Scopes with isolated / floating channels?

[Unixon]

Are there any scopes with floating channels (optionally isolated channel grounds) that don't cost like a space shuttle?

[wraper]

You are basically limited to portable scopes or using regular scope with differential probes.

[tautech]

Are there any scopes with floating channels (optionally isolated channel grounds) that don't cost like a space shuttle?

Few.

Handhelds can be floated but probe reference leads are often commoned just like bench scopes.
There is the fully isolated 2ch SHS1000 series however their capabilities are pretty basic.

Differential probes and a bench scope of your choice are the normal workaround but if your channel isolation requirements are not demanding the ISFE or a couple of them might be a cheap solution.
https://int.siglent.com/article/detail-144.html

[Unixon]

You are basically limited to portable scopes or using regular scope with differential probes.

Hmm... that's bad.

I wonder why it is that way... digital isolators these days are good enough to be used even after pretty fast ADCs.

[HKJ]

They are fairly pricy, some possibilities are:
https://cleverscope.com/
https://www.picoauto.com/products/automotive-oscilloscope-kit/picoscope-kits-hardware
https://www.siglent.eu/product/1141124/siglent-shs1062-60mhz-handheld-digital-oscilloscope

[Unixon]

They are fairly pricy, some possibilities are:

So, adding channel isolation results in at least 10-fold increase in unit price... Yeah, that's a lot.

[PartialDischarge]

So, adding channel isolation results in at least 10-fold increase in unit price... Yeah, that's a lot.

You are basically limited to battery scopes, differential probes and fiber isolated probes.
For the price of the Cleverscope that only gives 1kV isolation I'd definitively go for a fiber isolated voltage probe

[wraper]

You are basically limited to portable scopes or using regular scope with differential probes.

Hmm... that's bad.

I wonder why it is that way... digital isolators these days are good enough to be used even after pretty fast ADCs.

Certainly not at a reasonable price. Also then you need a PSU with outputs isolated for every channel, low capacitive coupling and not injecting significant noise. Using regular SMPS becomes an issue.

[snoopy]

Are there any scopes with floating channels (optionally isolated channel grounds) that don't cost like a space shuttle?

Tek THS and TPS series scopes. Agilent U1620a scope. MicSig MS series. Siglent SHS800 ??

[Unixon]

Also then you need a PSU with outputs isolated for every channel, low capacitive coupling and not injecting significant noise. Using regular SMPS becomes an issue.

Well, yes. Analog front-end is low power anyway, ADC not so much but still not a huge load, filtering at low currents sounds easier than filtering at high current for everything at once.

[tautech]

Are there any scopes with floating channels (optionally isolated channel grounds) that don't cost like a space shuttle?

Tek THS and TPS series scopes. Agilent U1620a scope. MicSig MS series. Siglent SHS800 ??

Commoned BNC's so not isolated channel.
SHS1000 models are the properly isolated channel models.

[Fungus]

Are there any scopes with floating channels (optionally isolated channel grounds) that don't cost like a space shuttle?

Most scopes allow you to connect two probes without ground clips then display the difference.

Whether that's a good idea will depend on the voltages you're trying to measure. If you're hundreds of volts away from the scope's ground then you might break it.

[tom66]

I wonder why it is that way... digital isolators these days are good enough to be used even after pretty fast ADCs.

Well to give you an idea of the problem, a regular 8-bit 1GSa/s ADC needs 8 x LVDS channels plus 1 frame sync plus 1 clock.  Those 10 channels need to be isolated and each runs (with the exception of the frame clock) at 500MHz / 1Gbit/s.  And the frame clock needs to be sync'd with the sample so you can't get away with a slower device there.

So you need LVDS input and output isolators.  10x.  Per channel.  Plus power supply isolation (which needs to be low-noise)  and low speed data isolation for channel controls.   No way to share a quad DAC for your offsets.  Or a dual VGA amplifier for 2 channels.  It all has to be isolated and independent.  A quick Google shows the state-of-the-art ADI isolators that run at 1.1Gbps are $7 at 1ku, each doing 2ch. So that adds $35 to the BOM right away for each channel.

You could also isolate the analog front ends before they reach the ADC but this has other issues; usually such a circuit introduces nonlinearity errors or the sample clock gets injected into the signal which means you need to keep your sample clock synced with your ADC clock, which then means you need to look at building an analog isolator that can sample and hold at 1GSa/s.

And heaven forbid you want to do something beyond 1GSa/s such as 2 or 4GSa/s which are becoming more common on cheaper scopes.

[tautech]

And heaven forbid you want to do something beyond 1GSa/s such as 2 or 4GSa/s which are becoming more common on cheaper scopes.

Yeah it ain't easy or everyone would be doing it.

[Per Hansson]

Are there any scopes with floating channels (optionally isolated channel grounds) that don't cost like a space shuttle?

Well a space shuttle sets the bar pretty high: a German museum offered $12m for one of the aging Buran space shuttles.
A Fluke 225c scopemeter on eBay will only set you back around $2000 so it seems like a real bargain

[BravoV]

When it comes to isolated, floating scope, not too ambitious, say like decent performance, even  used, they never come cheap, especially top brands. 

Example Tek THS series, its basically floating "AND" two isolated channel to each other.




With that isolated channels, its possible to probe like the example below, and no, ordinary scope even floated or just powered by battery can not do this if both channel are not isolated.

[bdunham7]

I think the problem is that with the proliferation of cheap scopes (big bangs for bucks) your expectations about price (costs like a space shuttle) have changed.

A Fluke 2-channel scopemeter was $3K in the early 90's.  And that was isolated, but not between the two channels.  Nice but relatively basic Tek scopes (no isolation) were $3K and up.  A $1K scope was an ultra-basic 20MHz 'educational' model.  Regular DSOs have become available at a small fraction of what they would have been worth back then, but isolated scopes haven't attracted enough attention for players like Rigol and Siglent to make a competing product.

Isolated channel scopes are not easy to make and are a niche market.  They are available, but generally a lot less bang for many more bucks.  The only users that are willing to pay the price for them nowadays are industrial users that need the ability to look at 3-phase power and motor control systems and they shell out $5K+ for a Fluke 190 series or a Tek TPS series without even blinking.  The $5K TPS2024B is a recycled 15-year old design, but apparently still sells.

You can find these on eBay in working order for $1K and up if you are patient.  But with today's expectation of feature-laden 350MHz hackable bargain scopes that cost less than a good bench meter, that probably seems like too much.

[Unixon]

Well to give you an idea of the problem, ...

I perfectly understand that it is not possible for high end scopes, OK, let it be a different kind of scope with like say only 500MSa/s or even slower, let it be only as high bandwidth as the isolators allow, fine, this would be already a huge help in a lot of cases. Many times I've run into situations where I had to simultaneously and synchronously measure 2+ isolated points that could not be brought to a common ground, even if it was safe the device would stop working if these reference points are connected.

[tautech]

Given that differential probes these days are affordable I believe the incentive for manufacturers to design new isolated scopes has largely dried up.
Eg. Take a $500 DSO add 4 diff probes and you're still well under $2k and with a modern capable DSO with good mem depth plus all the other features a modern DSO provides over old designs.

[commongrounder]

I think the least expensive isolated channel scope is the Hantek DSO1062S, for around $630us. I have the DSO1102S that I bought in 2013 as a stand-in while my Fluke 199C was being calibrated.  The firmware can occasionally be a bit flakey, but it worked fine for my bridging audio amplifier testing needs.
Also, differential probes aren’t truly isolated, are they?  The plus/minus inputs are referenced to ground, usually through about four megohms per leg.  I own two of them, though, and they are great for safe higher voltage measurements.

[shakalnokturn]

Well to give you an idea of the problem, a regular 8-bit 1GSa/s ADC needs 8 x LVDS channels plus 1 frame sync plus 1 clock.  Those 10 channels need to be isolated and each runs (with the exception of the frame clock) at 500MHz / 1Gbit/s.  And the frame clock needs to be sync'd with the sample so you can't get away with a slower device there.

So you need LVDS input and output isolators.  10x.  Per channel.  Plus power supply isolation (which needs to be low-noise)  and low speed data isolation for channel controls.   No way to share a quad DAC for your offsets.  Or a dual VGA amplifier for 2 channels.  It all has to be isolated and independent.  A quick Google shows the state-of-the-art ADI isolators that run at 1.1Gbps are $7 at 1ku, each doing 2ch. So that adds $35 to the BOM right away for each channel.

You could also isolate the analog front ends before they reach the ADC but this has other issues; usually such a circuit introduces nonlinearity errors or the sample clock gets injected into the signal which means you need to keep your sample clock synced with your ADC clock, which then means you need to look at building an analog isolator that can sample and hold at 1GSa/s.

And heaven forbid you want to do something beyond 1GSa/s such as 2 or 4GSa/s which are becoming more common on cheaper scopes.

I received a second hand battery powered Metrix (Chauvin-Arnoux) Scopix III today, analog bandwidth is modest although decent for a "handheld" with 200MHz on the top model.
The official specs are 4x 2.5GS/s simultaneous, 50k points/channel simultaneous and 12 bits vertical resolution which sounds pretty impressive, but is a half-lie.
The way they go about it is having (up to) 4 truly isolated channels containing the analog FE's and control, a fast CCD and a slowish 12 bit ADC on the floating side.
The 12 bits cross the gap to the common ground referenced side through cheap Ethernet type isolation transformers.
I'll do a teardown when I get round to it in 2022, just a little sceptical on the CCD having a good enough noise floor to feed a 12bit ADC, or maybe that's marketing wank.

[snoopy]

And if you can afford one of these

https://cleverscope.com/news/cs448/

https://youtu.be/I7ppDNLlEL4