Cheap Oscope Probe Quality

[TopherTheME]

Has anyone ever used some of the cheapo oscilloscope probes that you can get off ebay? For example:

http://cgi.ebay.com/Two-New-100MHz-Oscilloscope-Probes-Probe-Tektronix-HP-/160499354691?pt=LH_DefaultDomain_0&hash=item255e81cc43

I wrecked one of my Agilent probes last week and would like a replacement but I'm not going to pay $100+ for a new one. A lot of the cheaper probes appear to be of similar quality and have the same functionality. What is so great about the probes Agilent and Tektronix produces that makes them so expensive?

[baljemmett]

Has anyone ever used some of the cheapo oscilloscope probes that you can get off ebay? For example:

http://cgi.ebay.com/Two-New-100MHz-Oscilloscope-Probes-Probe-Tektronix-HP-/160499354691?pt=LH_DefaultDomain_0&hash=item255e81cc43

I've got a pair of what look like exactly the same thing -- they seem to work OK on my 20MHz analogue scope, at least.  More than that I can't say I'm afraid!

[alm]

Aberrations, circuit loading at high frequencies (flatness of the input impedance), mechanical quality. Tek/Agilent usually specify bandwidth at the probe tip, so the -3dB is including the standard probes. Even seemingly basic stuff like the input hybrid is complicated and proprietary technology (I believe Tek bought back the company that made them after spinning it off). Tek tunes the probe response to the front-end used in the scope, if the front-end has an impedance peak somewhere in the frequency band, the probe is designed to compensate for this. This is why they made various probes with almost identical specs. These effects tend to get magnified as you approach the bandwidth limits for Hi-Z passive probes.

Random example of what can be wrong in cheap probes: dielectric absorption or leakage in the input caps can lead to funny looking waveforms, which you may not attribute to the probe until you accidentally use a different probe. I've also had a cheap probe with a bum switch, even in ground mode, it would capacitively couple the signal. Safety might be an issue if you ever use scope probes on dangerous voltages. If the trimmer cap in the head is fails shorted due to high voltage, this may be dangerous for both the user and the scope (which suddenly sees the unattenuated signal).

If you're likely to kill them (I tend to treat them as delicate, expensive instruments, so I don't drop them or get the cord caught under my chair), a cheap(ish) one may be fine for simple/low-frequency use, just use a real probe for important measurements. In my experience, the cheap crap is not significantly cheaper than some used Tek probes, so I don't see the point.

[Time]

I think I have a pair of the probes that you show in that auction.  For 13 dollars, I thought I'd give them a whirl on my analog 40 MHz scope.  I got the scope for free and needed some cheap probes so I figured even if these crappy probes dont meet their 100 MHz spec they probably still have a better response than 40 MHz.  All I ever use that scope for is basic micro work and slow signal stuff when I run out of channels on my DSO.

[saturation]

Since they have no brand, I can't say all such Chinese cheapo probes are the same.  I bought a pair months ago too from ebay and they look identical, though.

For the money, you can't go wrong using them, but having function generator in hand, and if you have a better probe still around, characterize them and compare them, see if they live up to their specs.  I did them same to the limits of my Rigol, and they are good through at least 50 MHz.

[TopherTheME]

For the money, you can't go wrong using them, but having function generator in hand, and if you have a better probe still around, characterize them and compare them, see if they live up to their specs.  I did them same to the limits of my Rigol, and they are good through at least 50 MHz.

I think this is what I'll do. I do still have one good probe which I'll keep in a drawer for comparison.

[mikeselectricstuff]

I was disappointed with the quality of the probes that came with my £6K Agilent scope, as they were big and had rather stiff cables.  I much prefer the old Tek P6131's

which are very small and have very flexible cables that have less tendancy to pull the board off the bench when you have 4 probes hanging off it.

[qno]

I have used these probes and they are OK.

A bit flimsy switches,

The designers have looked to the older Tek probes carefully.

They last a long time but I have one that has a broken wire now.

Best choice for $13.

[alm]

For the money, you can't go wrong using them, but having function generator in hand, and if you have a better probe still around, characterize them and compare them, see if they live up to their specs.  I did them same to the limits of my Rigol, and they are good through at least 50 MHz.

Function generators are easy, they usually have an output impedance of 50 ohm (25 ohm if you terminate them properly). Real circuits usually have a much higher output impedance, which will make them more sensitive to probe loading. Function generators also usually can't supply signals with a high slew rate (large amplitude, fast rise/fall time). You also need a known good probe to compare aberrations like over/under shoot. Most people are not able to do these kinds of tests, this allows some companies to sell some really crappy probes (can't say about these particular probes, I've never used them, and even if I had, without brand you can't be sure that they're the same).

[saturation]

Thanks for the sage advice alm, and detailing the capabilities of more costly probes that aren't obvious unless fully tested characterized as say, a reputable scope manufacturer like Tektronix would do.  

Getting a used Tektronix or equivalent reputable 60-100MHz 10x passive probe isn't exactly $7 each.  Ebay typicals with clear manufacturer model numbers are $15-50 each delivered, via eBay, and about $50-200, new.  If you treat your probes with respect and delicacy, its anyone's guess how these eBay probes will perform without performance testing, given it could have been abused physically and/or electrically.

But your insight suggests to me that getting hold of a really known good probe as a reference and doing tests comparing the cheapo to it, to make it 'good probe' traceable, is prudent.  Once characterized I can compare future probes against referenced ones.

I'll revise a test suite with an impedance load to give the probes a workout modeled after Tek's more simpler to implement philosophy and expose what weakness Chinese cheapos maybe are hiding.

http://www.tek.com/Measurement/App_Notes/60_18324/eng/60W_18324_0.pdf

For the money, you can't go wrong using them, but having function generator in hand, and if you have a better probe still around, characterize them and compare them, see if they live up to their specs.  I did them same to the limits of my Rigol, and they are good through at least 50 MHz.

Function generators are easy, they usually have an output impedance of 50 ohm (25 ohm if you terminate them properly). Real circuits usually have a much higher output impedance, which will make them more sensitive to probe loading. Function generators also usually can't supply signals with a high slew rate (large amplitude, fast rise/fall time). You also need a known good probe to compare aberrations like over/under shoot. Most people are not able to do these kinds of tests, this allows some companies to sell some really crappy probes (can't say about these particular probes, I've never used them, and even if I had, without brand you can't be sure that they're the same).

 

Aberrations, circuit loading at high frequencies (flatness of the input impedance), mechanical quality...
...dielectric absorption or leakage in the input caps can lead to funny looking waveforms, ...Safety might be an issue if you ever use scope probes on dangerous voltages..I tend to treat them as delicate, expensive instruments, so I don't drop them or get the cord caught under my chair...a cheap(ish) one may be fine for simple/low-frequency use...the cheap crap is not significantly cheaper than some used Tek probes, so I don't see the point.

[gxti]

Those look identical to the probes I got with my Instek GDS-1062A. Guess I know where to find replacements, now. I've barely used it in the 4-5 months since I got the scope but the spring in one of the grabbers broke so it doesn't snap back correctly.

[alm]

Getting a used Tektronix or equivalent reputable 60-100MHz 10x passive probe isn't exactly $7 each.  Ebay typicals with clear manufacturer model numbers are $15-50 each delivered, via eBay, and about $50-200, new.

True, although I have gotten some like-new Tek probes for something like $15 each, which IMO is close enough to $7 to skip the cheap crap. Yes, as everything you buy from a non-authorized dealer without return policy, there is a risk that they're faulty, although the most common fault with probes appears to be damaged probe cables or tips, and this will easily show up with LF tests. Most low-frequency (<150MHz or so) probes don't have any adjustments except LF compensation to mess up either .

http://www.tek.com/Measurement/App_Notes/60_18324/eng/60W_18324_0.pdf

Good app note, thanks. Note that it does focus on multi-GHz active probes, at those frequencies there is no such thing as a high-impedance input. At low(er) frequences there is, which is why there are 10Mohm 100MHz probes, but no 10Mohm 8GHz probes. The stuff about how their philosophy is superior to Agilent is not so interesting, I think Agilent has a similar app note showing that their philosophy is better.

The Tek documents 'Oscilloscope probe circuits' and 'Oscilloscope probe measurements' (from the sixties) are probably also informative, passive probes are still basically the same (8Ghz active probes probably not so much), and they contained much more technical details compared to today's app notes which are written by marketing.

[Bored@Work]

The Tek documents 'Oscilloscope probe circuits' and 'Oscilloscope probe measurements' (from the sixties) are probably also informative, passive probes are still basically the same (8Ghz active probes probably not so much), and they contained much more technical details compared to today's app notes which are written by marketing.

Indeed. If you want to learn a little bit about today's GHz probes, you can look at Tek and Agilent patents in a patent archive (at the USPTO or some who republish them). Some fascinating stuff, e.g. http://www.freepatentsonline.com/7586318.pdf although they of course don't tell you every little dirty secret in those patents.

[saturation]

Comparing Probes.  Details later, but the simple test was revealing.  Since the speed changes are consistent it suggest the speed rating of the probes are close to its electrical performance.

NN = sold as no name eBay 100 MHz scope probes have names Yang Sun P6100 , $14 pair [ it has a name somewhere, will locate and revise post later].  Probe labels and documents suggests its CAT 1 and has a frequency to voltage derating curve. Rise time measurements are the tracing in yellow.

Rigol = RP 2200 [ shipped with 1052E ], 200 MHz rating, Rise time measurements are the tracing in blue.

Signal Sources, used for rise times
Instek = SFG 1003 , faster source, TTL output at least to 4.8ns
Hantek = DDS 3x25

Tektronix mentions the industry standard probe test uses a 1ns pulse with 50 ohms impedance.  See image.  Test probes were connected directly to the signal sources without adapters.  I don't have a 1ns pulse, so the Instek and Hantek substitute for it, for now.

The dot image shows the sample points, it suggests negligible aliasing or no erroneous interpolation created by an unmodded 50MHz 1052E.

[alm]

So the Rigol probe shows a faster rise time with the Instek FG, but slower with the Hantek AWG? The appears to be significant overshoot in the blue trace in the first picture, but this may also be the signal. 4.8ns shouldn't really be a hard test for these probes, it requires 70MHz or so of bandwidth, so the scope is likely to be limiting. At least in the first setup, the Rigol probe looks slightly better to me (less smooth), the opposite in the second. Note that this from a low-impedance source with a low amplitude, none are issues with Tek's current many-GHz probes, they're low-impedance inputs anyway, and won't survive a high amplitude. The scope also appears to sample too slow to really see aberrations, you should probably test one probe at a time so you can use the full 1GS/s.

[saturation]

Yes,  since I don't have a calibrated traceable reference for rise times or pulses, I have to deduce the performance. 

The simple test shows that an identical scope+probe setup produces different rise times from 2 different sources, one faster than another.  Changing the probe causes a proportional change, with the sources consistent, Instek is faster than Hantek.

One can say with certainty that the RP2200, rated at 200 MHz, is faster than the Yang Sun rated at 100 MHz.  But how fast they really are quantitatively, I can't say.  Since the Rigol probe provides a faster rise time vs the Sun probe, this suggests that the Instek source rise time is at <= 4.8ns, it could be better, but even if it faster than 1.8ns needed to test the 200 MHz probe to its limit, I then am limited by the scope's rise time.

The question raised by the OP, is are these 'low cost' eBay probes any good?

Using the Instek rise time 4.8ns for the Rigol probe, and 6.6ns for the Sun probe, and using 0.35/rise time estimates, gives applicable bandwidth between 73MHz and 53 MHz.  Since my scope is only rated to 50 MHz or slightly more, either probe is suitable to 50 MHz.

In addition to the rise time, you can also do a brute force sweep of sine wave frequency to check the -3dB limits of the probe+scope combination, and finally a qualitative appearance of a square wave distortion to complement all the measurements.

It would be good to have a reference probe and signal source traceable to a calibrated device, but I leave that task for another day.

On the sampling rate, yes, I checked each response at 1Gs/s when doing measurements, to insure I get the best results.  For these photos, I combined the images mostly to avoid posting so many individual photos.  I also use dot mode to confirm the scopes interpolation, and use vector to make the photos.

On the overshoot, yes, I could try to correct it, but I think it will add little more to determining whether these eBay probes are adequate for a 50 MHz DSO.


In summary, the probes don't limit the scope's performance, and the limiting step now is the scope.  If a user has a fast rise time square wave source, they can easily check the performance of a no-name passive probe by simply connecting the probes directly to the signal source, and measuring its rise time on the scope its to be used on.  The key is to insure it will allow the scope to perform at its rated bandwidth.

So the Rigol probe shows a faster rise time with the Instek FG, but slower with the Hantek AWG? The appears to be significant overshoot in the blue trace in the first picture, but this may also be the signal. 4.8ns shouldn't really be a hard test for these probes, it requires 70MHz or so of bandwidth, so the scope is likely to be limiting. At least in the first setup, the Rigol probe looks slightly better to me (less smooth), the opposite in the second. Note that this from a low-impedance source with a low amplitude, none are issues with Tek's current many-GHz probes, they're low-impedance inputs anyway, and won't survive a high amplitude. The scope also appears to sample too slow to really see aberrations, you should probably test one probe at a time so you can use the full 1GS/s.

[saturation]

Yes, what interested me more in that link is that Tek doesn't believe in measuring probe performance using a modeled load as Agilent does [which is what I was intending to slap together], and spends a good portion of the pdf discussing why; this reduces to the equivalent circuit I posted as an embodiment of their philosophy for characterizing probes.  Thus, the tests the OP or others can use can be the Tek model: its a simpler way to confirm the bandwidth of those nameless eBay passive probes, with validity.

I found those classic Tek documents and read them through as you suggested, and they are great refreshers, thank you!  The newer documents carry over a lot of that, but they touch on why there are so many probes, and how to focus your need: passive, active, differential, current, etc., and are the accessories worth it.  Many of those probes cost far more than the Rigol alone!

The Tek documents I read summarize 4 things worth remembering when buying passive probes:

500 MHz is a practical upper limit for passive probes, beyond which will likely require active probes, so that's Tek's upper limit for passive probes in their catalog.  Its a long story, its detailed in "ABC of Probes."

Probe+ scope should be rated 3-5x more than the bandwidth to reduce amplitude error;  at the -3dB mark, there will be an amplitude error ~ 30% from the roll off.  

Use the shortest shortest shortest shortest ground leads possible.

As always, beware the loading effects of probe capacitance and cable length.

True, although I have gotten some like-new Tek probes for something like $15 each, which IMO is close enough to $7 to skip the cheap crap. Yes, as everything you buy from a non-authorized dealer without return policy, there is a risk that they're faulty, although the most common fault with probes appears to be damaged probe cables or tips, and this will easily show up with LF tests. Most low-frequency (<150MHz or so) probes don't have any adjustments except LF compensation to mess up either .

http://www.tek.com/Measurement/App_Notes/60_18324/eng/60W_18324_0.pdf

Good app note, thanks. Note that it does focus on multi-GHz active probes, at those frequencies there is no such thing as a high-impedance input. At low(er) frequences there is, which is why there are 10Mohm 100MHz probes, but no 10Mohm 8GHz probes. The stuff about how their philosophy is superior to Agilent is not so interesting, I think Agilent has a similar app note showing that their philosophy is better.

The Tek documents 'Oscilloscope probe circuits' and 'Oscilloscope probe measurements' (from the sixties) are probably also informative, passive probes are still basically the same (8Ghz active probes probably not so much), and they contained much more technical details compared to today's app notes which are written by marketing.

[jahonen]

Test probes were connected directly to the signal sources without adapters.

Can you elaborate a bit more how the grounding was done, hopefully you did not use the long ground clip? I have always believed that when one characterizes the probe, BNC adapter (or similar) is usually used, which provides 360° ground right at the probe tip, minimizing the excess inductance.

For comparison, see my comparison between two ground connection styles.

Also, it does not make any sense to make very high-bandwidth 10M passive probe because capacitive loading gets way too high. For typical 10 pF probe, probe impedance drops to 160 ohms @ 100 MHz and 32 ohms @ 500 MHz. That is comparable to transmission line impedance in digital applications (something like 80-40 ohms). Anyway, 10 M input impedance makes sense only at DC. Even at 20 kHz, loading due to capacitance has dropped down to 800 k.

Regards,
Janne

[tinhead]

you can build your own cheap active FET probes, they are good enough up to 500Mhz.

http://welecw2000a.sourceforge.net/docs/Hardware/Aktiver_Tastkopf_mit_OPA659.pdf

I know the pdf is not in english, but in principle all you need to know is the schematics,
the PCB cab be ordered by the developer for few USD.

[Chasm]

It's a nice active DIY probe, the only thing missing is a scope that would warrant it for me.

Several measurements were taken during the development iterations of this probe.
The only thing still missing is AFAIK to hit it with some really high end measuring equipment.

[tinhead]

they are comparable with P6205 on a TDS5104B, so good enough for me.

[saturation]

Hi jahonen:

I use the spring ground clip on the Rigol's and compared the output on the 'long' alligator ground clips [its about 10cm long].  I also probed directly without the spring into the BNC adapter, making contact between the probe tip ground ring and the ground socket of the BNC adapter while probing inside the male socket.   It didn't make a difference, but I did it only once.  Then I just used the alligators clips clipped to the ground ring of the BNC jack for convenience.

I didn't try at all for the Sun probes, hmmm, I'll try again, it doesn't take much time or effort.

Test probes were connected directly to the signal sources without adapters.

Can you elaborate a bit more how the grounding was done, hopefully you did not use the long ground clip? I have always believed that when one characterizes the probe, BNC adapter (or similar) is usually used, which provides 360° ground right at the probe tip, minimizing the excess inductance.

For comparison, see my comparison between two ground connection styles.

Also, it does not make any sense to make very high-bandwidth 10M passive probe because capacitive loading gets way too high. For typical 10 pF probe, probe impedance drops to 160 ohms @ 100 MHz and 32 ohms @ 500 MHz. That is comparable to transmission line impedance in digital applications (something like 80-40 ohms). Anyway, 10 M input impedance makes sense only at DC. Even at 20 kHz, loading due to capacitance has dropped down to 800 k.

Regards,
Janne

[saturation]

Thanks Tinhead, have given that consideration too.  When the loading becomes too much, this is the only alternative.

FWIW too, the $7 100 MHz no-name Chinese probes are most cost effective.  The 200 MHz no-name probes are $25-30 each, and probes below 100 MHz are about $5-6 each.

you can build your own cheap active FET probes, they are good enough up to 500Mhz.

http://welecw2000a.sourceforge.net/docs/Hardware/Aktiver_Tastkopf_mit_OPA659.pdf

I know the pdf is not in english, but in principle all you need to know is the schematics,
the PCB cab be ordered by the developer for few USD.

[saturation]

I just repeated some measurements with the spring clip on both the Sun and Rigol probes, all 4 of them, they ship 2 pair in a box.  I also checked the spring clip's effect on a Velleman probe, rated at 60 MHz.

Examining it more intensely, there is a 1m 400mV reduction in the over shoot, and minor ringing in now flattened out.  The rise time is interestingly longer by 0.5ns on all probes compared to using the ground clip.  Very odd result, but the signal quality is improved, but not much.  Will post photos when able.

[FreeThinker]

I've just received a pair of these Chinese probes made by Yang Xun 100Mhz, 600v CatI type P6100. Cost £8 (UK) inc p&p. Would I us them on 600v? No way but for general use on my 50meg analogue scope you can't be robbed. For absolute precision probably not much cop but for run of the mill work a no brainer and remember the old saying "You can never have too many probes". Put the money saved towards buying something useful