Recommend a 4ch isolated scope

Thread Starter

strantor

Joined Oct 3, 2010
6,782
I'm looking for a scope around 100MHz with 4 (or more ;)) floating AKA isolated AKA differential inputs. I'm aware of Tektronix TPS2000 & THS3000, Fluke 190-104,-204,-504, and the crappy PC-based offerings from TIEPIE and PicoTech. Am I missing anything? I've searched extensively and I'm pretty sure that what I listed above are the only current products out there (correct me if I'm wrong) so I guess I'm looking for obsolete used equipment that's no longer advertised. Think back; can you remember coming across one in days gone by, that might be had on eBay? Handheld or portable form factor preferred.
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
Oh, and I'm also currently fighting a language barrier to find out the existence/availability/price of Chinese company MICSIG's (seemingly vaporware) "T-Book" product - a 4 channel supposedly isolated input (optional) tablet scope. I've been quoted $1600 for something, stamped with 200MHz - awaiting confirmation about the isolated part.
 

t06afre

Joined May 11, 2009
5,934
If I understand you correctly you need the channels to be galvanic isolated from each other. So using an isolation transformer (on the scope mains) will not be good enough in this case.
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
If I understand you correctly you need the channels to be galvanic isolated from each other. So using an isolation transformer (on the scope mains) will not be good enough in this case.
Correct. For example I need to measure 3 phase power, all 3 phases at once, plus another voltage/current; using math functions and multiple channels per measurement won't cut it.
 

t06afre

Joined May 11, 2009
5,934
What kind of frequency range do really need? Do you need the full 100MHz or can make some board using isolation circuits or audio transformers
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
What kind of frequency range do really need? Do you need the full 100MHz or can make some board using isolation circuits or audio transformers
I've been down enough rabbit holes to know when I'm staring into one. I am already on enough ADD-infused tangents. I think I would prefer to buy something than make something in this case. I'm in the market for a scope anyway so I figure I might as well buy what I need first thing, instead of making.

BUT... I will look into it. I did briefly look into it already (like for 5 min or less) and decided that there is a reason why differential probes cost so much, that high speed circuits are out of my league, and that it would probably cost me as much to make as to buy. It deserves a more thorough investigation though.

One other thing I might be interested in, is a rready-made isolation interface, like you suggested I make. Like a 4 channel differential interface. I also looked for a product like that and only found one company selling something like that; it was Siglent IIRC, it was only two two channels, the specs were pathetic, and it did not seem readily obtainable. Do you happen to know of something like this for sale? I really do not want to buy and carry around 4 cumbersome active differential probes.

Edit: to answer your question, I don't really know how much bandwidth I need. Most scopes sold for what I need to use it for (the ones mentioned in post #1 - for measuring 3 phase VFD output) are in the 100MHz range. I've seen VFD PWM go as high as 30KHz, but I'd also like to be able to scope comms on industrial field busses, which go much higher than 30KHz. I reach toward my anus and withdraw this 100MHz figure; I really have no idea if thats insufficient or overkill.
 
Last edited:

THE_RB

Joined Feb 11, 2008
5,438
That isolation requirement *4 sounds expensive.

It's going to require 4 separate isolated PSUs connected to 4 separate isolated front end amps, and 4 isolated precision linear analogue opto circuits OR 4 ADC front ends and 4 isolated high speed digital optos.

That's all in the realm of large, complex and very expensive.

I really doubt your $1600 China quote is going to resemble anything like that! It will be a display, an ARM processor, and a fast ADC digital front end that multiplexes between X input channels. Like other China digital scopes; cheap and cheerful.
 

GopherT

Joined Nov 23, 2012
8,009
Buy four scopes and add isolation transformers to power each. Note that the chassis and ground clips of a scope are connected to ground pin on the outlet so input isolation transformer will be required for power.

Scopes can be had on eBay for about $125 with shipping. I haven't priced the isolation transformer but back-to-back transformers from a pair of...

RB was right, any solution to this problem is going to be expensive.
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
They aren't as expensive as you guys are probably thinking. I'm not sure what you consider "expensive" for a scope, but I've seen non-isolated scopes approach $100K.

The Tektronix THS3014 (100MHz 4 isolated channels) costs $4250 brand new, and has not been out long enough for me to find one used.

The Fluke 190-104 (100MHz 4 isolated channels) costs $4000 brand new and I have seen sold used on ebay for barely north of $2500.

The Tektronix TPS2014 (100MHz 4 isolated channels) sells for $3800 brand new and can be found used for $2000.

These prices are more than I can afford to pay at the moment, but I think they are more than reasonable given the level of technology involved. If someone gave me a 4 channel mains-referenced scope, it would cost me more to buy 4 100MHz differential probes than what these scopes cost. (actually, I would only need 3, right? if I floated the scope for the 4th channel?)

Don't worry, I'm not sending $1600 off to china any time soon. If there were any 3rd party teardowns/reviews or even one-line blurbs then I might make a semi-serious consideration, but as of yet I have seen absolutely zero proof that anybody on the planet has ever purchased and received one of these tablet things. I cannot even find a supplier outside of unproven alibaba sellers.

Edit: found a used THS3014 for $3400 on ebay, from a seller with zero feedback.
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
Buy four scopes and add isolation transformers to power each. Note that the chassis and ground clips of a scope are connected to ground pin on the outlet so input isolation transformer will be required for power.

Scopes can be had on eBay for about $125 with shipping. I haven't priced the isolation transformer but back-to-back transformers from a pair of...

RB was right, any solution to this problem is going to be expensive.
I am looking for a scope I can use in the field for troubleshooting motor drives. Lugging 4 bench scopes around is kinda the exact opposite direction I need to go.
 

crutschow

Joined Mar 14, 2008
34,285
How about using four isoated amps such as these. They are low frequency devices (15kHz) but that should be adequate for power line measurements. They generate isolated power for the input (isolated) side of the amp from a single supply at the output. They cost about U$30 each but that's a lot less than a differential probe.
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
How about using four isoated amps such as these. They are low frequency devices (15kHz) but that should be adequate for power line measurements. They generate isolated power for the input (isolated) side of the amp from a single supply at the output. They cost about U$30 each but that's a lot less than a differential probe.
I'm not measuring power line frequencies; I'm measuring the output of VFDs which is a 3 phase PWM I have seen up to 30KHz, but usually around 2-20KHz. Probing the Analog devices website in the same section as the AD202, I found the AD215 which is a 120KHz version that sells for $75. If I go by the "10X bandwidth for measuring a square wave" thumb rule, that solution might be sufficient 75% of the time. But if I'm going to roll my own low performance device, I would hope the parts be cheaper.

But, looking at the functional block diagram in these data sheets, I see what they're doing and maybe I could emulate it. Looks like they take an analog signal into an ADC, send the digital data through a transformer, and on the other side of the transformer is a DAC that turns it back into analog. Discrete ADCs and DACs are available much cheaper and much much faster than these self-contained modules. I don't really dig the idea of using a transformer; I think optoisolation would be easier to implement if using discrete parts.

So what do you think? probe>super fast ADC>opto isolator>super fast DAC>scope input. ADCs to be powered by tranfsformer isolated supply. Something tells me there's one or more "gotchas" that are the reason why differential probes cost so much.

Thanks for the tip, but that thing is much too big. searching for scope isolator I found nothing smaller though. How big is too big? I haven't defined. Let's just say if it's bigger than a bench scope it's too big. (compact bench scope, ex: Tek TDS). Remember, this is a field device; ideally It (PLUS the oscilloscope ) fits in a backpack or small shoulder bag.
 

crutschow

Joined Mar 14, 2008
34,285
I'm not measuring power line frequencies; I'm measuring the output of VFDs which is a 3 phase PWM I have seen up to 30KHz, but usually around 2-20KHz. Probing the Analog devices website in the same section as the AD202, I found the AD215 which is a 120KHz version that sells for $75. If I go by the "10X bandwidth for measuring a square wave" thumb rule, that solution might be sufficient 75% of the time. But if I'm going to roll my own low performance device, I would hope the parts be cheaper.

But, looking at the functional block diagram in these data sheets, I see what they're doing and maybe I could emulate it. Looks like they take an analog signal into an ADC, send the digital data through a transformer, and on the other side of the transformer is a DAC that turns it back into analog. Discrete ADCs and DACs are available much cheaper and much much faster than these self-contained modules. I don't really dig the idea of using a transformer; I think optoisolation would be easier to implement if using discrete parts.

So what do you think? probe>super fast ADC>opto isolator>super fast DAC>scope input. ADCs to be powered by tranfsformer isolated supply. Something tells me there's one or more "gotchas" that are the reason why differential probes cost so much.
......................
For a 300kHz bandwidth (10 times 30kHz) the Nyquist minimum sample frequency is 600kHz. That will give you a serial bit rate of 4.8MHz (plus sync bits) for an 8-bit converter (0.5% resolution) which could be sent over a high speed 10Mbs opto-coupler (such as one of these) to a high speed D/A.

Note that each A/D will require a separate isolated power supply as well as one for all the output D/A's.

So how cheap and easy that is to do depends on your definition of those terms. I don't know that it meets my definition. :rolleyes:
 

GopherT

Joined Nov 23, 2012
8,009
Strantor, what are you hoping to see on the scope traces? The simultaneous voltage outputs of the three phases, duty cycle, wave shape, frequency, other?
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
For a 300kHz bandwidth (10 times 30kHz) the Nyquist minimum sample frequency is 600kHz. That will give you a serial bit rate of 4.8MHz (plus sync bits) for an 8-bit converter (0.5% resolution) which could be sent over a high speed 10Mbs opto-coupler (such as one of these) to a high speed D/A.

Note that each A/D will require a separate isolated power supply as well as one for all the output D/A's.

So how cheap and easy that is to do depends on your definition of those terms. I don't know that it meets my definition. :rolleyes:
So what you're saying is, that you see the rabbit hole too?
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
Strantor, what are you hoping to see on the scope traces? The simultaneous voltage outputs of the three phases, duty cycle, wave shape, frequency, other?
Im hoping to see everything I would see on a 100MHz isolated scope! Haha, ok enough smartassery... I want to compare voltage and current wve forms on 2 phases simultaneuously; I suppose this scenario would really only require one differential probe bare minimum - float the scope for one voltage input and use the diff probe for second voltage input, and the current probes are not electrically connected so no issue there. But I would also like to be able to see all 3 voltage output waveforms simultaneously; this is more of a desire than a requirement. I would like to be be able to zoom in to the individual pulses and examine the wave shape; I suppose I don't need to do this on more than one channel simultaneously. Ineed to be able to probe the individual output transistors and measure on/off time, turn on/turn off time and current (the Tek TPS2000 will do automatic switching loss calculations too) and it would be really helpful if I could display this info for 2 or more transistors simultaneously for comparison. I need to be able to scope the DC bus for dips/spikes and be able to correlate these anomalies with other floating measurements, including signals. I need to be able to scope input voltage for anomalies and correlate that with output anomalies. So I suppose I don't literally require 4 isolated channels but it sure would be nice. I could probably get away with a couple of diff probes, or maybe even just one diff probe in a pinch. But for every isolated channel I subtract from the picture, that adds something I must write down or take a take picture of, and then go back and do math, and most times that doesn't even work...

These are variable frequency devices and things are constantly changing. PID in the drive is continually adjusting the output frequency and PWM duty cycle (even the PWM frequency) to maintain a fixed speed and torque, and then you have variable speed/torque setpoints coming in from external process PIDs, so things are changining on several tiers and getting a scope to trigger on the drive outout is notoriously problematic. Best bet is to use single shot capture across all the signals you are interested in, and display a still stopped picture of what was going on the millisecond before last. Writing stuff down or taking screenshots, then moving probes around doing comparison can be a valid test or it can be apples to oranges. Things can change radically between test samples. So I need the same reasonable single shot capture rate and buffer and high res zoom that an ordinary scope provides, but isolated. not sure if the things we have discussed so far can deliver that, outside of a 4ch isolated scope. Even the diff probes seem quite crippled in comparison, unless I am to spend more for the probe than the scope.
 
Last edited:

Thread Starter

strantor

Joined Oct 3, 2010
6,782
For a 300kHz bandwidth (10 times 30kHz) the Nyquist minimum sample frequency is 600kHz. That will give you a serial bit rate of 4.8MHz (plus sync bits) for an 8-bit converter (0.5% resolution) which could be sent over a high speed 10Mbs opto-coupler (such as one of these) to a high speed D/A.

Note that each A/D will require a separate isolated power supply as well as one for all the output D/A's.

So how cheap and easy that is to do depends on your definition of those terms. I don't know that it meets my definition. :rolleyes:
Why does each DAC require a separate supply? As soon as I connect them all to the mains referenced scope they will be all made common, right?

Going back to my thoughts in post 14, its seems a little inefficient and silly to convert analog to digital and back to analog, to send to a scope that's just going to convert the analog back to digital. Having never examined a DSO front end (so don't laugh too hard) it seems like it would make more sense to hack the scope and send the optoisolated digital info straight into the scope's data bus. Ever seen this done? Thoughts on how painful it would be?
 
Top