Danfoss (Siemens) Mag Flow Validator

address@invalid.com says...

Because three points are all that is required to determine the offset,
gain, and linearity of the 4-20 output; there's little need to drive all
the way to 20 mA.  Also, the Verificator may not be capable of
generating the magnetic field equivalent to 20 mA (12 m/s).

--Gene
 

Gene S. Berkowitz wrote:

You can determine the linearity only of that part of the range
encompassed by the three points, and then only approximately. Only a
graph covering the entire range is ironclad.

Jerry
--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

proclaimed to the world:


A three point check is the minimum acceptable calibration check
generally accepted in any of the standards. While what you say is
correct in principle, reality is much different. Taking readings at
20, 50 and 80 percent and extrapolating the two extremes is preferable
to 0, 50 and 100 percent checks. There are several reason for this. We
can go into this later if you like. Some of them come from the legacy
of analog sensors.

To answer the original question, I would guess the validator was
engineered this way because they felt it more important to check the
meter output in the range where the most errors happen. Mag meters
have the most errors in the first 20 percent of it's range. As the
magnetic field increases, the signal to noise ratio increases, so the
upper range of the meter is more reliable and accurate than the lower
areas. Checking the accuracy of the electronics in this range makes
sense but I would still include a point near full scale. There may be
some limiter that I don't know of that made it difficult to simulate
higher flow rates. As I have said before, the question would best be
answered by calling the manufacture and talking to the guy who
designed the validator electronics. It would be nice if BIGEYE did
this and let us know what they said.

In the application here, the only way to check the function and
accuracy of a mag flow meter is to simulate the field conditions of
the coils at some predetermined rate. This only checks the
electronics. The other check that can be done on site is by measuring
the resistance of the coils, both the winding resistance and the
insulation with respect to ground. Some validators do both. Most of
the errors I have measured in mag meters have come from resistance
problems.

In mag flow meters you are always assuming something. If the
conditions in the primary measuring element are within design specs
then a voltage linear to the liquid velocity is produced. Calibration
is a matter of verifying that those conditions exist.

This is in opposition to the premise of calibration traceability in
it's purist form but real world calibrations are often this way,
particularly in flow measurement. Doing a direct comparison of flow is
very rare. Only a few labs exist in the world have the capability to
do direct comparisons. I had a rather detailed investigation into what
could be done to establish a better chain of tracibility for bunch of
18 and 24 in mag meters used to measure waste water flow being pumped
to a central treatment plant. Each meter was used to bill the
different municipalities. We had problems with the accuracy of the
meters, which were 20 years old. We used doppler meters to compare the
measurements, but doppler meter are not as accurate as mag flows, out
of the box. I usually try to have a "standard" that is four times more
accurate. The combination of this comparison and doing extensive mag
meter checks were the best solution we could find until the mag meters
could be replaced. We found that the errors were all caused by
breakdown in the coil insulation at the potted penetrations.

BTW we also attempted to do drop tests, measuring the flow by the
change in level of the wet wells. I did a bunch of measurements of the
concrete tanks and quickly learned that as little one inch change in
the dimensions of the tank would produce errors in my calculation that
exceeded the target accuracy we were trying to prove (0.5%). Still,
using two separate comparisons in combination with using a validator
(simulator) and having all thee measurements agree is a very powerful
pacifier. I would love to hear a better solution to flowmeter
calibration short of removing it and shipping it back to the
manufacture.

Sorry for the ramble.:-) At least I didn't tell any old navy stories
this time. Perhaps there is an answer of sorts in there somewhere.

Paul M wrote:

flow

Paul, You wrote a lot of important information; that's not rambling. A
three-point check is a useful and usually adequate way to verify that a
system that once worked properly still does, but it can provide no new
information outside the checked range. For example, a 4-to-20 ma
transmitter may be incapable of driving more than, say, 15 ma at the
burden it carries, but be quite linear at lower currents than that.

Emphasizing the lower parts of the range when calibrating is often a
good idea. To achieve anything like the relative error of the upper part
of the range, the absolute accuracy must be greater there.

Jerry
--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯