Destroying a perfectly good piece of test equipment by ignoring the ratings

Fluke 80i-110S AC/DC Current Probe Product Review


or how to destroy a Fluke current meter accidentally, but not in any obvious, straightforward manner
The little probe that could

You can reach me at fred.niell@gmail.com



[Home][Research]


Like most professional engineers/experimental physicists/males, I rarely read the directions for any new piece of test equipment that comes into the lab. I essentially read the specs, look at the controls, and if it makes sense, I start using it as I would any other tool. However, often, there is a little detail lurking somewhere - waiting to bite me. One afternoon this past summer, I was working on a 20kW switching supply with a friend who shall remain nameless and the probe stopped working without warning. But, we really like the probe, so we bought another one to replace it. Here's the strange story of what happened.



Probe Specs from fluke.com (copyright Fluke)

''The 80I-110s is compatible with any Fluke ScopeMeter Test Tool, Power Harmonics Analyzer, Oscilloscope, Multimeter, or other voltage measurement device that has the following features: BNC input connectors or if using a meter convert the banana plugs to a BNC connector using the adapter PM9081/001.''

• Current Range: 0.1 to 100A DC or 0.1 to 70A AC
• Basic Accuracy (DC to 1kHz):
100mV/A setting: 50mA to 10A +/- 3% of reading + 50mA
10mV/A setting: 50mA to 40A +/- 4% of reading + 50mA
40A to 80A +/- 12% of reading + 50mA
80 to 100A +/- 15% of reading
• Output Signal: 10A range: 100mV/A | 100A range: 10mV/A
• Bandwidth: 1Hz to 20kHz
• Working Voltage: 600V AC rms
• Maximum Conductor Size: 11.8 mm (.46 inch) diameter
• Safety Rating: IEC 1010 and CSA-C22.2 No. 1010-1 Installation Category II, Working Voltage 600V
Taken from http://us.fluke.com 4/14/08

I am a physicst, but I play a power electronics engineer at work. As such, I have lots of high power DC switching supplies to design, debug, and generally work on. So, I have a constant need for a good wide-bandwidth AC/DC current meter that can clamp on. Of course, for pulse and AC measurements, I always defer any other measurement device to an appropriate Pearson current transformer, but for DC switching supplies, you always have DC components that you can't really capture with a Pearson very well, so an AC/DC clamp probe or a current transducer like a LEM LTS-25 is preferable. The trouble is that there aren't really any clamp probes that have much bandwidth that can handle more than a few amps (Tek probes are great, but just try and look at 100A pulses with them...). The clamp part is pretty important, because the switching supplies I'm talking about are typically powered off the 480V 3 phase lines, and you really don't want to have to shut down the system just to place a current measurement. So having the flexibility of clamping on a nice, insulated probe is nice. So - when I saw the Fluke 80i-110s, I figured that all my dreams were answered. And for the most part, they were.


Click for larger



Green is the current in a boost supply, stepping up into continuous mode at 55A peak, 33kHz. Looks great - perfect display of the current in a boost supply.

Blue is the current again in a boost supply, stepping up into continuous mode around 30A, 33kHz again. Notice how nicely the current ramp is shown by the probe. No hint of rounding of the triangle or anything - more or less perfect representation of the current.

Here things are getting a little more hairy. Green is the Fluke, Blue is a pearson (note the sagging and centering around zero due to DC rejection), and the Yellow is a LEM LTS-25. The supply is running at 42kHz, but only 25A pulses. Still looks OK.

We really wanted to get a handle on what the real current looked like, so here is a comparison of all 3 current transducers. Magenta is the collector voltage in a boost mode power supply - textbook. The Yellow is the pearson - note that they are co-incident in time. The Green is the Fluke, and the Blue is the LEM. Note that the Fluke shows about a 4 microsecond delay - just as advertized. The LEM shows a longer delay, but better slope fidelity.

So we really loved it. The Fluke was great for early low-power development of our 20kW power supply. We spent probably 3 months using it for hours a day. Our only complaint with the thing during the entire development period was the battery life. It's really, really easy to forget and leave the probe turned on all weekend - it really kills the batteries. An auto-off function wouldn't be any better, because that would just confuse you if it stopped working without warning. Anyway - the Fluke performed very well, and only showed problems when you really wanted to see non-ideal behavior in the silicon in the power supply. All in all, an excellent current probe. The problems came in when we started bumping the power supply switching frequency up to 70kHz, and the pulse currents past 30-40A. We didn't realize it at the time, but that was just asking for trouble.


Fluke doesn't really say much about how this probe works, but just looking at the clamp-on laminated core, it's pretty clear that they are using an exciting field and measuring the Bfield in a gap in the core, probably with a closed-loop zero-field arrangement just like what LEM is using for their current transducers. The actual datasheet mentions that the bandwidth is rated at 0-1kHz at the basic accuracy (~3%), but then says useful bandwidth 0 to 100kHz, with rise and falltime set at <4µseconds. All that seems a little incongruous, as the bandwidth is greater than the useful accurate frequenct. They mention all this in a small table on the bottom of the 3rd page of the direction sheet. After having thought about it for a while, it makes sense, but here's what they mention:

''Max. non destructive current:
0 to 2kHz 140A peak
2 to 10kHz 110A peak
10 to 20kHz 70A peak
20 to 50kHz 30A peak
50 to 100kHz 20A peak''

So what does that mean? Well, the magnetic material that makes up the sense core looks like silicon steel laminations - about 14mil laminations. Since the core loss in a magnetic material is strongly frequency dependent, I would have to guess that the max permissible dissipation in that core is reached at 50kHz with just 30A peaks, even if the core isn't saturating at that current level. In fact, it stands to reason that the core probably doesn't saturate until 140A (since that's the peak current limit). The useable bandwidth is 100kHz because that's the useful bandwidth in the closed-loop flux-cancelling circuit they are using to actually measure the current. This is conjecture on my part, but I assume this is what Fluke is doing. Anyway - the max non-destructive current rating went right under my radar. It also went completely under the radar of my willing accomplice in all this, Mr. [name redacted to protect the innocent]. All this makes some sense, as silicon steel at 14mil isn't really all that good past 20kHz due to the incredibly non-linear heating with B2. So maybe if they went with smaller laminations, or maybe a ferrite for the clamp core the probe might have fared better. Anyway - scroll down for the carnage.


Here's the probe out of the box, with the de-facto reference pearson current transformer next to it for size comparison

Here's the side of the clamp - intact

One day, we had the power supply up and running around 10-15kW into a DC load, powered from a giant variac off the 480V mains. The supply was working great, and we were monitoring the current as it transitioned from discontinuous to continuous mode. The power supply was working great, but the current measurement wasn't looking right. We figured that the 50+A pulses weren't a big deal, and that the probe was somehow going out of scale. So we just ignored it and went with the already-attached pearson probe measurement. An hour passed, and on a whim we looked at the Fluke signal. It was still there, but not at all like what we were expecting. Then we smelled something.


Click for larger - this is the front of the probe after it melted. Note the gap in the core - it was still making a signal, but not the right one. Here's a photo of the good jaws of our replacement probe

Here's the side of the probe. Totally melted. The jaws won't open anymore. It's fused shut.

Click for larger. I would like to say that it had a good life, but the truth is that we abused the poor little thing daily for several months. It died a mean, smoking, melty death. It deserved better. But, to be honest, what better endorsement of its abilities? The damn thing made it way into its ''destructive'' current levels for months on end at bandwidths it wasn't intended for and it never complained. When we switched over to high power testing, we should have known better than to kill the poor little thing with >3x its rated current at 70kHz. Oh well. If only we had read the fine print.


So anyway- this promised to be a pseudo- product review. I think it's great. We killed our first one out of ignorance. We immediately bought a replacement because, really, there aren't any clamp-on high-current wide-band probes out there at any reasonable price. So - I give it 5 Lithium Atoms out of 5, if for nothing else, putting up with our abuse.

RATING:

-Fred Niell

Design, text, etc. copyright 4/2008 Fred Niell