but it does tell you how much your signal will degrade. With 8000 lb load cells seeing 20 lbs in the best conditions is not going to be easy. With an error factor of 1% on 8000 lbs that is 80 lbs. Factor in vibration and things can get real ugly.

A solution I have used in the past when distance and temp has been an issue is to install a junction box (also known as a summing box) as close to the load cells as possible. Then trim the signal at this point. This will allow for some compensation of signal degradation.

Also check your sheild ground. That is the most common thing I find. Someone either forgot to hook it up or grounded both ends. Anyway hope the info helps

Cable Length Effect
For high accuracy force measurement the effects of the cable on the measurement must be considered. For constant voltage excitation there are two effects of significance. These are:
1. An effect on the sensitivity due to voltage drops over the cable length.
2. An effect on the thermal span characteristics of the load cell due to the change of cable resistance with temperature.
CABLE LENGTH EFFECTS
If the load cell is sold with a cable of any length, the sensitivity is determined with the installed cable in calibration and this is not a problem. For load cells with connectors, or if the customer adds cable himself, there will be a loss of sensitivity of approximately 0.37% per 10 feet of 28 gage cable and .09% per 10 feet of 22 gage cable. This error can be eliminated if a six wire cable is run to the end of the load cell cable or connector and used in conjunction with an indicator that has sense lead capability.
TEMPERATURE EFFECTS
Since cable resistance is a function of temperature, the cable response to temperature change affects the thermal span characteristics of the load cell/cable system. For 6-wire systems this effect is eliminated. For 4-wire cables the effect is compensated for in the standard cable lengths offered with the load cells if the load cell and cable are at the same temperature at the same time. For non-standard cable lengths, there will be an effect on thermal span performance. The effect of adding 10 feet of 28 gage cable is to cause a decrease in sensitivity with temperature equal to 0.0008%/degF (an amount equal to the standard Interface specification). For an added 10 feet of 22 gage cable the effect is to decrease sensitivity by .0002%/degF (one-fourth Interface spec). In many cases a customer can tolerate the degraded performance since our standard specification is extremely tight. However, for long cable runs or high accuracy applications, this can be a significant factor. In such cases, the best approach to the problem is to run six wires to the end of the standard cable length and sense the excitation voltage at that point. This eliminates the problem.

Thanks for the replies, guys. I think I have them convinced that we need an alternate method of spice weighing/batching. They thought it would be easy to just use the existing load cells for automatically "proving" spice addition, but once I explained all the pitfalls and the existing innaccuracies, they accepted that answer.

So, I am not going to worry about the length of our cable runs for now, and will concentrate of figuring out how to squeeze one more scale/hopper onto an already congested platform.

Thanks for your replies.
Paul

Paul

One thing I forgot to mention that you already did. We did NOT try to measure items that were in the oz to lb range. These were weighed on a small scale.


Dan Bentler

Paul,

The systems I have seen like your's that worked well used a "pre-weigh" area for small amounts. They didn't weigh them at the mixer. They had them made up ahead of time and packaged them with labels that had all the data on them like: Actual weight, lot number, batch number, etc. When it came time for addition, they barcode the bag and the mixer (to verify addition) and dumped.

It souns to me as though
A/ you have a loose connection
B/ You have some hysteresis at the measuring cell
Try reading +mV to -mV at the cell then lift up and push down on the vessel to see if it always comes back to the starting point.
Regards
Roy

There is not a loose connection, more likely electrical noise on the cable from the load cell totalizer to the indicator. There is vibration on the equipment when the mixing starts, but even then, the hydraulic load cells don't alter the strain gauge reading very much.

This is not the typical electronic load cell system. It is four hydraulic load cells with a hydraulic totalizer summing box which has a strain gauge buried within it. It is very well insulated mechanically from the machine. This is the technology used to weigh rail cars and tractor trailers.

(Manual.pdf)

On the final blender, where the signal is solid, I might see a five to ten pound variability (0.1% of scale) when the mixer is running, and only 1 or two pounds with it stopped.

On the initial blenders, the signal looks noisy on the display and in software all the time, and varies up to 100lbs. (1% of scale)

I think the indicator is doing a very good job of filitering the noise that is probably being picked from one of our dozens of VFD conduits. It is not a problem for the intended purpose of weighing 6000 pound batches of pork, so my motivation to go chasing this goose has vanished.

I can't access the electrical connections during production, so next opportunity during a down day (we have several in August), I will see if I can put a scope on the mV signal going into the indicator.

That tip for checking hysteresis is a good one I hadn't thought of. Thanks, Roy.

If I find out anything interesting, I will post back.

Paul

Ferrite filter

You might try putting a ferrite core filter on the cable.

In the past EW has oversized the transducer in the bottom of the totalizer stack. This reduces mv output at any given point inthe range.

If you are using 14.8sq inch load cells (136-10?) on 4-legs you have about 60 sq. in. of acting area, you should be able to use a 250-lb transducer in the totalizer (there is probably a 500-lb'er in their now) based upon your mv output signal of 16mv.

If it is a blender application, there may have been fear of start up torque sending a spike through the system, but if the scales have been running OK from a reliability stand point you could probably go to a 250-lb load cell with no problems and double your mv signal strength at any given range.

NOTE: only use the EW load cells, as they have require less deflection then many off the shelf load cells.

BR

Can you use a signal conditioner? Mv to 4-20 and if your instrument on the other end needs Mv then 4-20 back to MV a signal conditioner on each end. Kinda janky but it does work.

bilros,
Is this a shameless plug? This thread is a couple years old. Good info though.

I would think you would see a loss from each conversion that would add up to an unacceptable variation.

But...since this thread was started almost two years ago...I am guessing OkiePC's problem is no longer an issue.

Doh! Didn't even look at the date. Got a routing off a google search on Emery Winslow and thought I could help.

Thanks for the heads up.
BR