Any Hall Effect gurus out there?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Can you help?

I'm looking for a very cost effective method of sensing low current ac and dc in a 12ga copper conductor which cannot be disconnected, tapped into, or looped. This requires the use of some form of split core ferrite/collector, and a very sensitive chip. The current in the cable is constantly moving, ranging from one amp to approx 250 amps. It is water cooled (explaining the small ga.) and the total size of the cable with water jacket is approx 8mm. This cable supplies a welding current to a torch. No metal shielding on the cable.

I'd like to apply a non-contact sensor to this cable to determing when there is current and when there is not. Threshold of approx 2 to 3 amps. Basically, enough to know if an arc is present at the torch or not.

There are many, many forms of HE sensors out there, and many of them bidirectional (to my understanding this indicates that they are capable of sensing ac and dc current). There are as well some magnetoresitive sensors out there which are more sensitive yet, but appear to be less straightforward in ease of use. Important since I'm not an engineer and must rely essentially on what I can glean from application notes accompanying the devices.

Here's one HE device I've identified as possible,
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=620-1213-ND

Here's one MR device I've found as well,
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=480-1997-ND

And finally, a solid core unit that seems to be capable and only 18USD, but alas, not split core accessible,
http://sensing.honeywell.com/index.cfm?ci_id=140301&la_id=1&pr_id=146381

I have eventually found versions of the above, completed for use, and split core. They are extremely expensive. For some reason, perhaps there is only a very small demand in the market, there are no budget variety ac/dc split core sensors on the market. It looks like I could fab one pretty easily with the first two devices I've linked by epoxying them into a gap in a split ferrite ring.

It's difficult for me, with limited electronics language skills to ascertain from the data sheets if these are actually capable of meeting my needs.

Can anyone comment on whether I will be successful at this --given that I am not capable of much in the way of signal conditioning once I've put these two items together?
 
Rogowski is ac only. There are other very simple (relatively) methods of detecting just ac. Flux Gate devices can be whipped up with various bobbins and lacquered wire.

The HE chips like the first one I linked to have all the bells and whistles contained within for ac and dc detection with just a little signal conditioning necessary afterward to take advantage of both sensitivities. Datasheet..
http://www.allegromicro.com/en/Products/Part_Numbers/3245/3245.pdf

I'm just hoping that someone out there has actually fiddled with one of these chips in a similar application. I get, essentially, the method of physical assembly. The chip is cemented into a gap that is ground into the face of one side of a split core, or it is inserted into a gap cut out of a solid ring. Then is it a matter of just an opamp or transistor, and one or two resistors and caps to turn it into an ac/dc current detector.

If anyone knows of a schematic for such a circuit, I'd be ever so grateful.
 
Hi BB
Just some thoughts...
The DC capabilty is the real driver for using a HE sensor. If AC current sensing is the only reqirement then your task becomes much easier. I don't know much about welding but perhaps the DC has some AC component that could be taken advatage of. Maybe a dual current sensor could be devised to take this in consideration. I think if you don't care about absolute current accuracy, might free up the requirement of a HE?
 
There's a good possiblity that the DC welding current will be impure enough to do so. Even the arc itself probably would contribute to that. But with such low amperage, I'd just have to do a lot of experimenting to see about that. I'd rather aim for something that is dedicated to the task. The spec sheets on some of these devices indicate this ability.
 
F. W. Bell and/or Sypriss has sensors that will do the job. There WILL be some signal processing and circuitry to build.

Wherever split cores are involved... price will be an issue. It is simply because of the precise machining that is involved to provide any decent level of interchangability/accuracy.

Fluke makes reasonably priced current clamps. Assuming that you see the 2 Amp reading on a meter (easy)... what would you then DO with that signal?

I ask because your biggest stumbling block here is not the sensor... but rather how you need to interface that sensor to something else...

:)
 
bluebeard said:
Can anyone comment on whether I will be successful at this --given that I am not capable of much in the way of signal conditioning once I've put these two items together?

If you're lucky and the threshold is fairly well defined then you may get away with something as simple as a comparator to drive an LED, but until you get some idea of what kind of output you can generate then there's no way of knowing.

These Allegro components are pretty cheap, and there's quite a lot of conditioning built in. I'd buy one, cobble up some kind of clamp (maybe out of some regular [iron] plumbing fitting or maybe you can cut a ferrite with a diamond disk), power it up in situ and put a scope across it and see if anything is visible in the conditions you are interested in. You might need to change the device orientation to get the best (any) result. Give it a local supply with a 7805 and a couple caps and you won't blow it up.

If the trace is lumpy then you may have to think some more (maybe a unity gain buffer and an RC filter), but if there is a reasonable offset well distinguished from the noise then you're home and dry other than checking that it's stable with temperature...

w
 
flux concentraors may require a little care in material selection - you need hysterisis < 1% to detect 2A followng some time at 250A

a quick look at NVE sensor chips suggests that they might work without flux concentration - or with a low concentration factor making the shape and tolerances easy
 
What is often overlooked or unstated with these sensors is that:

A sensor with a stated range of 1 to 50 Amps... may be perfectly capable of carrying 300 or more amps. The reading will, of course, be erroneous. But, no damage will be sustained. You must check with the manufacturer.

:)
 
Administrator
Joined 2004
Paid Member
Hi bluebeard, Conrad,
I think you are looking for the old HP 428B DC amp meter. The aux output will respond accurately to an AC signal. I think it was limited to 200 Hz or something in that area. I have one I'm going to restore and use.

These are vacuum tube type instruments, but I think it's the principle you are after. The service manual explains that in detail. Basically, it's an AC excited coil (chopper amplifier) with a type of discriminator. You aren't looking for extreme accuracy or repeatability. Therefore this might answer your questions. You can easily scale the range to what you need.

Have a look on the HP web site HP archive site under equipment. My link dumps you there. I see there are two versions for this meter.

Good luck with your project!

-Chris

Edit: I forgot to mention how accurate this meter still is. It surprised me, so much so that I feel it's a keeper. Yep, I intend to rebuild and continue using it.
 
Thanks so much for all of the input.
The HP articles are very interesting reading for some of the physical considerations of this thing. Poobah, that really is an overlooked aspect of these things. The application sheets seem to take for granted this understanding, but it is in no way universal. I had more than a couple of engineers tell me that amperage in excess of the rating would destroy the sensor. Most, however, told me that the device merely 'saturates' harmlessly and since I'm not concerned with accuracy above my nominal threshold, not to worry.

I've got some samples of the Allegro Hall sensor A3245LUA-T due in the mail as well as a few samples of an asahi-kasei sensor
http://www.asahi-kasei.co.jp/ake/en/product/linear-ic/file/eq-711l_e.pdf

I've got enough ferrite rings laying about the shop, but I ought to look around for a well made split core of the proper end profile to fit the sensor best. Absolute machined tolerance is not that important since real accuracy is not essential -- only sensitivity. The asahi distributor even faxed me a little circuit necessary to get me up and running. There's a trim pot identified that will allow me to adjust my operating threshold after I get everything clamped together, and then I can simply replace that pot with the right resistor value.

I'll drop back in here in a week or so when I've had some success (I'm optomistic). I never fails to impress me how deep and wide the range of electronics skills are in diyaudio. Thanks so much.
 
Keep in mind that all other things aside, your reading will be roughly proportional to the width of the gap in the ring. So... while absolute accuracy in machining is not required... a stable mechanical setup is. You will want to keep that gap as stable (and then some) as the measurement you seek. You'll also want that gap as small as possible to get useable flux.
 
Administrator
Joined 2004
Paid Member
Hi bluebeard,
That big hairy fella knows more about this subject than I do - for sure.

What you could do is cut the ring in half, then cut one of the halves again to mount your sensor. Then align it and use epoxy to make a solid half assembly again. After that, it's only a matter of maybe more epoxy to mount and align a hinge and snap assembly for the other side. That should be stable enough to give repeatable measurements. If you use a test loop and pass an amp through it, you can calibrate the unit every time to make sure. If you loop twice, one ampere will read as two amperes.

Finally, your results will be better if you polish the mating ends very smooth and parallel so they come together with minimal gap. Should work fine until it's dropped.

Hi Poobah,
Really good to see you!

-Chris
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.