EEVblog #932 - How Does A HV Differential Probe Work?

[EEVblog]

How does a high voltage differential probe work?
How is it safe?
Is it isolated?
A teardown and some reverse engineering of the Lecroy AP031 25MHz HV Differential probe. a.k.a Sapphire Instruments SI-9001

[Cliff Matthews]

Very nice Dave! (and DIY maybe it's cheaper than a big isolation transformer too).

I'd like to perfboard a similar front-end with gaps and conformal coat and tie it to a simple back-end for non-critical use under 2 MHz and no more than 400v. If I understand this correctly, the input resistor dividers are about 150:1 at DC? And the original design puts no more that 2.5v p-p through to the back-end?

Sorry for marking-up the D-CAD, but is there any low-cost DIY project hints you might offer? (safety first of course..)

[e100]

Why does something like this retail for hundreds of dollars?
Is there a 20x mark up?

Mike

[EEVblog]

Why does something like this retail for hundreds of dollars?
Is there a 20x mark up?

Someone has to design it and make a profit, and they don't sell in huge volumes.
Anyone is free to compete.

[nwvlab]

Your teardowns are very nice, also because many people can't afford to spend hundreds of dollars each time they need something (a probe, a precision resistors box, etc).

I was impressed that actually those probe were not isolated. I was thinking they were using the HCNR200/201 (link to datasheet: http://www.avagotech.com/docs/AV02-0886EN ).

cheers

[selim13]

Thank you for the teardown! Really nice to see what is inside of those probes.

Also while searching for the schematics  I found a thread on diyaudio where this particular model was reversed engineered: http://www.diyaudio.com/forums/equipment-tools/248505-differential-probe-reverese-engineered.html

[McBryce]

Hi Dave,
       why is the PCB digitised/scrambled in the Video link? Did I miss something?

Regarding the more recent low-cost Diff-Probes, such as those from Pintek. Several users have posted teardown pictures here on the forum including myself: https://www.eevblog.com/forum/testgear/pintek-dp-100-100mhz-differential-probe-teardown-mini-review/

McBryce.

[EEVblog]

       why is the PCB digitised/scrambled in the Video link? Did I miss something?

It's a spoiler

[McBryce]

       why is the PCB digitised/scrambled in the Video link? Did I miss something?

It's a spoiler

Yeah, like a small picture of the PCB is going to give away the entire content of the video!?

McBryce.

[Jeff_Birt]

I was puzzling over the heat shrink bands around the HV input caps ands resistors. Most heat shrink only has a 300V rating and it was not covering the entire component so it was not insulation. It much be a blast shield, i.e. if you blow up the HV front end the pieces of resistors and caps won't go flying.

[Jay_Diddy_B]

Hi,

There are some measurements that I made on one of these probes in this thread:

https://www.eevblog.com/forum/projects/high-voltage-differential-probe-design-for-review/msg600694/#msg600694


Regards,

Jay_Diddy_B

[C]

Why not use one of the many fully differential op amps

[Kleinstein]

The discrete JFET + BJT stage is still quite quite good when it comes to noise and bandwidth with a high impedance input. The discrete part is like a high impedance, low noise differential amplifier with a differential (current) output.  So it would need something like a JFET input instrumentation amplifier to replace is.  For me the strange thing is how much effort for just making the current source.

[nctnico]

I was puzzling over the heat shrink bands around the HV input caps ands resistors. Most heat shrink only has a 300V rating and it was not covering the entire component so it was not insulation. It much be a blast shield, i.e. if you blow up the HV front end the pieces of resistors and caps won't go flying.

My guess is that it is there to get a minimum creepage distance which is good to get into the low kV range of isolation and still be able to have it assembled quickly and consistantly.

[rob77]

The discrete JFET + BJT stage is still quite quite good when it comes to noise and bandwidth with a high impedance input. The discrete part is like a high impedance, low noise differential amplifier with a differential (current) output.  So it would need something like a JFET input instrumentation amplifier to replace is.  For me the strange thing is how much effort for just making the current source.

i think it's a pretty simple current source + current mirror there.

[C]

found this from electronicdesign.com 2003 a while back.
any thoughts?



http://electronicdesign.com/analog/increase-common-mode-range-fully-differential-amplifiers

uses THS4130  one each $8 to $9 from mouser
http://www.ti.com/lit/ds/slos318i/slos318i.pdf

[MauriceS]

There is fully reverse engineering info here:
Someone reverse engineered them:
http://www.diyaudio.com/forums/equipment-tools/248505-differential-probe-reverese-engineered.html

http://bardagjy.com/?p=1664?

There are SMD and through hole versions...

[TrioTest]

Thanks Dave for choosing this topic. Safety for both the operator and the test-equipment is extremely important and differential probes are designed to ensure safety as well as enabling measurements where the signal ground is floating from earth or on balanced circuits.

Using an isolation transformer is filled with danger on normal oscilloscopes unless the operator is very careful and fully understands the measurement environment. Personally I'd never recommend it. If you float the scope from ground then the exposed metal BNC's on the unused channel(s) and EXT-Trigger input all rise to whatever voltage you've connected to the ground lead on the passive scope-probe. Plus you can't work on more than one signal if the other signals' grounds are at different absolute potentials relative to earth. If you do, the ground plane in the scope's front-end will have to carry the full short-circuit current caused by the potential difference in the two external signal grounds. I've seen the result and it was catastrophic.  It was on a scope that was used on delta configured 3-phase. The operator used 100:1 passive probes for "safety." ...  There's no attenuation in the ground wire of a scope probe as this company found to its cost.  I also have actually seen in my travels a large company servicing UPS units using an oscilloscope powered by a mains lead with the Earth pin removed.  Even worse, this power cord was actually carrying a recent test tag to say it was safe!!

Another point in favour of differential probes is that buying a couple of differential probes and a normal oscilloscope is VERY MUCH cheaper than paying for an isolated channel oscilloscope. 

FYI... especially for Aussie's on the forum....     Sapphire, as Dave said, is the OEM for the probe reviewed. TRIO Test & Measurement is the Sapphire distributor for Australia.  We keep the SI-9001 (700V) and  the beefier SI-9002 (1400V) and a range of other Sapphire differential probes in stock. Sapphire is the OEM for more big-name scope companies than just the LeCroy branded version reviewed.  We have seen them branded with Yokogawa and Agilent too.

The other lower cost probe Dave pointed out at the end of the teardown is the Pintek DP-25. It is a good value product.  The reason we carry both brands (Sapphire & Pintek) is to offer a choice in price/performance. The main difference is that Sapphire costs a little more but it also has a higher voltage CAT rating at 1000V CAT-III or 2500V CAT-II.   Pintek has a lower  CAT- rating at 600V CAT-II and a lower price to match. Pintek is also the OEM for some of the big-guys too. Sapphire is a more sturdy construction and lower noise.

Be careful with blaming all the noise on the probe though. If you are using the probes on lower level signals and with lower probe attenuation settings, those unshielded input leads do make good antennae in the presence of interference sources and the probe's high input impedance is great for turning those small induced currents into significant noise voltages. It's perhaps also a good idea when using this type of probe having 25 MHz bandwidth, to use the 20 MHz bandwidth-limiter found on most scopes unless you really need the full 25 MHz of the probe.

If any one wants more info on these probes then check-out our website. Our pricing for the Pintek probe Dave showed (at 23:55 in the teardown) is similar to the eBay price shown and mentioned by Dave. 

Pintek DP-25  http://www.triotest.com.au/shop/search?controller=search&orderby=position&orderway=desc&search_query=pintek+dp-25&submit_search=Search

Sapphire SI-900x  http://www.triotest.com.au/shop/search?controller=search&orderby=position&orderway=desc&search_query=sapphire+si-900&submit_search=Search 

For the upper end of the Sapphire differential probe range check out this one:
http://www.triotest.com.au/shop/oscilloscope-differential-probes/1539-sapphire-instruments-si-9010a-differential-probe-70mhz-7kv.html
Up to 70MHz and up to 7000V differential... but not at the same time   

[Kleinstein]

The simple circuit with TSH 4130 is a poor choice, like a beginners try. The voltage noise of that amplifier is really good, but current noise is way to high for the high impedance. Also common mode gain can not be adjusted (you need to, due to layout effects).

[rx8pilot]

Why does something like this retail for hundreds of dollars?
Is there a 20x mark up?

Mike

it may be 20x divided by a whole bunch of entities....

It goes from OEM to LeCroy. LeCroy sells to a distributor. The distributor sells to a dealer. The dealer sells to you. Every step adds significant cost. Very inefficient but that is the way commercial products still move, especially the ones that are not considered particularly price sensitive. The target customer expenses these things and does not think twice about the cost.

I wish that all distributors and dealers would go away and let me buy from manufacturers direct. Everything would literally be 50% less overnight and the manufacturer would have the exact same margin.

[PA0PBZ]

I wish that all distributors and dealers would go away and let me buy from manufacturers direct. Everything would literally be 50% less overnight and the manufacturer would have the exact same margin.

And the unemployment rate would double and the support would suck. The structure is not there just because someone thought it was a good idea, but it is there because it works...

[rx8pilot]

And the unemployment rate would double and the support would suck. The structure is not there just because someone thought it was a good idea, but it is there because it works...

 
It worked pre-internet, the model is outdated in my opinion. As I have been a buyer and seller in high-end commercial electronics - the distributors do nothing to support me - they just facilitate an intermediate transaction that is primarily to get dealers signed up. The dealers rarely have more than basic sales data that is already on the data sheets and other material generated by the manufacturer. After the sale - all my customers come direct to me because the dealers cannot help them as well as the people that designed the product. Even the good dealers can only deal with minor issues.

Manufacturers spend the bulk of the money on marketing, trade shows, support material like white papers, data sheets, videos, etc. I don't think it works at all for test equipment. It does work for components and that is why I use Digikey instead of going to 50 different vendors each week with separate PO's.

[rrinker]

 Pretty much the same in the computer industry. What were dozens of distributors have merged together and there are now only a handful. But unlike the distributors of yore, they don't act as distribution warehouses, they are mostly just there to process orders and add a markup - the equipment is in most cases is now drop shipped direct from the manufacturer. Our primary software distributor (now THERE'S a real useless entity - they software is downloaded from the manufacturer's web site and they email you the license keys - the distributor literally does nothing. And our main software distributor is now actually a competitor - they added support services to their offerings, although the quality is pretty poor compared to our people and in several cases they have subcontracted with us to do the actual support work.

 Tits on a bull.

[MartinX]

I suspect the 14pin DIP IC is the good old LM733, high bandwidth and easy to set the gain, it would match the age of the construction that seems to be from the eighties.

[Kleinstein]

The LM733 or NE592 (essentially the same) was also my first thought, but it is not. There are links to detailed reverse engineering : it is a transistor array, to make a current source. The +-9 V supply would also be not compatible to the LM733.