Interface Isolation on Entry Oscilloscopes? (+USB Scopes, A Hard Lesson Learned)

[Danist]

Question:

Do entry level bench oscilloscopes (i.e. Rigol DS2k/1k series, Tektronix TDS series, etc.) have any sort of isolation on their data interfaces, such that dumping of current to the probe shield goes to Earth rather than the ground reference of the interface?

Background:

In specific reference to EEVblog #279 How NOT To Blow Up Your Oscilloscope! and the other (very excellent) advice not to bother with cheap USB oscilloscopes… I had an incident, more than several years ago, involving a Hantek DSO USB oscilloscope, which I now find mildly amusing. Please, giggle along, after you watch #279 again. Yes, go watch it again now.

As the most memorable of all these stories start, I was working with power equipment. To wit, I had taken two legs of three phase mains AC (208V, in the US that’s two 120V lines of 120 degrees) at service amperage and transformed them to a balanced 240V system bonded to ground. Then, I had some isolation transformers in various configurations, some bonded, some floating. These, in turn, were connected to various DC power supplies (on the order of 15V 100+A each). Now, to be clear, I had taken very, very specific safety precautions in how this was setup, with various layers and technologies of breaking and ground faulting. It's not a bad idea for a study, I suppose, if done right; the basic purpose was to measure end DC power quality among various ground noise scenarios. Do not try this at home -- if you do proceed against all rational advice, do so at someone else's home with someone else's stuff. Ahem, excuse me, my attorney has advised me not to comment further. *** (Psst, have good health coverage, and a life policy if you have dependent family) ***

Everything was going along just fine... I had started probing, and collecting data… etc., etc. All well and good.

Then the lights flickered. And a very impressive poof of smoke came from... my computer rack! ?!?

Wow! How cool is that?!? (if by “cool,” you mean the destruction of a high-end workstation)

Stupid! Stupid! Stupid! Gaaah!!! Facepalm!

Of course, the power equipment was still humming along -- the safety mechanisms did exactly what they were supposed to do. And on resetting and cycling, every piece of power equipment came back online exactly as it was intended to.

Well… except my high end workstation. Whoops! 

I figured I’d save some money on an oscilloscope that was “good enough” to look at basic power quality -- even just using rough analogue TLAR (the official acronym for “that looks about right”). I mean, hey, what’s a cheap little oscilloscope when you can afford to let it blow up, and buy another -- right?

Except when you attach that little USB gizmo to your PC -- when you blow up the oscilloscope, you can blow up the PC motherboard, and several pieces of attached USB equipment. The CPU, RAM, and expansion boards were salvageable... maybe because my motherboard was made to go down with the ship, with folks such as myself in mind as repeat customers.

On diagnosis, I found the originating problem was a combination of an unexpected ground loop and a cable I had miswired. That does happen. And yes, it was my fault. I am not perfect, nor will I ever claim to be. It will happen, and you plan for it. (Hopefully, on paying much attention to the material here, it will happen much less frequently -- but it will still happen). And in my case, the intended mechanisms correctly triggered to protect both operator and equipment. But I hadn’t given even the slightest consideration to the USB bus on the oscilloscope -- that is the last thing on your mind when you are wiring up these high power whackadoodle sledgehammer contraptions.

I thought to myself “oh yes, that is good, thank you, thank you!” when Mr. Jones mentioned it in #279. Similar situations are true when your USB bus serves as the oscilloscope earth ground. Thank you sir, thank you very, very much -- you said it... here, here! (Danist raises his glass and motions for a proper toast)

Still, an error which results in a fault through one’s tools is better than one which faults through oneself.
In retrospect (hindsight being more perfect than foresight), some sort of differential probe setup might have been a much better (but more expensive) idea.
But, as remains to my question -- do entry level scopes have isolated data interfaces… lest one dump to ground through the probe shield, is there any mechanism for it to bypass the interface ground in favor of mains Earth?

[BravoV]

Possible cheap & proven solution -> http://www.analog.com/en/evaluation/EVAL-ADUM4160/eb.html

[Danist]

Yes, I may start isolating the USB devices of anything I use in electronics. I also would not normally assume that an ethernet interface to a scope would have a problem. But is it known if any of the entry level scopes already have a degree of isolation?

[nctnico]

The real lesson learned should be to use differential probes. Using an isolation transformer is a really bad practise from the days a GFI was still horribly expensive. Nowadays its better keep things properly grounded and be sure to use a (tested) GFI.

[Balaur]

Thank you for the story.

I also have a Hantek USB DSO (the 2150 to be precise) that I've bought almost 5 years ago. While the usual remarks about the quality of the device are perfectly applicable, it works for me until I'll be able to find more bench space (I don't even have a bench really).

Yeah, the DSO is not electrically isolated. I only use it with a notebook that is floating. Even so, most of the time, I'm using the notebook on batteries and wireless.

[kfitch42]

I also would not normally assume that an ethernet interface to a scope would have a problem.

Hmm silly question, the shield on STP (shielded twisted pair) ethernet cables (as opposed to unsheilded/UTP), is that normally connected to earth ground?

Could you create a ground loop by connecting STP between two earthed devices? e.g. Running ethernet directly from a scope to a PC.

I guess normally that ethernet cable goes to a switch with a two prong power plug, so not too big a deal, right?

[Monkeh]

Could you create a ground loop by connecting STP between two earthed devices? e.g. Running ethernet directly from a scope to a PC.

Yes. This is why STP (and most shielded data cables, actually) is inadvisable for an amateur setup. If you don't know what you're doing it just makes things worse in a variety of ways which are either difficult to trace or very, very obvious when things catch fire.

I guess normally that ethernet cable goes to a switch with a two prong power plug, so not too big a deal, right?

*snort*

Real switches are earthed, not made of plastic. And the shields would be connected even on the few floating switches I've seen. Cheap plastic routers are another matter entirely, of course. It's cheaper just to make the whole thing out of PVC, who cares about signal quality?

[Danist]

The real lesson learned should be to use differential probes. Using an isolation transformer is a really bad practise from the days a GFI was still horribly expensive. Nowadays its better keep things properly grounded and be sure to use a (tested) GFI.

Yes, spot on with the differential probes. That's on my list if I ever want to try a similar project in the future. I am also now in the habit of neutral/ground bonding all of my isolation transformers -- it's actually one of the few exceptions to the practice of only bonding at the mains panel (aside from being code exempt as a "separately derived system"). I like the safety concept of having a common reference and a low-resistance fault path which is separate from anything carrying working current.  To wit, some of the transformers I was using were for hospital use, where floating equipment would be physically isolated. I actually can't recall when having ground isolation has been useful, except perhaps for some ham radio experiments where the service ground was a long dirty run up several stories.

[Danist]

Thank you for the story.

I also have a Hantek USB DSO (the 2150 to be precise) that I've bought almost 5 years ago. While the usual remarks about the quality of the device are perfectly applicable, it works for me until I'll be able to find more bench space (I don't even have a bench really).

Yeah, the DSO is not electrically isolated. I only use it with a notebook that is floating. Even so, most of the time, I'm using the notebook on batteries and wireless.

Yes! That was it! I would still be using it, had it not blown up. I wouldn't say it is bad, since it was really, really useful for a lot of projects. The experience of using any scope opens up a much wider world, and is magnitudes better than having no scope at all. I fell back on a Tektronix 465 which recently died of CRT failure. That was a great piece of equipment, though it did have a big footprint. From a value perspective, I could both buy another good second hand Tek (maybe even a 22xx series) for about $100, or a new Hantek DSO 2150 for $200 -- and of the two, I would go with the Tek due to the price vs. features beyond what I could do with a multimeter alone. But ultimately, for a value piece of bench equipment which I would take care of and not treat as blow-up-able, putting that money towards a new digital storage scope in the ~$800 class seems like a better long term idea... especially since I am doing less power stuff and getting more into microcontroller work, and usb interfaces, etc.

[Danist]

Could you create a ground loop by connecting STP between two earthed devices? e.g. Running ethernet directly from a scope to a PC.

Yes. This is why STP (and most shielded data cables, actually) is inadvisable for an amateur setup. If you don't know what you're doing it just makes things worse in a variety of ways which are either difficult to trace or very, very obvious when things catch fire.

I guess normally that ethernet cable goes to a switch with a two prong power plug, so not too big a deal, right?

*snort*

Real switches are earthed, not made of plastic. And the shields would be connected even on the few floating switches I've seen. Cheap plastic routers are another matter entirely, of course. It's cheaper just to make the whole thing out of PVC, who cares about signal quality?

A few really good points in this exchange!

I don't normally use shielded cable, mostly because the price is not worth it for what I'm doing. However, I do generate lots of powered radio signals from HF to microwave. So I just stuff my ethernet bundles into grounded conduit when necessary -- per foot, conduit is much cheaper than the price difference between shielded and unshielded cat 6. And really, who cares about ethernet quality when it's just a backup for the infiniband anyway? (Yes, I like infiniband, so there)