> 1 GHz DIY differential probes

[JohnG]

I'm wondering if anyone has a working design or knows of one for a DIY wideband (DC to > 1 GHz) DIY differential probe. I've search for this on and off, and found some attempts, but have not found a successful design. I have come across references to a 2 GHz design published in Elektor July/Aug 2015, but I don't have the article so I don't know if this is a working design or not. I'd buy the issue if I had some indication that this was indeed a working design.

I've used these types of probes before, so I understand the voltage and impedance limitations they present.

There are a few reasons I'd like to have a DIY design. First, to learn from. Second, to have a design that is not tied to a particular brand scope  . Third, to extend or modify for different applications. In theory, one could have a few different probes for different voltage ranges, or similar.

Any help appreciated!

John

[joeqsmith]

I attempted to design a DC coupled wide band differential probe last year.  Got as far as coming up with requirements.  In the end I could not find parts suitable for the job.   

I had seen a few homemade probes people had came up with but nothing DC coupled and nothing as fast as I was looking to build.     

Beyond the DC-1GHz diff, what are your requirements? 

https://xellers.wordpress.com/electronics/1ghz-active-differential-probe/
https://xellers.wordpress.com/2014/09/28/diy-active-differential-probe-characterization-round-2/

[JohnG]

Since you asked about requirements:

Bandwidth: > 1GHZ (risetime < ~300 ps) or so, faster is better
Input: >= 1k resistive, <= 2pF capacitive (higher resistance would be preferred
Attentuation: 10:1
INput voltage range: >10 V DM, >5V CM
CMRR: >20 db at 1 GHz, if possible

If I could do a 100:1 with a variation on the same circuit, that would be great. I could take some hit on speed at the higher voltage, but the input resistance would have to go up to at least 10k. It would not need to meet 100V continuous operation, but the ability to get capture short pulses (up to 1-2us wide) at 100V or close would be a big plus. Not many commercial probes can do this at this kind of BW.

Here is what I really want, but it will only work with a Tek Scope, which I don't have. It will probably cost as much as the scope, or close: http://www.tek.com/dl/51W-60485-0%2520IsoVu%2520White%2520Paper%2520%2520TN%25203-25-16.pdf


All the CMRR you could ever want. I have seen it in action, and it works. But, I won't be getting one anytime soon .

John

[Marco]

Here is what I really want

Not really. That's a high frequency 50 Ohm input isolation amplifier, not a high input impedance differential probe. You can simply add two resistors for a differential version of the passive attenuated probe, but just as with the passive attenuated probe that's not optimal as far as noise is concerned.

The Elektor design uses a high frequency fully differential amplifier (ADA4927-1) as a differential to single ended converter, witch matched resistors to provide attenuation. Boring Also they didn't actually measure CMRR, you can't rely on datasheet CMRR because that's for differential->differential and common mode errors for the two outputs could very well cancel out in differential mode.

[joeqsmith]

When I saw 1GHz to DC I was thinking digital, not something high voltage dual purpose.  I was hoping to keep loading above a couple hundred ohms with  maybe +/-2.5V differential and same for common mode.  Maybe 0.2pf at the tip sort of thing, couple of GHz BW.  I was thinking that with modern parts I could pull it off but I spent a few weeks looking and gave up.  I was also having problems trying to figure out how I was going to tackle the DC part.   My plan was to use 2 AC paths and two DC to something paths.  Then somehow combine the whole mess and not ruin signal.  The circuitry would have been fairly complex, at least for me.   

Keep us posted if you try it.   

On the bright side, at least for digital work the used probe market seems to be growing.  Prices are all over the place.   Thousands for broken probes with missing parts, to under a K for this baby!

http://www.ebay.com/itm/Teledyne-LeCroy-D13000PS-13GHz-Differential-Probe-System-/131895698595?hash=item1eb598a4a3:g:4WEAAOSw~otWce6k

I could see this for my hobby use.
http://www.ebay.com/itm/LeCroy-D600-7-5-GHz-WaveLink-Probe/162154103823?_trksid=p2047675.c100005.m1851&_trkparms=aid%3D222007%26algo%3DSIC.MBE%26ao%3D1%26asc%3D20131003132420%26meid%3D1b91768d4ae64145a92f0a3a455d4605%26pid%3D100005%26rk%3D2%26rkt%3D6%26sd%3D311133558055

 
That Tektronix probe looks slick.  Having a 2KV CM standoff or galvanic isolation is not something I would need at home.  I did find another article on it.  Sounds like it may not yet be available. 

http://electronicdesign.com/blog/1-ghz-isovu-leads-tektronix-probe-performanceusability-charge-apec

And a video on their facebook page
https://www.facebook.com/tektronix/videos/vb.50990632804/10153395327767805/?type=2&theater

[Marco]

Teledyne-LeCroy-D13000PS-13GHz-Differential-Probe-System

How do these actually work? From the pics and the impedance specs I would guess each signal is probed by a separate passive divider probe (with 500 Ohm tip resistance). Because it seems to have two separate transmission lines coming out.

[joeqsmith]

Teledyne-LeCroy-D13000PS-13GHz-Differential-Probe-System

How do these actually work? From the pics and the impedance specs I would guess each signal is probed by a separate passive divider probe (with 500 Ohm tip resistance). Because it seems to have two separate transmission lines coming out.

I have not idea but am guessing the changeable interconnect to a custom ASIC to the scope.      Spent some time to see if they had a whitepaper or some other notes.  May be of interest:

http://cdn.teledynelecroy.com/files/whitepapers/wp_differential_measurements.pdf
http://cdn.teledynelecroy.com/files/whitepapers/probing_technology.pdf
http://cdn.teledynelecroy.com/files/whitepapers/technologies-for-very-high-bandwidth-real-time-oscilloscopes.pdf
http://cdn.teledynelecroy.com/files/manuals/d13000ps-om-e.pdf

[Marco]

That's the Lecroy, the IsoVu was at 80 dB at 1 GHz. As I said though, it's not a differential probe, it's a 50 Ohm input isolation amplifier.

[tggzzz]

A novel very simple and robust differential probe: http://emcesd.com/pdf/cd94scr.pdf

Not DC, the article measures refers to 500MHz.

[Marco]

Simple, but not cheap. It should be possible to make a combiner out of coax and ferrite beads.

[nctnico]

Does it need to be DIY from scratch? Tektronix P6860 logic analyser probes can be found cheap (say $50) and they have a differential amplifier for the clock / qualifier inputs and regular FET probe inputs for the other signals. All in all one probe gives you 2 differential and 16 regular fet probes with >2GHz (probably 3Ghz) bandwidth.

[lukier]

Does it need to be DIY from scratch? Tektronix P6860 logic analyser probes can be found cheap (say $50) and they have a differential amplifier for the clock / qualifier inputs and regular FET probe inputs for the other signals. All in all one probe gives you 2 differential and 16 regular fet probes with >2GHz (probably 3Ghz) bandwidth.

Interesting. Tektronix doesn't share much detail on the inner workings of their probes (but some companies seem to know more - e.g. http://www.movingpixel.com/SQUIRE.html). How do you power this probe (without TLA card/mainframe I assume in this thread) and terminate it for the scope (50 Ohm). Did you build an adapter from the probe high density connector (the side that goes to the TLA7AA4) to BNCs?

[Cerebus]

Please ignore this, just marking thus to come back to.

[Someone]

I'm wondering if anyone has a working design or knows of one for a DIY wideband (DC to > 1 GHz) DIY differential probe. I've search for this on and off, and found some attempts, but have not found a successful design. I have come across references to a 2 GHz design published in Elektor July/Aug 2015, but I don't have the article so I don't know if this is a working design or not. I'd buy the issue if I had some indication that this was indeed a working design.

I've used these types of probes before, so I understand the voltage and impedance limitations they present.

There are a few reasons I'd like to have a DIY design. First, to learn from. Second, to have a design that is not tied to a particular brand scope  . Third, to extend or modify for different applications. In theory, one could have a few different probes for different voltage ranges, or similar.

You'll have challenges simply driving the cable and scope input capacitance at 1GHz, you could start with a simple implementation of the THS3217 to get a feel for the project but it won't hit 1GHz. Cheap probes without manufacturer dependence are possible: https://www.eevblog.com/forum/projects/diy-100mhz-differential-probe/

[joeqsmith]

A novel very simple and robust differential probe: http://emcesd.com/pdf/cd94scr.pdf

Not DC, the article measures refers to 500MHz.

Interesting idea.  Looks like the Mini-circuits 1-500MHz w/SMA cost about $65.  Isolation is worse on their surface mount parts.  I wonder why he did not have a resistive probe in his comparison.

[nctnico]

Does it need to be DIY from scratch? Tektronix P6860 logic analyser probes can be found cheap (say $50) and they have a differential amplifier for the clock / qualifier inputs and regular FET probe inputs for the other signals. All in all one probe gives you 2 differential and 16 regular fet probes with >2GHz (probably 3Ghz) bandwidth.

Interesting. Tektronix doesn't share much detail on the inner workings of their probes (but some companies seem to know more - e.g. http://www.movingpixel.com/SQUIRE.html). How do you power this probe (without TLA card/mainframe I assume in this thread) and terminate it for the scope (50 Ohm). Did you build an adapter from the probe high density connector (the side that goes to the TLA7AA4) to BNCs?

I have not looked into the technical details. I'm just pointing towards an existing solution which could be hacked.

[JohnG]

Thanks for the feedback, everyone.

1 GHz is a hard requirement, and is probably barely enough for my application. I've seen the link to the design with the transformer, and it looks intriguing, but the bandwidth is not there and I would really like the ability to go to DC.

I've looked at buying used probes and hacking at them, but this is for work, and my priority is not to design a probe, but get measurements. At some point it becomes cheaper to buy, just not there yet. I would like the flexibility of having a design - recently I have built transmission line probes into some designs and they work surprisingly well.

The IsoVue may not look like a differential probe, but it sure acts like one. Yes, it looks like a 50 ohm isolation amplifier, but it is a high resolution analog one, and they are not exaggerating the CMRR. Because the CM impedance is so high, there is very little conversion of CM to DM. I'm not sure about the innards, except that I know they use an electro-optic modulator. These can be designed to be inherently balanced, so it may be that the only imbalanced part is the input coax/attenuator. So unless you are sure about the design, it may be premature to say it is not a differential probe. Here is a link with some interesting information: https://www.google.com/patents/US7310455


John

[tggzzz]

I've looked at buying used probes and hacking at them, but this is for work, and my priority is not to design a probe, but get measurements. At some point it becomes cheaper to buy, just not there yet.

Have you considered renting a probe? Last time I looked at renting, a long time ago, the monthly rental charge was 10% the purchase price.

[Marco]

So unless you are sure about the design, it may be premature to say it is not a differential probe.

I thought that because they used MMCX connectors all the input impedances of the probes were 50 Ohm, but apparently they just ignore the mismatch (not big deal at 1 GHz) and the attenuated probes have higher input impedance. They seem to be simply using simple passive dividers for attenuation, with no termination at the probe end.

So yeah, it's a high impedance differential probe ... but a noisy one, because of the attenuation. 25x for 1.25k input impedance. I guess they want to sell an active probe separately in the near future, undoubtedly for a ton of money.

[joeqsmith]

Rental may not be a bad way to go but I too have not done this in many years.   I wonder if the cost of a used one would be close to the rental fees.   Something to consider if its a one off measurement.     

I looked around to see if I could find a better 180 degree combiner but did not find anything.  May have been fun to try and replicate his tests. 

[David Hess]

Source impedance is a major problem with wide bandwidth differential probes and a major limitation of performance.  If you are probing a differential transmission line, then it may be better to build a test point into the circuit with 50 ohm outputs to connect directly to the oscilloscope which can then do the differential measurement itself.

[joeqsmith]

Source impedance is a major problem with wide bandwidth differential probes and a major limitation of performance.  If you are probing a differential transmission line, then it may be better to build a test point into the circuit with 50 ohm outputs to connect directly to the oscilloscope which can then do the differential measurement itself.

  I guess that depends on what you consider a major problem.   My old DC coupled diff probes are 300 ohms at 4GHz.  Tip capacitance is a 1/10th of a pf. 

[David Hess]

I guess that depends on what you consider a major problem.   My old DC coupled diff probes are 300 ohms at 4GHz.  Tip capacitance is a 1/10th of a pf.

How much source impedance mismatch are you dealing with though?  A 300 ohm differential probe is not intended for general purpose measurements; it requires closely matched source impedances to maintain common mode rejection ratio.

This Tektronix article, Making Single-ended Measurements with a Differential Probe, refers to the problem on page 11:

A 50 ohm source impedance mismatch for a single-ended signal measurement has only a minor effect on the DC CMRR of a probe with a 50 Kohm probe input impedance.  As the signal frequency increases however, the probe input impedance begins to decrease and eventually this 50 ohm source impedance mismatch for a single-ended measurement can become quite significant. High frequency AC common mode voltage transients that may be largely rejected by the relatively high CMRR in a differential measurement may show a noticeable effect due to degraded CMRR in a single-ended measurement.

[tronde]

I'm wondering if anyone has a working design or knows of one for a DIY wideband (DC to > 1 GHz) DIY differential probe. I've search for this on and off, and found some attempts, but have not found a successful design. I have come across references to a 2 GHz design published in Elektor July/Aug 2015, but I don't have the article so I don't know if this is a working design or not. I'd buy the issue if I had some indication that this was indeed a working design.

I don't want to break Elektor's copyright, but this is the specs:

Technical specification
• Attenuation: 10:1 with a differential signal and 50 ? termination in the ‘scope
• Differential input resistance: 5 k?
• Single-ended input resistance: 2.5 k?
• Output resistance: 50 ?
• Bandwidth: 1.9 GHz (–3 dB)
• Rise/fall time: 300 ps
• Power supply: ±8 to 12 V DC.

They also write:

For readers who are interested the author
offers ready-to-use and tested PCB modules, also a kit consisting of case, RF cable
with BNC connector and power supply
lead with plug. Further information from:
alfred_rosenkraenzer@gmx.de.

[joeqsmith]

How much source impedance mismatch are you dealing with though?  A 300 ohm differential probe is not intended for general purpose measurements; it requires closely matched source impedances to maintain common mode rejection ratio.

Differential digital, LVDS etc.