Close
0%
0%

A 10X 100MHz Differential Probe

A DIY oscilloscope probe to dig small differential signals embedded in high common mode voltages.

Similar projects worth following
To make...or not to make...that is the question. I needed a probe that can dig out a 1V differential signal riding on a 60Hz 125VAC mains voltage. Many of the commercial probes aren't suitable for my purposes and are pricey. Therefore, the answer is...make.

Twice in the last couple of years I have needed a probe to diagnose problems with a non-isolated power supply that connected directly with the 125VAC mains voltage here in the USA. I have been able to work around it by using less than satisfying methods. My cheap oscilloscope (a Rigol DS1102E) is good enough for most of my purposes, but the math function to subtract Channel 1 from Channel 2 really sucks. And even though it is a cheap scope, I'd rather not have to buy a new (more expensive) scope because I connected the scope ground incorrectly.

Needing a tool on a yearly basis hardly qualifies as necessary, but I had some down time waiting for components and PCBs for another project and began to consider this as an interesting project.

There are now three versions of this probe. My original version that is powered by a 5VDC adapter, Paul's new version is powered by any adapter in the range of 7-12VDC, and a 1X version -- see Christoph's Project. Paul made some changes to increase the creep distance between the input pins since he needs to use it with 240VAC mains. You can find the Gerber files, BOM and schematics for Paul's version on his Github Page. Christoph's 1X probe uses my design, but changes the attenuator to achieve higher differential gain at the expense of lower common mode range.

I'd been following Paul Versteeg's progress to make a differential probe on his blog. I recommend that you read it -- he covers most of the basics and has some good references of other approaches to make this kind of probe. Paul has done more thorough testing as well (his equipment is better than mine.) He also spent a lot more effort explaining how to assemble the probe and calibrate it. There are also several persons within Hackaday that have completed projects successfully, or are still ongoing.

The Specifications:

  • Input impedance: 20MegΩ// 1.25pF - differential, 10MegΩ//2.5pF each terminal to GND.
  • Differential Gain = 1/10 V/V. Any lower than this and the DS1102E could not resolve a 1V signal with any clarity. (Many of the cheaper differential probes are switchable between 50X and 500X attenuation. This would be good if the signals in which you are interested are very large, but that's not the case so far with my needs. The better probes ($400 and up) provide 10X/100X attenuation.
  • Common Mode Range > ±340V (240VAC produces a sine wave 679V peak-peak)
  • CMRR >90dB @ DC, ~60dB @ 1MHz
  • Differential Voltage Range > ±24V for 240VAC common-mode,  ±24V for 0V common-mode (Paul's version increases both ranges to ±25V)
  • Bandwidth ≥ 100MHz (Dependent upon signal amplitude) This will be a stretch for my current skill set.
  • DC offset < 20mV
  • Noise: 2.2mVrms at output.
  • Cost: ~ $50

The completed probe:

The above photo is my completed probe. You will need a 5VDC adapter to power Bud's probe or a 7-12VDC adapter to power Paul's probe, and a SMA to BNC pigtail to connect the probe to the scope. You should also consider making a few leads for various situations. The leads use a Dupont 0.1" (2.54mm) 2-pin female connector to attach to the probe. The green tie-wrap provides a bit of strain relief for the power supply leads into the probe.

The Schematics:

There are now two versions of this probe.

The 5V adapter version (Bud's):

The 7-12V adapter version (Paul's):

[Edit 2020-01-30: Improved the differential range with a 240VAC input to ±24V with small changes to component values. No change to the PCB layout.] 

[Edit 2020-02-05: More conservative design for the first gain stage stability. Replaced BAV99 diodes with BAV199.]

[Edit 2020-02-13: Increased R23 and R24 to 1.5k from 510Ω to reduce ringing with fast input step.]

[Edit 2020-02-25: Changed C6-C8 and C11-C13 to match lower capacitance of the 10pF input capacitors. Your mileage may vary.]

[Edit 2020-05-10: Changed C19 to 0.4pF±0.05pF to match Paul's schematic. He found the bandwidth not quite 100MHz so reducing this...

Read more »

BOM.xls

Bill of Materials spreadsheet for 240VAC diff probe. Includes comments and sources for components. Corrected an error for C7/C12 on 2020-09-29

ms-excel - 20.00 kB - 09/29/2020 at 23:37

Download

DiffProbe.zip

Gerber files for the PCB. Can upload this zip file to OSH Park to have PCBs made. Board thickness should be 1.6mm.

Zip Archive - 31.96 kB - 02/25/2020 at 15:45

Download

DiffProbe.asc

Basic LTspice simulation schematic for the differential probe. This is the 240VAC version

plain - 8.42 kB - 01/31/2020 at 16:24

Download

  • Noise Analysis

    Bud Bennett03/06/2024 at 22:50 0 comments

    I had a PM from someone who built the probe and then wondered why the output noise was so bad. They were measuring about 100mV p-p at the output with the inputs shorted together. At first I had a difficulty believing the noise was that bad, but I got the same result when performing the same measurement. In fact one of my probes was a bit worse than 100mV p-p. So why is it so bad -- I don't know the answer yet, but I can confirm that LTSpice predicts the same result.

    To get an answer I ran a .Noise analysis on the entire probe (with parasitics, but that doesn't matter much). LTSpice predicts that total RMS noise voltage at the output is 2.1832mVRMS. I use a fat-finger rule to get p-p from RMS by multiplying the RMS value by 6, so peak-to-peak output noise is about 13.1mV p-p at the scope input, if it is measuring with 1X. If you select the 10X probe setting then the noise will jump to 131mV p-p, which is what I'm seeing. 

    This doesn't bother me too much because most of what I measure with these probes is repetitive in nature and I can average the noise to a much lower value. But you might have a problem if you need to dig a small, non-repetitive signal out of this noise.

  • Some measurements

    Christoph02/18/2022 at 21:31 0 comments

    A fellow maker put the diff probe under test, with some serious equipment: https://electronicprojectsforfun.wordpress.com/a-100mhz-differential-probe-from-hackaday-with-keysight-autoprobe-connectivity/

    We are now working on a slight change to the offset current injection that won't impact CMRR as much, and an enclosure.

    The design that was posted previously (with +/- 2V5 dual supply) requires voltage regulators that are pretty specific, expensive and hard to get, so we're looking into building our own from TL431 references, TLV911 OpAmps, and a pair of FETs. Here's the positive rail:

    These parts are comparatively cheap, and replacements should be easy to find. The simulation results are looking promising, and we'll just test it.

    A simple dual output DCDC converter can be used to generate the required +/- 5V from a scope's USB output - at least for one probe. Or two, if the DCDC can supply enough current.

  • Addition to the previous log: compensation

    Christoph09/05/2021 at 18:48 0 comments

    In the previous log (https://hackaday.io/project/169390-a-10x-100mhz-differential-probe/log/197059-slim-probe-build-log-and-first-findings) I ranted about having to swap caps to get the input attenuator into the trim cap's range. Here are some more details about that.

    In step 2 of the compensation procedure (see here: https://hackaday.io/project/169390-a-10x-100mhz-differential-probe/details far down) we try to match positive signal path's AC gain with DC gain. If they don't match, we get this kind of output error:

    (Ch1: single ended reference input, Ch2: differential probe)

    There's an overshoot in the probe's output, which means that the AC gain is too high.

    The probe was equipped with

    C5...C8 = 330p + 22p + 220p + (2...10)p = 574...582p.

    This was too low even with the trim cap at the upper end. The probe peaked to 2.2 V with an actual input of 2V. We can guess the right overall capacitance as

    574 * 2.2/2 = 631

    which seemed rather high to me, so I swapped the small cap to get 

    C5...C8 = 330p + 68p + 220p + (2...10)p = 620...628p.

    The result:

    Since I got too low AC response now, the 68 pF cap was too large. And, unfortunately, estimating the right cap values from one scope shot apparently isn't an exact science. 

    I ended up with

    330+220+33+10+trim = 595...603 pF

    to get this result:

    That was ok and I was also able to match both signal paths.

    Swapping these caps was very difficult because the layout doesn't have thermals for the caps' ground connection, which means that the solder joints were very hard to heat up even with low temp solder.

    One obvious solution to this would be to increase the trim range, but the temperature coefficient of the JR100 (or JR030) trim caps is lower than that of trim caps with a higher range. We'd be sacrificing performance in the application for convenience in the assembly and compensation, and that doesn't sound like a good compromise for a device that is only built in singles (and caps are cheap).

  • "Slim" probe build log and first findings

    Christoph08/24/2021 at 20:23 1 comment

    In the previous project log (https://hackaday.io/project/169390/log/194914-adding-a-more-sophisticated-scope-interface) I described a modified probe that allows to inject an offset current to adjust the probe. Some scopes have an interface that has additional connections to supply active probes, communicate with them and adjust them, and we're trying to make this probe compatible with that interface - at least the analog and supply part (communication will be added too, but closer to the scope connector).

    While Bud was already pointing out some shortcomings of the modified schematic, namely gain error and lower CMRR, I had already ordered PCBs from aisler and assembled one once they arrived:

    My plan was to assemble one even if it's not perfect, send it to another tinkerer who has the necessary equipment to test it thoroughly, and use his and Bud's feedback to adjust the next revision of the schematic, the layout and also the overall shape (for a 3D printed enclosure).

    Note:

    1) I didn't place protection diodes

    2) The layout didn't contain spark gaps and their ground trace

    3) an extra cap had to be added to the attenuator to compensate the probe. We'll get to that later.

    Assembly was done with low temp solder (see here for the details about how I solder small/low temp stuff: https://hackaday.io/page/10792). However, swapping out caps was still tedious and time consuming because it's an iterative process.

    Caps, Caps, Caps

    I had previously built a probe as per Bud's original design and knew that the effective capacitance of C1...C4 is significantly lower than 2.5 pF (4x 10 pF caps in series). On his layout, the effective capacitance was around 2.31 pF (4x 9.25 pF in series), so C6...C8 had to be around 572 pF (+trim cap 2...10 pF) to get a 1:250 attenuator.

    On my layout, the capacitance of C6...C8 had to be 593 pF (330 + 220 + 33 + 10 pF), which is significantly higher than on Bud's layout. We came to the conclusion that it's most probably the absence of the spark gaps and their ground trace that leads to this significant change in parasitic capacitance (obviously, this is something that can be found out by comparing with a layout with a ground trace). The trim caps were at 1/3 and 2/3 of their range. I created a spreadsheet to calculate a probable range of the capacitance of C1...C4, taking into account the tolerance of C6...C8:

    At the bottom, the spreadsheet calculates the resulting attenuation given the values for C1...C4 and C6...C8 (+trim). Since both trimmers were somewhere around their center, I wanted the resulting error at both ends of the trim range to be somewhat symmetric around zero (highlighted in green).

    The left block shows the result for the case that C6...C8 were at their lower end (-1 %). To get into the right trim range, C1...C4 had to be set to 9.52 pF (top left). Similar for the center and right block.

    Next step was to select a set of capacitor values that allow to build a probe where

    • the "base" capacitance is known to be not too large for the lower end of C1...C4 and +1 % in C6...C8 (that's the one extreme of the overall spectrum)
    • a maximum of 2 additional caps with known values can be added to get into the right trim range even for the upper end of C1...C4 and -1 % in C6...C8 (the other extreme) and all other combinations
    • so no cap would have to be swapped - we'd just add more and know that it'll be good.

    Some experimentation with the values made it obvious that these goals can't be achieved with just 3 footprints, so I added a fourth footprint to the next schematic and layout revision.

    Good cap combo using values from the E12 series

    The base values are 560 + 18 pF, and one or two 12 pF caps can be added.

    For the low value of 9.52 pF for C1...C4:

    For the center value (9.62 pF):

    for the upper value (9.72 pF):

    So during the compensation procedure I could start with just 560+18 pF, and add one or two 12 pF caps until the trim caps can do their job...

    Read more »

  • Adding a more sophisticated scope interface

    Christoph07/02/2021 at 18:20 2 comments

    The probe as it was designed by Bud and Paul works well and does the job. However, as always, there are ways to make it do more: Wolfgang was looking for a modified version that can connect to his keysight scope and make use of the scope's AutoProbe interface. Keysight scopes have an extra interface (a row of pogo pads below the BNC connector) that provides an I2C interface and supplies various voltages and an adjustable offset current to the probe. This can be used to supply the probe, auto-detect it, and set the offset. That's all pretty neat, but requires a couple of modifications on the probe and an external adapter that connects to the scope. The interface is described here: 

    https://electronicprojectsforfun.wordpress.com/using-the-keysight-autoprobe-interface-in-your-own-projects/

    I found this quite interesting and started a design that can be used with or with the AutoProbe interface. I also wanted a more "pen"-shaped probe and Wolfgang suggested adding a white LED to the tip. Other modifications (starting off with Bud's 5V design):

    • The spark gaps had to go, because they add extra input capacitance
    • wider gap between the tip pads
    • not a layout modification, but the clamping diodes D1 and D2 also add some capacitance that is not desired. They are not populated for initial performance measurements.

    The modified probe needs a different external power supply if used without the AutoProbe interface, but that's not too hard to accomplish. It can now even be a simple 2S LiIon or LiPo battery, or a DCDC converter with +/- 5V output (those black blocky things)

    Power supply

    Power supply comes either as +/-12 V or as +/-3...6 V from the AutoProbe interface (some scopes apparently have +/-5 V only). This requires a different power supply on the probe. Instead of creating a virtual ground from a floating 5V suppky, we now have to create +/- 2V5 from the fixed dual supply with fixed ground. So I replaced the OpAmp that creates virtual ground by two low noise linear regulstor (TPS73125 and TPS72325).

    This part of the schematic also shows that the probe is connected to the AutoProbe adapter with a 4-pin connector.

    Offset injection

    The AutoProbe interface can inject an adjustable offset current limited to +/- 1 mA and a low voltage (5 V? not sure). This is sent through a 10R resistor at the bottom of R10 (instead of wiring R10 to ground). A 100 nF cap and a pair of tantalum caps were also added:

    This allows the offset to be adjusted by +/- 1 mA * 10R = +/- 10 mV. That's just our first guess, and the offset resistor is large enough for comfortable rework (1206). It can also be left out and placed externally for experimentation.

    Layout

    Here's the new slim layout:

    (84.5 mm x 13 mm)

    The input header is a 3-pin header with the center pin removed, I was too lazy to modify the 3D model accordingly. Same for the 4-pin connector which can be a 5-pin type with the center pin removed. In this case I substituted for 2 2-pin 3D models because the 5-pin model would have obstructed the view of the output resistor that feeds the waveguide.

    LED

    The actual LED is not on the PCB, because that would have interfered with the attenuator layout. It's connected to a pair of through-hole pads (labeled "A" and "K") right next to the 4-pin connector instead, so it can be connected from either side of the PCB, depending on what would better fit the enclosure. There's a little switch in the back to turn it on and off.

    SMA connector

    The SMA connector is a bit smaller (6.35 mm wide) than the one on the initial layout (9.5 mm). It's also further away from the output OpAmp, so we added the waveguide you can see above

    Other modifications

    R8 and R21 were rotated by 45° to make the overall layout a bit prettier. Looks like a spider now.

    The voltage regulators are on the bottom:

    There are cutouts for threaded inserts in a 3D-printed enclosure. The threaded inserts on the input end would be a couple...

    Read more »

  • Acid Test

    Bud Bennett02/26/2020 at 21:24 0 comments

    I wanted to see how well this probe would perform in its intended application. I took the case off my old Macintosh 512k and measured a few waveforms. Here's the power supply schematic:

    The primary side of this SMPS power supply is not isolated from the 125VAC mains. All of the circuitry to the left of the multi-tapped transformer are moving up and down 180V at 60Hz with respect to the ground reference on the right side. There are a lot of voltage swings that exceed the probe's ability (±25V), but you can get a pretty good idea about what is going on by measuring lower voltages. The first point that I measured was from node "5", the base of Q11, to the virtual ground at the base of Q9. It looks pretty much as it should.

    Then I put the probe leads from node "3" to the virtual ground. Again, as expected.

    The last node that I probed was the gate of Q10 w.r.t the virtual ground. This is a lot uglier, but seems appropriate.

    There is an artifact from the large common mode voltage. I recorded this to show the amount of bleed through from the common mode:

    It is the same signal at the gate of Q10, but the baseline voltage shows the 60Hz common mode. It's on the order of 5mV (50mV @ 10X), so it is tolerable for this situation.

    I'm pretty happy at this point. I wish that I had this capability when I was trying to figure out what was wrong with this circuit a few months ago.

  • Test Results Using SMA to BNC pigtail

    Bud Bennett02/25/2020 at 16:22 5 comments

    My SMA connectors that I ordered from AliExpress have not arrived, so I ordered two SMA connectors from Digikey for $3.44 each, along with the resistors to correct the gain error in the first gain stage. I am using a 20 inch long SMA to BNC pigtail to connect the probe to the scope input. The power is provided by an old wall adapter that I modified some time ago to output 5.25VDC. The extra 0.25V allows the differential inputs to swing past ±25V before the amplifiers saturate.

    Here's a few waveforms showing performance at lower frequencies:

    Channel #1 is the diff probe and channel #2 is the 10X scope probe. Both are connected to the scope's 3V calibration signal.

    This is a 25V 1kHz square wave. Same channel inputs as above.

    A zoom in on the rising edge of the 25V square wave. The diff probe captures the same waveform features as the 10X scope probe.

    Faster Rise/Fall times:

    I don't have a laboratory grade signal source to test the probe, so again I'm using a switching step-down converter to obtain ~10ns transition times.

    This is what the 10X scope probe produces on channel 2 when connected to the inductor.

    The above is what the diff probe thinks the inductor waveform looks like.

    This is the falling edge measured with the 10X scope probe.

    The falling edge measured by the diff probe.

    The rising edge measured by the 10X scope probe. This edge is faster -- about 10ns -- there is ringing from the probe.

    The rising edge measured by the diff probe.

    I can't say for sure which probe is telling the truth about the waveforms. My equipment is not up to the task of properly evaluating the performance of the differential probe. But it is good enough for my purposes at this time.

  • Custom Leads

    Bud Bennett02/12/2020 at 16:54 0 comments

    The inductance of a 3 inch long 26AWG wire is 1.23µH. It plays havoc with the diff probe if not properly accounted for. I ginned up a simple circuit to help me visualize how to dampen the ring associated with the lead:

    A small value of series resistance in the lead will remove most of the ringing caused by the lead inductance. The largest value of series resistance that can be inserted into a 3 inch long lead is 620Ω. Any more than that will reduce the bandwidth of the differential probe. There is also a small, 0.006%, reduction in gain.

    With 620Ω, the response to a 1V input step with a 3.5ns risetime is:

    There is a 10% overshoot in the probe's response to this fast step. Clearly, if you need fidelity on signals with very fast rising edges the lead length must be as small as possible. But if the rise time is reduced to 10ns the overshoot decreases to only 3.5%:

    I'll be making couple of sets of custom leads, with unique values of series resistance. There is no need to add any resistance to the PCB. The lead set to use will depend upon the characteristics of the signal to measure.

  • Ringing & Swinging

    Bud Bennett02/11/2020 at 23:10 0 comments

    I can’t leave well enough alone. I still don’t have the SMA connectors to allow direct connection of the probe to the scope input, but I wanted to evaluate the probe in some more real world environments. I thought that a good test was to connect the diff probe to the gate drives of my T16 battery charger. These signals are 5-6V transitions with expected rise/fall times on the order of 10ns. In order to capture all of the information, you must have 35-70MHz of bandwidth. I have seen, and dealt with, quite a few buck converter gate drive issues over the years. 

    The top gate drive of the buck converter has a slower rise/fall time than the bottom gate drive. This is usually due to the top FET having a bit more total charge, Qt, because the top FET is a PCh FET vs. the NCh bottom FET. I connected the diff probe positive lead to the top gate signal and the negative lead to the GND of the battery charger. This is what was output from the diff probe:

    The yellow trace is the output of the diff probe, measured by CH1 of my scope. The blue trace is Ch2 of the scope measured with the 10x scope probe. This waveform doesn’t show much ringing associated with inductive parasitics. That could be because of the slower, 40ns, rise time or the care that I took with the PCB layout :). 

    I then connected the probe across the bottom gate signal. This signal is much faster, 5ns rise/fall, than the top gate.

    There is quite a bit of ringing. The ringing can be a result of many things: poor probe grounding, marginal stability of the diff probe amplifiers, unknown bad stuff? Usually, when confronted with unexpected circuit performance I tend to go back to the simulator and see if I can reproduce the effect and get a handled on the problem. I was looking for something that affected the second gain stage, or all gain stages, that would present as shown. I found two things: R23 & R24 had an effect on the ringing due to a fast rise/fall event, and the lead inductance that I was using should have caused ringing when the lead inductance interacted with the input capacitance of the attenuator. Or perhaps it was both, or something else?

    In simulation, I found that the first stage amplifier ringing reduced when R23 & R24 were raised above 1kΩ. This damped oscillation was not present in the simulation until the stimulus had some common mode component. And this ringing was quite high in frequency and did not pass through to the probe output. 

    I also found that the longish leads that I was using to connect the diff prove to the Buck converter may be causing a problem. A 12 inch long 26 AWG wire has a self inductance of around 5µH. If you combine that inductance with the ~2.5pF input capacitance of the probe it will tend to ring at a frequency around 50MHz. If there is no resistance to dampen the Q of the input LC then the ring can be pronounced. A small amount of resistance in series with the probe inputs could remove most of the ringing at the input by reducing the Q of the LC circuit. If too much resistance is added to the diff probe inputs then the bandwidth would be compromised.

    I added 470Ω to each of the diff probe's inputs, using leaded metal film resistors. I did not see a significant reduction in the ringing. So what's really happening?

    I turns out that channel #1 of my Rigol 1102 appears to ring a lot more than channel #2. At this point I don't know which channel is bad, but my money is on channel #1.  When I connected both channels of my Rigol scope to the diff probe output I obtained these results:

    My Rigol scope has a problem with ringing on Channel #1. I think the diff probe performance is acceptable. [Caveat: I haven't directly connected the diff probe to the scope yet.]

  • Results from First Prototypes

    Bud Bennett02/09/2020 at 21:49 0 comments

    PCBs and components arrived within a day of each other. I had the first prototype assembled and working in an afternoon, with some compromises: I used 3.6kΩ/390Ω resistors for the first gain stage instead of 3.92kΩ/412Ω, and I don’t have any SMA connectors yet, so I can’t tie the output directly to the scope input. So the gain is 3% low.

    Power supply current = 52mA.

    First thing I checked was the ground supply to make sure that the M7301 opamp was not oscillating — no buzz. I did not populate the 22µF ceramics at the input and output of the LM7301, and I made sure to use 10µF tantalum capacitors with an ESR spec of 2.5Ω to damp any potential ringing at 100kHz.

    Next, I tied the inputs together and connected a scope probe to the output. No buzz there either...I’m breathing a bit easier now. I measure the output offset voltage to be -18mV. This seems a bit high (low, really) to me, but it is within the envelope of the amplifier’s Vos x Gain. The easiest way to see a trace is to connect the probe inputs across the scope’s square wave calibration output. The output is 386mVp-p for an input of 3.96Vp-p — 0.0969V/V. The lower gain is expected because of the temporary resistors installed. 

    But something is not right. The probe output is showing a pronounced rounded edge instead of a sharp square edge. Adjusting the trimmer capacitor in the positive signal path has a negligible effect. I can calculate the probe’s AC gain from the difference between the input and output waveforms — it’s about 7.3% too low. I remove the 15pF 0603 capacitor in the attenuator. This helps, but the AC gain is still 5% too low, so I replace one of the 300pF capacitors with a 270pF 1%/0805. Now the DC and AC gains almost match perfectly and I can use the pot and the trimmer capacitors to fine tune the probe. [This is a strange result for me. I bought 10pF±1% NP0 capacitors for the attenuator, but they appear to be 7.5% low after assembly onto the PCB. I believe this is the first time in my life that capacitance is lower than expected!]

    At this point I decided to build a second prototype, with the same component values now in place in the first one, to see if there was some big error somewhere. The second prototype performs nearly identically with the first, except the output offset voltage is only -11mV. In a way this is a good thing...consistency is good.

    Let’s measure CMRR:

    I believed the simplest way to measure low frequency CMRR was to connect both probe inputs to a 125VAC signal and see what the output did. It was not very pretty:

    It’s not really bad — about -90dB — but I was expecting a better result. Maybe I had not calibrated the CMR and AC gain correctly. So I connected both inputs to a 30VDC supply and rotated the potentiometer to cancel the DC CMR error while looking at the outputs with a DVM. I got the DC CMRR to better than 100dB. Here’s a screen dump of the HP DVM showing the DC output voltage of the probe changing with the the applied 30VDC CM input. Each large spike is me turning off/on the 30V supply. The only thing that's important is the deviation between the large spikes. It’s 0.25mV! That’s more like what I was expecting.

    I set up a test circuit to provide a 1kHz 30Vp-p square wave (a more controlled waveform than the mains input) to test the CMRR. This was just a 2N7000 FET with a 3.3kΩ load resistor connected to a 30V supply, with the 2N7000 gate driven by the 3V scope calibration output waveform. Rise/fall times improved from ~1µs to ~100ns. I was able to calibrate out the DC and AC common-mode contributions to where the probe output looked like this to a 1kHz +30V CM square wave:

    Ch2 shows the 30V square wave. I would call this less than ±1mV p-p variation. That’s > 83dB CMRR at 1kHz.

    Unfortunately, I don’t have any appropriate equipment to thoroughly test the differential probe performance beyond this point, unless I get access to better...

    Read more »

View all 10 project logs

Enjoy this project?

Share

Discussions

Keith wrote 01/15/2024 at 16:31 point

Hi Bud,

I made one of these a while back and it works well,  using Paul's version of the design. Testing it on my HP4396B VNA, I see a dip in s21 (through response) of about -6dB at around 35MHz. I believe its due to the parasitic capacitance coupling between the two input divider chains, which are quite close together. The series inductance of the resistor chain and the parasitic capacitance between them are the likely cause.

Commercial probes like the MicSig one space the two divider chains out, to the extent of making the PCB U-shaped to minimise the coupling capacitance. I wonder if it's worth trying a new PCB layout to test the theory out. 

BTW, apart from that dip the frequency response is pretty good (i.e. within 3dB) out to about 250MHz . 

Keith

  Are you sure? yes | no

Peter wrote 12/29/2023 at 23:58 point

I've successfully built a couple of these probes based on Paul's version, and they appear to be working well. You can check them out here: Link to photos

Thanks.

  Are you sure? yes | no

Jan Rownicki wrote 11/04/2022 at 13:17 point

Are the design files available? If not could you please generate a pick and place file? I'd like to use JLCPCB to avoid assembling it by hand.

  Are you sure? yes | no

Bud Bennett wrote 11/17/2022 at 22:11 point

The Gerber files are available in the "Files" section of this project, if you wish to build my 5V version. You will have to see PaulV's Git page for his Gerbers. I will attempt to generate a pick and place file (and JLCPCB specific BOM) and upload them to the Files space soon.

  Are you sure? yes | no

Bud Bennett wrote 11/17/2022 at 22:12 point

There are no design files yet for Christoph and Wolfgang's approach.

  Are you sure? yes | no

Bud Bennett wrote 11/17/2022 at 23:22 point

Well...I tried to redo the BOM for JLCPCB, but it is a useless exercise since none of the necessary parts are in stock. They will not be able to assemble it with all of the components and you will have to solder a majority of parts anyway. Not happening.

  Are you sure? yes | no

Wolfgang Griebel wrote 04/13/2021 at 00:25 point

Hi Bud,

I like your probe. What I would like to do is integrate it to my Keysight scopes autoprobe system so power supply and offset can be provided by the scope and it is automatically recognized and calibratable.


I hacked something like this on my webpage

https://electronicprojectsforfun.wordpress.com

My problem:

I am too old for small SMD stuff. What I could provide is the adapter PCBs, some 3D printed enclosures, and the like. But I would need to get a completed probe board.

Interested ?

Best regards, 73

  Wolfgang DL1DWG

  Are you sure? yes | no

Bud Bennett wrote 04/15/2021 at 14:31 point

I'm not interested, since I have two working probes. But someone else might.

  Are you sure? yes | no

raviteza88 wrote 04/06/2021 at 18:45 point

Hi Bud, I have 2 questions.

1) The GBW of LTC6269 is 500Mhz, and you have a 20x gain in your first stage. Meaning don't you get only 25 MHz bandwidth @ 20x gain? 

2) I simulated your LTspice and wanted to play with the first stage feedback resistors but it started to oscillate. By adding a 0.2 pF the oscillations died down, Is there some math which can be used to calculate the capacitor instead of trail and error.?

Thanks

Ravi

  Are you sure? yes | no

Bud Bennett wrote 04/07/2021 at 03:42 point

Hi Ravi, The first gain stage uses LTC6269-10 amplifiers, which have a GBW = 4GHz. The second stage uses the LTC6268, with a GBW = 500MHz, but only has a closed loop gain of 1.25. Can't help you with the simulation...don't know where you added the capacitor. There are a few application notes about how to compensate these amplifiers. Look for them in the support documents offered by ADI (or LTC).

  Are you sure? yes | no

marcinszajner wrote 04/02/2021 at 10:59 point

Based on your scheme I have my own, but 1:8000 for higher voltage (10kV):

https://www.youtube.com/watch?v=ZJHYIH9O-JU

Your project and description help me a lot!

  Are you sure? yes | no

marcinszajner wrote 01/16/2021 at 09:07 point

In LTspice you have use 0.15pF capacitance for 0603 resistors. You have calculate it or just some intuitive value? Because I build 10kV probe 1Mhz and I have problem with capacitance for voltage divider. I could put 100< resistors if they have really 0.15pF +/- 100% parasitic capacitance, but I cannot find anywhere how estimate capacitance for resistors. Could you help with estimating resistor parasitic capacitance?

  Are you sure? yes | no

Bud Bennett wrote 01/18/2021 at 03:59 point

There is a lot of stuff out there to help you estimate parasitic capacitance. It depends upon the dielectric constants, board thickness, trace width, and distance between components. You must take you best guess, unless you have some very expensive software that can do it for you. The 0.15pF is discussed in the details section regarding the design. You need some feedback capacitance to cancel the capacitance at the opamp inputs. See the data sheet and app notes for more details.

  Are you sure? yes | no

Jon S wrote 11/15/2020 at 01:37 point

I'm very curious about the common mode voltage limit. You didn't really specify one(Though you had a requirement of at least 340V) and I suppose that makes sense since you can't test it.

That said, don't a lot of switch-mode PSUs use a boost converter active PFC circuit? I would expect the bridge voltage on a normal power supply to reach as high as 400V.  Looking at Dave Jones' teardown of a LeCroy diff probe showed that you have /more/ attenuation in your input stage. Safety concerns aside with needing a case and proper insulation, any reason you think it wouldn't be able to deal with 400-450V common mode voltage?

  Are you sure? yes | no

Bud Bennett wrote 11/20/2020 at 20:17 point

Safety aside...

The power supply is approximately ±2.5V. A 340V peak CM voltage +25V differential input voltage is 365V peak. If you divide that by the attenuation (250x) then the input voltage presented to the amplifier is 1.46V. The input voltage range of the LTC6268-10 is -0.1V to VCC-0.5V, so there is some room left (another 125V) before the input voltage range is exceeded. So yes, it can probably hit 450V CM before the input is clipped.

  Are you sure? yes | no

jnesmoht wrote 11/05/2020 at 08:30 point

Hi!

I realize something strange here. On the pic I we can see the power adapter that has two pins. 

And in Europe the scope are all grounded. So that means the "isolation" and the trick with the single power opamp rely on the isolation of the power adapter. 

I personally do not find that super cute.

  Are you sure? yes | no

Bud Bennett wrote 11/06/2020 at 00:55 point

I believe all power adapters are isolated — both here in the US and in Europe, where Paul lives. They are cheap and ubiquitous. US scopes are all grounded. I’d like to see a schematic with your super cute idea.

  Are you sure? yes | no

jnesmoht wrote 11/06/2020 at 02:07 point

Hi,

sorry, don't take it personnal :) I appreciate the probe of course but the powering stage is bothering me. I have a background in power electronics where isolation is a big topic both for personal and HW safety. Just check how many people got shocked from cheap USB charger.

Beside that I have no real cheap and elegant solution. Perhaps battery powered.

  Are you sure? yes | no

nike9307 wrote 10/27/2020 at 20:30 point

This is an awesome project. I want to design something similar, and this has been a gold mine for mine. 

I just have one small note, that could probably help with your design. I remember reading an Analog Devices article or application note about this. When using split rail supply, not a dedicated plus and minus power supply, one should never decouple the op-amps to the GND on the split rain, but they should be decoupled from one power pin to the other, because that is where the return current is going to. Unfortunately i cannot find the article i'm referring to, if i do i will link it.

EDIT1: I can give an example of a well know schematic bypassed that way. Dave Jones's uCurrent Gold has a very similar virtual ground schematic and he bypassed the op-amps that way.  

  Are you sure? yes | no

Bud Bennett wrote 10/27/2020 at 21:35 point

Thanks for the tip. I referenced several articles on high speed design and there are several opinions about it. It make sense to only bypass to the supplies in this circuit since most of the current is sloshing between the two. But the ground plane is also the largest, lowest impedance piece of metal on the PCB. I’ve got it covered both ways. The circuit works...what can I say?

  Are you sure? yes | no

nike9307 wrote 10/28/2020 at 07:14 point

You are welcome. I'm not sayin the circuit does not work, i'm saying it may be possible to make it oscillates less. Its your project, the final design decision is yours. I was just pointing something interesting i've read from a reputable company's app note/article.  

Thank you again for the great project and for describing how and why you made the decisions in it. 

  Are you sure? yes | no

Tyler Gerritsen wrote 06/19/2020 at 19:13 point

This is awesome!  Nice work!

I'm developing a high-speed LED flash that strobes for about one microsecond.  Just started pricing out a differential probe for more accurate current measurements when I stumbled on your design.  Thanks for sharing, I ordered a few boards from JLCPCB.  Due to the minimum order quantity I'm going to have a few left over. 

If anyone wants a PCB and it's OK with Bud & Paul, shoot me a PM.

  Are you sure? yes | no

Bud Bennett wrote 06/19/2020 at 20:44 point

It's OK with me (and probably Paul). Did you build my version or Paul's? Be aware that it seems that nobody has any LTC6268 (not the -10 version) until sometime in August.

  Are you sure? yes | no

Tyler Gerritsen wrote 06/19/2020 at 21:21 point

I used your design (5V supply).  Good to note, I guess it's going to be a wait...

  Are you sure? yes | no

solanki.pareshkumar wrote 06/12/2020 at 06:31 point

Hello Mr. Bennett and others

I have found one more JFAT op-amp OPA659 from TI. It has stable unity gain with bandwidth of 650 MHz and slew-rate of 2550 v/us with 4V step. It has Vcc to Vee range of 12V.

I would like to measure the signals with highest measuring bandwidth of 10 MHz and peak signal amplitude of 2KV on either side of zero.

Does this op-amp makes a good candidate for this design with x250 or x500 probe. As I am new to this measurement electronics design, i would appreciate any feedback from experienced members.

Thanks for your reference design and best regards

  Are you sure? yes | no

Bud Bennett wrote 06/16/2020 at 12:58 point

From the information you provided above, it doesn't seem like you need to use this topology since there is no common mode voltage that needs rejection. Therefore a simple voltage divider would suffice. If you insert the OPA659 into this topology (for whatever reason), and increase the voltage ranges, many things would need to change: voltage ratings on resistors/capacitors in both the attenuator and the supply bypassing, feedback resistor/capacitors for stability, PCB spacings for voltage standoff, etc. The open loop gain of the OPA659 is only 52dB, around 300 V/V, so closed loop accuracy might be poor. Additionally, I previously tried to simulate a TI OPA657 in this topology and I could not prevent the two opamps in the first stage from oscillating. You might have success with the 659 in a lower bandwidth requirement. Good luck.

  Are you sure? yes | no

Keith wrote 05/09/2020 at 09:04 point

ALso (sorry for all the questions!) did you use a stencil and/or solder paste?  I'm new to SMT and still trying to get my head around the best way to solder such small components - especially as those opamps aren't cheap! Thanks.

  Are you sure? yes | no

Bud Bennett wrote 05/09/2020 at 13:55 point

I generally don’t use a stencil, unless the pad pitch is 0.5mm or less. I use Kester 37/63 lead solder paste in a syringe with a pink or purple needle size. I keep the syringe in the fridge between uses. When applying solder paste to DFN parts, like this amplifier, I squeeze out a bit of paste and then use a pin to get just the right amount on the pads. I would suggest that you use a stencil since things tend to go a lot faster. The reason that I don’t use a stencil is because the cost when making several PCB revisions. 

Paul Versteeg is working on a slightly different version of this probe. I suggest that you wait until his gerber files are available before ordering PCBs. I put a link to his blog in the project details.

  Are you sure? yes | no

Keith wrote 05/09/2020 at 02:07 point

Did you solder the board using an iron, or use an oven? It looks incredibly fine for an iron.

  Are you sure? yes | no

Bud Bennett wrote 05/09/2020 at 02:11 point

I used a hot air rework station — made by Quick. It seems to work for me. No need for an oven. My hands shake too much to attempt soldering a 0402 capacitor with an iron.

  Are you sure? yes | no

Frits Jalvingh wrote 03/26/2022 at 09:41 point

I managed to do two with an iron. It is challenging but if you make sure to keep your hands on the table (not holding them high) and have a good magnification thing it is possible. You also need a good and fine tip for the soldering iron.

I will not say the result is beautiful as far as the 402's are concerned ;) but it was fun to do.

  Are you sure? yes | no

Bud Bennett wrote 03/29/2022 at 03:10 point

Did it work after you finished?

  Are you sure? yes | no

Frits Jalvingh wrote 03/29/2022 at 06:50 point

I just finished my second one. Both work ;)

Thanks a lot for this very nice project!

  Are you sure? yes | no

Dan Maloney wrote 01/10/2020 at 16:46 point

I'm only just starting to get into oscilloscopes, and from what I've read there are way too many ways to kill a mains-powered scope. Makes me a bit nervous using mine to diagnose anything but battery-powered circuits. I've been considering buying an isolation transformer to power the scope - still thinking that over.

I'll be following this with interest. Thanks!

  Are you sure? yes | no

MS-BOSS wrote 02/12/2020 at 20:32 point

An oscilloscope on isolation transformer has one major problem. Attach the probes to some HV circuit and then touch one of the unused BNC connectors. Rubber shoes are mandatory.

  Are you sure? yes | no

Bharbour wrote 03/29/2022 at 13:52 point

Stacking other test equipment on top of a transformer isolated scope can also be "problematic"

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates