Group Actions
Engagement
  • Author Author: shabaz
  • Date Created: Date Created
  • Views 16687 views
  • Likes 17 likes
  • Comments 37 comments
Related
Recommended

Building Solderable In-Circuit Oscilloscope Probes

shabaz

 

Introduction

The probes attached to oscilloscopes are fundamental to getting useful captures and measurements from the ‘scope. The normally seen ‘scope probes (technically they’re called passive probes) have a special construction that is not easy to DIY; it’s best to purchase these. They have a distributed resistance running all along the wire (it is not ‘normal’ coax!) and without that the oscilloscope will display a distorted signal.

 

Many such passive probes come with a pointy tip and with attachments such as grippers. The probe end can sometimes be inserted into sockets too. However, it is rarer for supplied probes to come with a solderable wire attachment (although some probes do come with that). This blog post documents an experiment to build solderable probes, the idea being that the probes can be quickly tacked with solder directly onto the board under test.

image

 

I had written some notes for building custom probes several months ago, but I’d not got around to documenting it because I did not have a decent way to test it all. Since then, Building a Fast Edge Square Wave Generator allowed me to collect some measurements.

 

This blog post may be useful for those who need low-cost ‘scope probes, or for those who want solderable probes for attaching to a circuit board, temporarily or even semi-permanently. For another technique for using probes, see DIY $10 Solder-in Oscilloscope Probe

 

It can sometimes make a huge difference to usability, not needing to hold probes in position manually! To cut a long story short, all it entails is soldering a 950 ohm (or 953 ohm - easier to find!) resistor inline with the center conductor of a coax cable. It is inside the yellow pod in the photo below. Connect the other end to your 'scope and set it to 50 ohms, and that's pretty much it : )

image

 

If you’ve recently purchased an oscilloscope, hopefully, it came with ‘scope probes but if it didn’t, you can make these! However, there are limitations; the probes described in this blog post have an impedance of 1 kohm. This is lower than typical passive probes that come with oscilloscopes, so it may load the signal to be measured too much. The best approach to using the solderable probes described in this blog post would be to first briefly check with a normal passive probe to see what the signal should look like.

 

Why not just use Coax Cable

One school of thought could be to get a coax cable, attach one end to the ‘scope and the other end to the signal on the device under test. Ordinary coax cable can be used in such a scenario but only if the signal is intended for a 50 ohm load, and the ‘scope is set accordingly to a 50 ohm termination.

 

It is best to avoid ‘unusual’ attachments like BNC-to-banana or BNC-to-alligator, or ordinary coax cable, unless you’re very sure of what you’re expecting to measure, otherwise, one will get burned sooner or later with confusing ‘scope trace results.

 

If the oscilloscope is not set to 50 ohms load, then fast-changing signals will appear distorted; there is an explanation for this here.

Also, it’s good to avoid setting the ‘scope to 50 ohms unless you’re really sure that the signal source can handle this low impedance, and that the ‘scope input can handle the power going into it’s internal 50 ohm resistor - to prevent blowing up the 'scope input.

 

Some oscilloscopes do not have a 50 ohm input. For such ‘scopes with up to around 200 MHz bandwidth, it is ok to use either an inline 50 ohm resistance or a BNC T-piece with 50 ohm terminator. Note however that nearly all low-cost 50 ohm BNC terminators are designed for old-school local area network or LAN connections and will not work very well; it’s best to buy a terminator specifically intended for operation at high frequencies, or make your own!).

 

How Do Probes Work?

‘Normal’ Passive Probes

Normal oscilloscope probes (also known as 10X or 100X passive probes) use a specialized, custom cable for ultra-low capacitance. This is needed in order not to attenuate the signal too much at high frequencies. At either end of the cable, passive components are used to create a potential divider circuit, so that the circuit under test is not impacted much when the probe and ‘scope are attached to it. The potential divider circuit is a bit specialized because it includes a compensation capacitance within the divider, so that the (unwanted but inevitable) ‘scope input capacitance, combined with the probe cable capacitance, does not act as a low-pass filter. This type of ‘scope probe is better purchased than DIY’d!

 

Resistive Probes

An alternative type of probe, known as a resistive probe, has a simpler topology. It relies on normal coax cable. Typical ’50 ohm’ labelled coax cable has a property that, if the far end of it is connected to a 50 ohm resistor, means that any signal input at the other end will also see a pure 50 ohm resistance and actually won’t see any reactance, regardless of the capacitance and inductance of the coax. As a result, the potential divider can be greatly simplified compared to a conventional passive probe! (The feature of coax that causes this behavior is known as its transmission line property, and that property enables many operations transporting signals in systems, especially communications circuits, networks and computers).

 

Resistive probes are described in online and offline literature. The Art of Electronics suggests using a 950 ohm (or 953 ohm since that is an E24 value) resistor in series with the coax cable, at the signal end. The potential divider formed with the 50 ohm oscilloscope input setting, therefore, reduces the voltage by 20 times. The resistor value is a compromise; too high a resistance and the signal is attenuated too much for the oscilloscope to display well, and too low a resistance and the circuit under test gets loaded too much. Since a resistor is not perfectly resistive and has some capacitance and inductance too, the construction of the resistor can have an affect too, and the impact of that may differ depending on the resistance value as well.

 

Using a 950 ohm resistance, the circuit will see a 1 kohm load when the far end of the probe is terminated with a 50 ohm resistor. 1 kohm isn’t great for all scenarios, however for many circuits, especially logic gate outputs, it can be acceptable to attach to such a load.

 

With microcontroller and digital circuits, it can often be necessary to probe several connections for serial busses, and it could be awkward to use many probe clips while debugging. A solderable option for the probes could be attractive for that scenario.

 

Building It

Construction can be very straightforward. Any 50 ohm coax cable can be used. Attach a BNC connector to one end. If you don't feel comfortable doing that, it is possible to purchase BNC-to-BNC cables and cut in half, or chop off one end.

 

For the resistor, a 1% tolerance 953 ohm thick film (not thin-film) resistor can be used, or perhaps a 5% tolerance 1 kohm resistor (many could be purchased, and individually measured to select the ones closest to 950 ohm).

 

For my implementation, I used RG-178 coax. If you’re unfamiliar with it, it’s one of the most useful coax types for electronics labs as general-purpose hookup-coax, because it is so easy to work with, and excellent for short connections.  It is thin (1.8 mm overall diameter) and the outer insulator is easily removed by lightly scoring with a knife, bending the coax slightly to tear at the score, and then gripping with wire cutters and sliding it off. Another excellent property is that it’s almost impossible to melt the inner insulator (dielectric) when soldering.

 

I wanted to make several of these probes, so I used heat-shrink tubing at each end for color-coding. I used 2.5 mm2 ferrules to terminate the outer braid, however, it’s not necessary; you could just spin the braid around the outer insulation and tin it with solder to form a circular termination, all ready for attaching a ground cable. Since the coax is so thin, the outer termination could also be directly soldered to the nearest ground point or ground plane on the circuit board under test.

 

The diagram below contains measurements in case anyone wants to assemble it in a similar way, to get comparable results as I got. I'm not saying my method was good or bad; it's just one data point, so experimentation could improve the results further.

image

 

I used an 0805-sized 953 ohm thick film resistor, and fairly thin 30AWG stranded PTFE-insulated wire. This will allow me to solder the probes even in fairly dense locations on a PCB.

image

 

The resistor and the connections to it were covered in ‘instant’ epoxy glue (it sets in a minute or two), and I tried to get glue over the insulation of the wire too, to secure it all properly. Maybe this step isn’t required.

image

 

Next, to make the colored plastic blobs, Polydoh was used. If you’ve not used it before, it is like a manual version of 3D printing plastic; it melts at a low temperature and can be formed with ease. It comes as a bag of white granules/pellets and can be purchased bundled with pigment granules too, to be mixed in. It is really cheap – about $15 for a bag containing thousands of granules.

image

 

Ordinarily, the granules are supposed to be used with boiling water, but I didn’t want to get my resistor wet, so an alternative procedure was used. I took two granules and placed them on a surface (I used a metal block) and heated it with a hot air tool. After a few seconds, it was possible to place the wire and resistor between the two granules and fold it all around into a kind of pod or egg-shape, using bare fingers (it doesn’t burn the fingers, just feels slightly hot briefly). If it hardens too soon before it is fully shaped, then it can be re-heated with the hot air tool.

image

 

If you don’t want white blobs and want to color-co-ordinate with the ‘scope channel colors, then instead of just heating two granules, add a single pigment granule too, and then mash the molten color into the molten granules with fingers, and then (because it will have cooled and solidified) place back on the block and re-heat, and then wrap it around the resistor and form the blob. It’s really easy!

 

Using It

For peace of mind, as mentioned earlier, if you’re unsure what sort of signal levels to expect, a normal passive probe should be first used to be aware of what the signal should approximately look like.

 

To use the resistive probe, if your oscilloscope has a 50 ohm termination setting, that needs to be enabled. Otherwise, a 50 ohm terminator and a BNC T-piece is required for each probe. This is mandatory.

 

Next, set the oscilloscope scaling to 20X. If the oscilloscope does not support such a setting, then the scaling will need to be done mentally because the displayed voltage levels will be scaled down by that factor.

 

Solder on the probe to the circuit under test, attach the probe to the ‘scope and power up the circuit under test!

 

Comparison with Other Probing Methods

For a test signal, I used a square wave generator circuit (see Building a Fast Edge Square Wave Generator  ). The reference signal (captured with a plain 50 ohm coax connection to the ‘scope) is shown at that blog post, and will be used for comparison purposes as a near-ideal capture.

 

I tested three different probing methods:

 

  1. Simple plain coax connected to the square wave generator, and the oscilloscope set to its normal 1 Mohm setting. This will be a bad method as will be visible in the ‘scope traces.
  2. The home-made resistive probe, connected to the oscilloscope set to 50 ohm input
  3. Active FET probe. This probe should give good results but is a pricey option. It will be good to compare the results.

 

The photos below show the three different methods. For all three methods, the square wave generator output was directed into a 50 ohm load, which can be seen screwed onto the SMA connector at the top of each photo.

image

 

The oscilloscope traces below show the results for each of the probing methods. The reference trace (from the square wave generator blog post) is shown on the right for easy comparison. The coax method was, as expected, bad. Coax 1 and Coax 2 were attempts with different lengths of coax, and it made a difference, again as expected. In summary, both Coax 1 and Coax 2 look bad, with Coax 2 happening to look extremely bad. This is the danger of just using coax like this; the result can look extremely distorted.

image

 

The resistive probe results are shown below. There is a small anomaly visible about 8 nanoseconds from the time of the rising edge, but aside from that the result looks fairly similar.

image

 

The result from the active probe is shown below. There are a couple of small differences here and there, but generally the trace looks similar to the reference.

image

 

Summary

Solderable ‘resistive’ oscilloscope probes are easy to create, and although they won’t be suitable for all circuits, they can be handy for attaching to awkward locations for some circuits under test. The results show there really isn’t much difference between the expected reference trace, and the result with the solderable probe looks fairly good! It is not as general-purpose as a normal passive probe nor as high impedance as an active probe. The resistive probe will not be suitable for all purposes due to the 1 kohm impedance it presents to circuits, however, it could be handy from time to time. As mentioned earlier in the blog post, another solderable option is to try to attach the normal passive probe into a circuit using a PCB probe socket.

 

Thanks for reading!

Parents

  • This is interesting. I've seen the idea mentioned, but never tried it myself, so it's very useful to see your results.

     

    Resistive probes are described in online and offline literature.

    Howard Johnson and Martin Graham, in 'High-Speed Digital Design', simply go for a 1k resistor (ie 21:1) and use the oscilloscope to turn that back into sensible values [presumably they had a fairly good oscilloscope to play with].

     

    Henry Ott has an interesting variation on the theme in 'Electromagnetic Compatibility Engineering'. He suggests two of them, side-by-side, with 450R resistors (so 10:1), to make a differential probe for looking at differences between ground signals in separate parts of a plane (two 50R input channels, with subtraction for the differencing). He tacks the outer of the shields together every few inches [there's no connection for the screen at the probe end - the return currents just run around the end to form their own loop, if you see what I mean - if this isn't very clear, say, and I'll draw it - I don't want to reproduce his picture because it's still in print]. He also suggests using it with a 'high-frequency 180 degree combiner' to feed it to a single-ended spectrum analyser [that part looks very radio-engineerish to me, because he starts going on about insertion losses, and that kind of thing] which is probably more up your street than mine [as I don't even have a spectrum analyser].

     

    One reason 20:1 might work well is to do with the kind of capacitance you get at the input of the oscilloscope. For my scope, it's 11.5pF. If the end-to-end capacitance of the probe resistor is of the order of 0.5pF, then that's in the right kind of area for a 20:1 capacitive divider to match the resistive one. If you wanted to experiment, simply trying 1206, 0805, and 0604, and seeing which worked best might be one approach, though the difference might be too slight to see. Just a thought.

  • in reply to jc2048

    Hi Jon,

     

    Thanks for the info!

      wrote:

     

    One reason 20:1 might work well is to do with the kind of capacitance you get at the input of the oscilloscope. For my scope, it's 11.5pF. If the end-to-end capacitance of the probe resistor is of the order of 0.5pF, then that's in the right kind of area for a 20:1 capacitive divider to match the resistive one. If you wanted to experiment, simply trying 1206, 0805, and 0604, and seeing which worked best might be one approach, though the difference might be too slight to see. Just a thought.

    That makes sense, I hadn't thought of the 'scope's input capacitance on the 1Mohm setting which would get paralleled up with the 50 ohm terminator. I did think of trying different size resistors, but was sufficiently surprised the first attempt (0805) worked quite well, so didn't get around to trying others (I didn't have any larger resistor anyway). The 1206 would be interesting to try!

     

    Henry Ott has an interesting variation on the theme in 'Electromagnetic Compatibility Engineering'. He suggests two of them, side-by-side, with 450R resistors (so 10:1), to make a differential probe for looking at differences between ground signals in separate parts of a plane (two 50R input channels, with subtraction for the differencing). He tacks the outer of the shields together every few inches [there's no connection for the screen at the probe end - the return currents just run around the end to form their own loop, if you see what I mean - if this isn't very clear, say, and I'll draw it - I don't want to reproduce his picture because it's still in print]. He also suggests using it with a 'high-frequency 180 degree combiner' to feed it to a single-ended spectrum analyser [that part looks very radio-engineerish to me, because he starts going on about insertion losses, and that kind of thing] which is probably more up your street than mine [as I don't even have a spectrum analyser].

    I think  I see what you mean! I'll try to dig up the document you mention, just to be 100% sure, but I think I can follow your explanation. I do have a splitter/combiner (was just using it today for a specific exercise) mine covers 5 MHz to 200 MHz range, it is Minicircuits PMT-1+ (PDF datasheet).  I wish I had one that covered a higher range, but it's so expensive ordering from Minicircuits (nothing is cheap there). It is an interesting device, it could be worth also occasionally checking ebay for these too. You're right it's popular for radio uses, but it can also be used with a 'scope instead of a spectrum analyzer, although for that use-case the spectrum analyzer would be needed for the sensitivity. It can be used to check for things like delays i.e. phase differences between the two inputs, by seeing how much the output changes by. Or, it can be used to generate two out of phase signals, if the same signal is input to both inputs. It's quite a versatile device. I'll try to assemble something like this at some point (likely in a couple of weeks time) because it sounds super-useful! Maybe it can be used with a 'scope with an amplifier. I'll experiment a bit. It seems a really clever concept, I would never have thought to apply two 450 ohm probes in this way.

  • in reply to shabaz

    To reduce the tip capacitance further, putting two resistors in series at the probe end would do. Not sure it's worth fiddling around with, though, because what you've got is so close to the active probe already and you'll likely just be degrading what you've already got rather than improving it. You're already in the area where you can't really tell which is the 'right' waveform (unless you've got a 10GHz scope hidden somewhere and haven't told us).

     

    The combiner he talks about in the book is a Mini-Circuits Model ZFSCJ-2-1. He says it has a response of 1 to 500 MHz and about 4dB insertion loss. Obviously, with a spectrum analyser, you'd need to be very careful of the dc levels involved if you tried to use it for anything other than differential noise measurement on a ground plane [I'm sure you understand that - just throwing it in as a note of caution for a more general audience].

  • in reply to jc2048

    Hi Jon,

     

    Thanks for this! This is useful information. It would be nice to have response down to 1 MHz, or even lower!, it would be useful for examining noise from (say) DC-DC converters. That device is about £60, but there is PSCJ-2-1+ for 1-200 MHz, which works out to £34 including postage (it's a through-hole version, so would need connectors attached, or even easier, to permanently connect the coax to it, and dedicate it for this role. It sounds like a cool project. I found a link to a coupler (PDF doc) that works all the way down to 40 kHz, but it would need experimentation (and not the easiest diagram in the world to follow but not insurmountable), the cores it refers to are not easily available it seems.

  • in reply to shabaz

    Sounds like an interesting project, though not the sort of thing I'd have the knowledge or skills to do [I'll happily be a spectator whilst you do all the hard work].

     

    For the area up to a few MHz, a custom differential low-noise preamp might be an alternative approach.

     

    To go back to the single-ended probe, do you have [or have access to] an rf signal generator? It might be interesting to see what the frequency response of your probe looks like. Whether it's flat lower down and where it starts to roll off at the top end.

  • in reply to shabaz

    This is essentially what his drawing shows for the differential probe.

     

    image

     

    This next one illustrates what I was trying to say about the return current.

     

    image

     

     

    If you connect the braids at the probe end nicely, the return current that develops one side, on the inside of the braid, as the wave propagates along, folds round and becomes the return current for the other [for a differential mode signal, which is what the probe is trying for]. I think that's right, though I'm not always very good with this kind of theory. I suppose, if you were being a bit perfectionist about it, you'd extend the return path alongside the resistors and fold it round a bit nearer the tips of the probe.

  • in reply to jc2048

    Hi Jon,

     

    Thanks for the diagrams, nicely drawn! I visualized the behaviour by mentally removing the coax and replace with two 50 ohm resistors in series, and then the circuit looks like 1 kohm impedance across the probe ends.

    I made a bit of it, still need to complete it. I'm thinking to wrap the outside in tape or insulated braid afterwards otherwise it's a shorting hazard : )

    For the 450 ohm resistors, I can try 150 ohm and 300 ohm resistors in series.

    image

Comment
  • in reply to jc2048

    Hi Jon,

     

    Thanks for the diagrams, nicely drawn! I visualized the behaviour by mentally removing the coax and replace with two 50 ohm resistors in series, and then the circuit looks like 1 kohm impedance across the probe ends.

    I made a bit of it, still need to complete it. I'm thinking to wrap the outside in tape or insulated braid afterwards otherwise it's a shorting hazard : )

    For the 450 ohm resistors, I can try 150 ohm and 300 ohm resistors in series.

    image

Children
  • in reply to shabaz

    Yes, to the circuit it will look like 1k.

     

    Agree about the shorting hazard. I'd be quite nervous about all that bare metal waving around.

     

    An further experiment you might do, when you get it going, would be to add a ferrite fairly close to the probe end to stop external rf currents running along the outside of the shields, in the way that people often do with EMC aerials, and see if that makes any noticeable difference.

     

    I should have read a bit further in Ott's book. Over the page from the simple differential one, he describes how the frequency can be extended to 1GHz [from 500MHz]. He points out the problem with the oscilloscope's input capacitance "typically 3p to 10pF" affecting things once you get above 500MHz. His solution is to better terminate at the probe end, so that any wave coming back gets absorbed, rather than reflecting off the 50R/450R mismatch at the input and setting up standing waves.

     

    "The solution is to place a 50R termination at the probe tip (between the centre conductor and the shield) on the cable side of the 450R resistor. To reduce any inductance in series with this shunt resistor, the termination is usually constructed from four 200R resistors in parallel."

     

    He references a probe design by Doug Smith, which seems to be this one http://www.emcesd.com/1ghzprob.htm

     

    Then he [Ott] says: "For the balanced differential probe, use a 450R tip resistor instead of the 976R resistor suggested in the construction detail, leave the ground lead off, and modify both tips of the balanced probe as described".

     

    That looks like it would be 19:1, to me, rather than 20:1. Shouldn't it be 475R?

     

    Anyway, a bit more information there to throw into the pot.

  • in reply to jc2048

    Hi Jon,

     

    Interesting link, and I too suspected the probe as it is would only be useful up to around 500-600MHz or so, which is great for 'scopes but spectrum analyzers can go much further (I did try to get the frequency response of the soldered probe, but my setup was very crude and I could see that after about 600 MHz, the response increased, due to either the test setup or the capacitance of the 950 ohm resistor, or some interaction! but it was a bad setup so I couldn't conclude much - and certainly there was impact from reflections, so the match was not perfect everywhere in the testbed and/or probe, although it only impacted by about +-1dB up to 600 MHz or so).

     

    The comments at that site makes sense (and as you say, there must be a typo, it should be 475 ohms). I'm going to try out the design with just the 450 ohm resistor, since the other design lends itself much better to his construction method with the BNC barrel so I'd need to redesign anyway. But definitely worth doing for the better match if using the spectrum analyzer. The 40:1 ratio however would further reduce the chances of using a 'scope without any additional amplifier to see the result.

  • in reply to shabaz

    ... Interesting link...

    It is! When I opened it, I felt thrown back to 1996 HTML. But the content is great, and the design is simple and efficient.

  • in reply to jc2048

    I tried a quick experiment, a bit inconclusive currently! Not sure I'm doing this the best way...

    Anyway, from left to right, I drew a trace in copper tape, to simulate a poor thin ground plane in a C shape. There's a 50 ohm load (two 100 ohm resistors in parallel) at the bottom-right end of the C shape. The circuit is fed from the signal generator, at 35 MHz, -10 dBm. The differential probe red and blue wires, are just held in place with copper tape for now, and blu-tack to secure it a bit more. Then a load of ferrites before going into the PMT-1+ (terninated with a 50 ohm load on the unused zero degree port) and the 180 degree port of it goes directly into the spectrum analyzer.

    With the probe wires as shown in the photo, I see -50 dBm received on the spectrum analyzer. If I short the C shape by placing some copper on it, then the received signal drops, and I can vary the amount of drop by moving the copper short. I'm not sure I'm doing this correctly - perhaps I'm supposed to keep the distance between the two probes identical, so that the phase difference doesn't change, otherwise it is to be expected that the output will change.. but surely not by much at this relatively low frequency, were it a good ground plane and not an inductance like this long C shape. By shorting the C shape, I'm changing the distance between the probes but also the inductance, so the phase will have changed. Anyway, it's late, I need to think about this!

    Also by the way the ferrites did make a difference, if I connect just one probe and leave the other in free air, I still receive some signal. The ferrites in the photo reduced it by a sum of about 6 dB.. not a lot, but these were random ones that had a large enough aperture for the blobs to pass through. I also forgot to put tube/braid around the coax, now it's a bit harder to do : ( I believe the reason the ferrites are not making much difference, is that the majority of the pickup could be from the long red and blue wires. Perhaps I should reduce those, and not join the coax right near the resistor blobs (perhaps even put small ferrites on both of the coaxes separately where they split).

    image

  • in reply to shabaz

    Aren't the ferrites undoing all the hard work to capture fast edges?