Current Sensor With Vision 700

[Chris_1984]

Hi All,

I have a V700 with snap in V200-18-E3/4XB.

I'm looking to use an analog current/hull sesnsor to measure the DC current draw in several branch circuits in the panel (and feed it to analog inputs on the V700).  I know there are several devices available that specifically spit out 4-20ma or 0-10v but they are quite expensive and management is pushing to lower cost.  I came across the attached Honeywell sensors (planned on using CSLA2DG) and was wondering if, with the right scaling, they would work with this V700 controller.

Thanks in Advance!

Chris

honeywell-sensing-current-csla-series-catalog-pages(1).pdf

[Joe Tauser]

I'm guessing your baseline is DC transducers in the $150 each range.

Those are raw sensors and you're going to have to build a circuit board for them.  And come up with a dedicated 12 VDC power supply.  And test each one's output with a known meter at several points  to calibrate them and set up your program.

But yes, they should work.  Give yourself a couple of days to do it.

Joe T.

 

 

[Chris_1984]

Thanks for the feedback Joe!

The best I was able to find price wise were from ATO (probably Chinese) for about $80 per, but we needed something like 100 count of them.

We initially thought that we struck gold and ordered some Dwyer CCT50-200 ( $30/) only to realize that  we didn't read the cut sheet well enough to see that they're AC only 🤦‍♂️.

Would you recommend just using any the "Raw Value" Analog Input Type in the HW Config?

 

Chris

[Chris_1984]

BTW, I would like to say that I'm new to the Unitronics Platform (Mostly done Siemens/AB last 15 years) and these last 3 months I have used these forums for a lot of self teach.  With that being said, you stand out in these forums as a guy that is extremely knowledgeable and very patient/helpful with everyone.  Just want to say thanks a million from the green guys like me!

 

Chris

 

[kratmel]

Something like this with Modbus

https://www.aliexpress.com/item/32977623982.html?spm=a2g0o.productlist.0.0.29fd4687btW6qC&algo_pvid=40fea82f-c7a6-4431-b668-0b2e8cd0c535&algo_expid=40fea82f-c7a6-4431-b668-0b2e8cd0c535-33&btsid=60263821-ef36-474f-97f1-b2fdc5c9830c&ws_ab_test=searchweb0_0,searchweb201602_8,searchweb201603_52

 

 

 

[Joe Tauser]

I love browsing aliexpress.  So many bargains!  Delivery time?  I forgot to look at that!  

I do think that adding Modbus to this mix would overly complicate the code.  But it is an option.

@cleman84 - I'm a Dwyer dealer and I use and recommend those CCTs all the time.  When the first came out I put them on my old-tyme Fluke 760A meter calibrator in AC current source mode and found them to be amazingly accurate for the price.  But as you found out - AC only.

As far as making your Honeywell sensors work, I see the CSLA2DG 150A sensor you've chosen has a 16.2 mV / amp output assuming one pass of the wire through the core.  Cranking the math gives you a 2.43V change for 150A 

It also has an offset of Vcc/2, so when you apply 12 VDC power it will have an idle output of 6 VDC. 

There's a reason the DC transducers cost a bit more - you really need to amplify the output signal and remove the offset to come up with a standard process value.  Or not.

I've never tried the Raw Value configuration and the E3XB datasheet says something about frequency when you select this.

I'd use the 0-10V range in 14 bit mode.  This gives you 1 count per 1.63 mV, so you'll get about 10 counts per amp.  Also take into consideration that the A/D will "chitter" a couple of counts all the time.  Plus +/- 0.4% full scale error, which works out to an uncertainty of 66 counts, or 0.6 amps on a 150A full scale device.

Just make sure your eyes are open before you jump into this project.

Joe T.

[kratmel]

16 hours ago, cleman84 said:

we needed something like 100 count of them.

Please explain is it 100 sensor need to connect to one panel or 25 panel with 4 sensor on V200-18-E3/4XB?

Maybe modbus version of current sensor is cheapest variant. Even if you use something like that

http://www.loreme.fr/fichtech/HCM d33_eng.pdf

Analog version need 100 analog input... and minimum 25pcs  V200-18-E3/4XB.
Modbus has more than 100 address and do not need V200-18-E3/4XB.
But as Joe explain, this task is not simple  in both - analog or modbus version.

 

 

[Ausman]

10 hours ago, Joe Tauser said:

I love browsing aliexpress. 

This is very off topic, but worth a mention.

I too like browsing there, and source stuff on and off, delivery to Aus is sometimes only a few days.  If you choose carefully the quality is often terrific.  But one of my great sources of enjoyment is the review translations, which are fantastic entertainment.

This one, whilst looking for a new set of wire terminal crimpers, was the best so far!

"Perfect, I will ask for another for grimparme the testicles".    😬     I still haven't figured out how "uninsulated terminals" translates into nuts.

cheers, Aus. 

[Chris_1984]

5 hours ago, kratmel said:

Please explain is it 100 sensor need to connect to one panel or 25 panel with 4 sensor on V200-18-E3/4XB?

Maybe modbus version of current sensor is cheapest variant. Even if you use something like that

http://www.loreme.fr/fichtech/HCM d33_eng.pdf

Analog version need 100 analog input... and minimum 25pcs  V200-18-E3/4XB.
Modbus has more than 100 address and do not need V200-18-E3/4XB.
But as Joe explain, this task is not simple  in both - analog or modbus version.

 

 

Not all the current monitors will be tied into the snap in IO, I will have several EX-RIO with Analog Cards.  I was just using the snap-in IO in my question to simply.

I'm creating a fully controlled Power Distribution architecture for several labs

           - Each Lab will have a V700 in a Main PDP (Power Distribution Panel)

           - The Main PDP will have 4 Branch Circuits to Remote Power Panels.  Current Monitors For Branches tied into 4AI on Snap-In Module)

           - Each Remote Panel will have EX-RIO with an 8AI Module to monitor 5 Circuits

           - 20 Sensors For Lab (5 labs planned as of now) = 100 pieces

I'm already going to be using UniCAN, so adding the Modbus adds another layer of communication/complexity which will outweigh the cost savings of using such hardware.

[Chris_1984]

13 hours ago, Joe Tauser said:

I love browsing aliexpress.  So many bargains!  Delivery time?  I forgot to look at that!  

I do think that adding Modbus to this mix would overly complicate the code.  But it is an option.

@cleman84 - I'm a Dwyer dealer and I use and recommend those CCTs all the time.  When the first came out I put them on my old-tyme Fluke 760A meter calibrator in AC current source mode and found them to be amazingly accurate for the price.  But as you found out - AC only.

As far as making your Honeywell sensors work, I see the CSLA2DG 150A sensor you've chosen has a 16.2 mV / amp output assuming one pass of the wire through the core.  Cranking the math gives you a 2.43V change for 150A 

It also has an offset of Vcc/2, so when you apply 12 VDC power it will have an idle output of 6 VDC. 

There's a reason the DC transducers cost a bit more - you really need to amplify the output signal and remove the offset to come up with a standard process value.  Or not.

I've never tried the Raw Value configuration and the E3XB datasheet says something about frequency when you select this.

I'd use the 0-10V range in 14 bit mode.  This gives you 1 count per 1.63 mV, so you'll get about 10 counts per amp.  Also take into consideration that the A/D will "chitter" a couple of counts all the time.  Plus +/- 0.4% full scale error, which works out to an uncertainty of 66 counts, or 0.6 amps on a 150A full scale device.

Just make sure your eyes are open before you jump into this project.

Joe T.

Thanks for the math on this.  I don't think I would've been able to get to that 10 counts / amp.  I will run with that to start and then just use some trial and error with it.

I'll report back when I finally get it all implemented.

 

 

[Joe Tauser]

On 9/6/2019 at 8:55 AM, cleman84 said:

Thanks for the math on this

As a final comment, what I put out is just that - math.  Your final results may vary.  Don't order 100 pieces - start with something like 5.

On 9/4/2019 at 1:33 PM, cleman84 said:

they are quite expensive and management is pushing to lower cost.

<rant>

Define "quite expensive".  Is this data kind of important or very important?  I think management is freaking out over the overall cost of the project and struggling to get the cost down and pointing to a large dollar line item without quite understanding what's involved to roll your own sensor.  When I was a young man I tried rolling my own 4-wire ohmmeter that was an integral part of a wire plating process.  The short version is we tried to make my circuit work for a month in production only to determine that what I had come up with was not accurate enough nor could overcome the inherent noise in the process.  We had to buy a $5K meter that worked in the process right out of the box.

So often I see a company hire a young engineer and put him in charge of something that is really out of his league.  "Oh, he's and engineer!  He can figure it out!"  But what's missing is the crusty old engineer to give him guidance.  So he comes up with a process controller / monitor that is based on an Arduino or a Raspberry Pi.  And he thinks he's a genius, just ask him.  And then he goes and gets another job and nobody has a clue how it works.    

Consider the value of your time and how much damage you will cause if this doesn't work or is flakey.  Also consider who gets to maintain this thing.  If you built it, you are married to it.  If you do go forward with this be sure to document the shit out of it, including ordering spare parts and writing a calibration procedure.  That's what professionals do.

</rant>

Joe T.  

[Chris_1984]

Totally agree Joe.  Being in automotive industry this long, I've just become numb to the bean counters and their unwavering attention to the bottom line.  The way I see it is these people don't have the technical background to appreciate a KISS method and the costs ensued with future downtime.  It's almost like they have the thought that if/when something does happen, it will be someone else's problem.