TSP

In this episode Shahriar reviews the Saker fiber-optically isolated active voltage probe XFVP. The fully galvanically isolated probe can reject common-mode signals in the tens of thousands of volts range. Signals with fast rising common-mode can also be rejected thanks to its high bandwidth of beyond 250MHz. Information about the Saker probe can be found here:
saker-mv.com/fiber-isolated-v...
This review is organized as follows:
00:00 - Introduction & optical probe overview, principle of operation, dividers and accessories
05:07 - Full teardown of the probe head, probe body and 500x attenuator box
09:28 - Differential probe versus an optically isolated probe architecture
11:22 - Excess noise, SNR, dynamic range, THD and auto-calibration experiments
18:13 - Measured bandwidth of probe, S-parameters, gain ripple, phase response and group delay
19:47 - Measuring a GaN switching behavior with 400V of common-mode signal swing
27:03 - Concluding remarks
www.TheSignalPath.com
/ thesignalpath​​​
www.Patreon.com/TheSignalPath

[@twicecookedpork6220]

based on the silkscreen font, they are designing their PCBs in KiCAD. Nice to see open source software being used in the wild!

[@godfreypoon5148]

You can get better CMRR by having your assistant yell the results across the lab. But they usually don't like being floated to 50kV, and the bandwidth is terrible.

[@IanScottJohnston]

I know it's high bandwidth stuff, but my eyes were crying looking at the traces on the Pcb's at all sorts of wierd angles and routes!

[@skataskatata9236]

why? it is better.

[@ehsandaneshvar]

I was trying to capture high side gate waveform from the 400V inverter I'd built recently. then I searched about isolated probe performance and find you,,, Nice explanation an thank you. right now I would check my Budget again.. :)

[@garci66]

The conector for the control fiber looks like your "pof" (plastic optical fiber) connector that was being pushed as a standard for in home use a few years ago. It was a 2.2mm plastic fiber using visible light and no special tools for terminating. look for "optolock" cobnectors and POF internet...

[@mtwieg24]

Thanks for the teardown Shahriar, maybe someone will donate an IsoVu probe so we can do a shootout. 250MHz is actually terribly slow for GaN-based converters, especially lower voltage devices (rise/fall times ordinarily are

[@zaprodk]

Oh all those funky angle PCB tracks ... It's hard on my CDO!

[@Chriva]

I'll never understand the difference between tsp and tnp. Do I watch both? -Heck yeah! :D

[@Darkknight512]

If you can't tell the difference, does that mean the SNR is too low?

[@Chriva]

@@Darkknight512 Best dad joke I've read in ages lol. Thanks for the chuckle :D

[@EdwinSteiner]

It has something to do with the mood of his cat, I think.

[@PeregrineBF]

Video length usually. There are a few exceptions, but generally 20 is TSP. It's intended to be based on how much detail he goes into, with the TSP videos having more detail.

[@andersvandegevel8355]

One negative point of AA's over PP3's is they are far more likely to leak

[@zyeborm]

I'd really love it if with these reviews there was a "budgetary price", mostly so I know how much I can't afford it by but also so I know if it's worth bothering to even look. With as many disclaimers around it as needed of course. It's just I'd like one of these but even their website doesn't mention a price and has an email as the main point of contact. I don't know if this thing is a $1K or $50K, both of which seem like a plausible order of magnitude price for the product. One of which I could justify, one I could not lol.

[@w2aew]

I wonder what the CMRR vs frequency is on that probe.

[@Thesignalpath]

I was going to measure it, but didn't get a chance to. In the datasheet, it is listed as: CMRR: < -110dB from DC to 10MHz ~ -90dB at 100MHz ~ -65dB at 200MHz

[@James-wb1iq]

I'd love to see what the MXO can do with that measurement using 2x probes and math on the 12 bit signal. If you calibrate the the channels before the measurement, I think it can do it almost as well as the isolated probe? Edit: I'd love to see an investigation of the MXO's high resolution mode now... Looking at the datasheet it claims 18 bit resolution at 10 MHz bw - but this requires a 14 bit ADC at 5 GSps. If the ADC was only 12 bit, getting an extra 6 bits would require oversampling by 4^6 = 4096, or ~82 GSps. Presumably their claim of a 12 bit ADC is just modesty.

[@xinzhang851]

did you ever try to tear down Tek's optical isolation probe?

[@Thesignalpath]

Not yet!

[@kenhenderson7858]

It would be interesting if you could repeat the measurement, but use a shorted MMCX connector (all pins soldered together) and connect it to the Vsw node (high side source). That way you could see the actual CMRR with the 400V edges.

[@xiankangwang9283]

Modulating the light intensifty is nearly the simplest way to achieve optic-isolated probe.Just wonder if the gain of the proble changes when you bend the fiber or the room temperature changes? I guess Tek's probe adopt frequency modulation with much more compliated compensation and controls. Hope to see you tear down tek's probe and make a comparision.

[@zyeborm]

You'd need a fairly substantial bend to get a large impact on the output power I think. You could be very fancy fairly easily if you wanted to control for that though. "Just" send one signal through at a constant output power on one light wavelength and use that to adjust the gain on the modulated signal on another wavelength The optical components to do that are pretty common in the fibre communications world. Of course you've now added another source of noise. I imagine the "cal" button there exists mostly for that reason.

[@supernumex]

micsig has a new optical probe, too. It would be cool to see how that one compares.

[@fantasticoadidas]

On our sales rep website they mention that the MicSig can only be used with MicSig Oscilloscope. Looking at the images the MicSig oscilloscope do have a weird connector like the old LeCroy equipment had. This is nice, but it also is cumbersome because it hardly is possible to mix probes from other manufacturers. A sort of 'vendor lock-in' so to say.

[@supernumex]

@@fantasticoadidas That's true of most scope specific probes unfortunately. I have a tek current probe, and even though tek scopes aren't my favourite, I have to use one.

[@supernumex]

@@fantasticoadidas Actually, check out the new EEVBlog video on the micsig optical probe. Standard BNC interface!

[@chenjon4907]

@@fantasticoadidas Micsig's optical probe is standard BNC interface though.

[@markp5726]

I wonder how much current the head draws, and if it could be powered by pumping relatively intense light through the "oob" fiber. Like the Dallas 1wire protocol (power and data on one line), but with fiber.

[@pelor92]

I am fairly sure tektronix does that in their isovu probes, however they are 3/4x the price

[@joelstienlet1641]

I just let my USB oscilloscope float 😅(I'm very careful, no panic!). I use an USB 2.0 high speed (not fool speed!) isolator, you can find these ready to use for audio applications for ~150$ (I had to modify mine though, didn't work out of the box) (fiber usb isolators also exist, but I've not tested that yet). I built a highly insulated floating 5V power supply by coupling a stepper (used as a generator) to an mains AC motor with an insulating shaft. I successfully used this setup to measure signals with 400VDC common mode offset on some power board.

[@jakubniemczuk]

Nice idea. Better then trafo isolated classic scope that you have to touch.

[@zyeborm]

I mean I wouldn't trust it to 10Kv, but I use the $5 USB isolators off aliexpress for development just so I don't take out my computer doing something dumb. They put out modest 5v power and do actually appear to be truly isolated. Just for anyone who wanted to try it on a budget.

[@joelstienlet1641]

@@zyeborm just my 2 cents: one thing to be careful about when chosing the right isolator: USB 2 has three speeds: low, fool and high speed (I spell it "fool" instead of "full" on purpose, many get confused that "full speed" is actually much slower than high speed!). Some isolators only handle low speed USB 2 packets (which are also USB 1 packets). Some isolators handle low and fool speed. The one that I have does high speed. Depending on the oscilloscope, low/fool speed may or may not be sufficient.

[@user-bs1hp2cu5u]

What tripped me up when trying to follow along with the setup at the end of the video is that you're saying that you're measuring the gate-source voltage of the low side GAN with the passive probe, but that wouldn't show the CM swing that's present on the high side measurement (you'd see the 8-ish volts produced by the isolating power supply). Checking the app note confirms that you're connected to TP3, not TP2, which measures the drain-source voltage, which indeed corresponds to the CM swing.

[@paulpaulzadeh6172]

I have R&S RTH1004 which used to design microcontroller soft start main power supply , I notice then isolated oscilloscope is must , specials for designing EV cars , with 600v on one side and 3 volt to controller side , there is no other good scope from other brand except R&S, could you review RTH1004 ?

[@ZergZfTw]

How does the transmitter encode the measurement over the fiber? Does it just modulate the intensity of the light? Or is it more complicated?

[@Thesignalpath]

It is intensity only.

[@ZergZfTw]

@@Thesignalpath That's what I was thinking. How does the probe calibrate DC values? Does it have its own voltage reference to compare measurements too?

[@marknations9651]

@@ZergZfTw There is an optional module that you can plug into the input to do offset and 2 point gain calibration. The Lecroy HVFO108 is around the same price as this probe and has a much more sophisticated architecture which is more resistant to offset and gain drift.

[@andrewyanke1787]

I'm in over my head but now I have to watch it all since W2aew is in the house. I've binged watched him so much he's like my daddy and I'm his son he doesn't know about.

[@ikocheratcr]

In order to reduce noise , could an optically isolated probe work better if the signal on its input is modulated, say with 1GHz, put in the fiber, and then demodulated? My guess is, it will have better linearity, but I do not know if it is really worth for noise and delay versus complexity and battery life.

[@zyeborm]

You'd also have the added noise from those conversions too, nothing is free ;-)

[@akosbuzogany2752]

What surprised me is the flat phase response. What is the main contributor/culprit to it?

[@Axotron]

With a relatively large delay through the probe and its cables, the delay will dominate the phase response and result in a steep slope that makes it pretty hard to see any deviations from a perfectly straight line. It would be more interesting to see a plot of the group delay than the plot of the phase. Or a plot of the phase where port extension/electrical delay has been applied to get rid of the part of the phase that is caused by a pure delay.

[@AndrewZonenberg]

@@Axotron A measurement that I like to use for this type of thing is "deviation from linear phase". First step is to unwrap the phase, so it keeps on going instead of wrapping modulo 360 degrees. Then you fit a line to it between two frequencies and plot the delta between that line and the actual measurement. This makes small nonlinearities much more evident. Surprisingly, this isn't available as a math function out of the box in most software so I had to write it myself for ngscopeclient which is what I use for this kind of analysis.

[@motherjoon]

Adding external dongles to add further attenuation just sounds a little weak. For 6k, there should be an attenuation selector on the front end.

[@skataskatata9236]

high voltage, bro.

[@tamaseduard5145]

👍🙏

[@hinz1]

WiFiScope might be the simple option, ditch optical cable, just float the entire scope and connect by WiFi ;-)

[@jeroenk3570]

Is there a technical reason the designers didn't put a BNC socket on the receiving end but put a bit of coax cable with a BNC plug on instead?

[@lbgstzockt8493]

Probably want to prevent the probe body from breaking the bnc connector of off some insanely expensive scope, just think of the leverage that thing could create when compared to just a cable.

[@jeroenk3570]

@@lbgstzockt8493 Thank you.Sorry, I was not clear. I meant to say a BNC panel receptacle connector so you can attach your own cable to go to the instrument or replace it easily when it's broken.

[@Connexelectronic]

I'm going to purchase the CYBERTEK OP6032 which is 200MHz isolated probe similar with the reviewed one and i was wondering if anyone have some feedback about it. Any chance to see such a detailed review and see if they worth the money, sold in China for about 3k USD.

[@chenjon4907]

Micsig MOIP02P would be a better choice, even cheaper.

[@MicsigOscilloscope]

No offense, but our MOIP02P worth better, tell me where you are see if we can introduce you a local distributor.

[@mikeselectricstuff]

For 6K I'd expect something that looked a bit less homemade inside. The PCB layout looks pretty ugly.

[@txd]

Yeah I irked at the not orthogonal lines :(

[@bartschrodernz]

Hi, at [26:06] I see the edge induced common mode rising to a dip of about 2.5V. Quite large. Assuming 310V power supply, this is a CMRR of 20 x log (310/2.5) = 42 dB. Allow 20 dB for the 10x divider, and that is 62 dB. Still much worse than the spec. What is going on?? You can see it is a probe thing, as the dip grows with the power supply voltage, while it should stay constant if it was a problem in the power supply. The statement that nothing is changing is not correct, the other parts of the signal are much lower bandwidth.

[@Paul-sc2ue]

I don't think we can be sure it's a 'probe thing' that is being seeing at 26:06. The gate that's being looked at is itself being driven by an isolated high side driver which is also having to deal with the 310V common mode signal. These drivers are not perfect, there is for instance around 2pF across the drivers isolation barrier, this is not ideal when looking at nS rise times. I feel that what the Probe is showing us is exactly the type of signal that these fibre isolated probes will be ideal for looking at.

[@bartschrodernz]

Hi Paul, Thanks for the reply - you are completely right, the power supply can indeed have a dip in the output, not a good thing as it starves the gate of current. A simple way to check which is which is to short the probe tip to the common and see what happens then as you probe the upper switching transistor drain. But I guess you no longer have the probe. @@Paul-sc2ue

[@Paul-sc2ue]

Hi Bart Yes some more experimenting would help to clarify exactly what is being seen. I have never had one of these probes to play with, I just watched the video because I've been looking at this type of galvanic isolated probe and the video came up via google. It could be a dip in the Power supply to the gate drivers, these are actually produced via two transformer isolated DC-DC converters. in the Eval Board app note they mention they have tried to keep the barrier capacitance to a minimum by reducing the number of windings and capacitively doubling the voltage. The isolated gate drivers also use some form of internal transformers to isolate the low side from the high side drivers. this again will have some barrier capacitance. Although the designer has tried hard to minimise the capacitance there will still be some capacitance across the barrier. With 300V 1MHz square waves on one side of the barrier and the other side being effectively at earth potential the fast edges will get across the barrier and will cause a nuisance of themselves. For instance the gate driver on the high side is driving the gate of GaN device to around 5V wrt its source, as both the gate and source are riding on the 300V square wave, if the gate happens to 'see' more of the barrier capacitance than the source then this will cause the gate drive signal to change (for the worse) possibly causing the spike we see as the voltage increases. When you are designing / working with circuits like this they are never perfect because of all sorts of factors, however it looks like these isolated probes are going to be a big help looking for all the little subtle problems that arise. Paul @@bartschrodernz

[@EdwinSteiner]

Wow, the external battery pack you can get for this probe costs almost twice as much as my oscilloscope. Don't ask about the actual probe. The levels of gear-envy this channel induces are inhumane.

[@hinz1]

But what use do I get from such a probe?? How often do I work on 50kV and need 200MHz scope?? Much better buying an old excavator, might actually get used for something ;-D And even if I would need it somehow, cheap battery scope from china floated @ 50kV would be simpler/cheaper solution.

[@EdwinSteiner]

@@hinz1 The probe is a specialty, yes. I'm more envious about the scopes, etc. However, I did recently have a use case: debugging brightness fluctuations in an old analog scope. I needed to look at small fluctuations between voltages that were about 2kV below ground in common mode. I could do it with a self-made optocoupler circuit, though, that is an ultra-cheap version of what this probe does much better.

[@hinz1]

@@EdwinSteiner Nice new gear certainly is nice, but it's a big cash grab too. Old, cheap ebay Tek scopes and HP gear from 1960-80s you could almost always repair yourself for little cash, have fun and learn something. Highend stuff 2000s onwards, spend big cash buying it and maybe 50% repair chance, if something simple breaks. Otherwise dump more cash to manufacturer, to get it repaired, hopefully. Calibration and everything, digital, nothing you can simply fiddle yourself with screwdriver. And for the not so sophisticated stuff I sometimes do, a good Tek7000 (usually repaired many times over) and a bunch of plugins is all scope I need, perhaps some cheap Rigol or HP logic analyzer for digital stuff. I.E. I'm not that envious to keep all that stuff running and calibrated, let alone for the dumpload of cash that went into acquisition...

[@vincei4252]

Uses AA batteries so should last a long time. Not really, if Duracell and other battery manufacturers have their way with the device.