Oscilloscope probes

[Howardlong]

Folks

I recently bought some alleged 500MHz scope probes, but when I plugged them in they definitely weren't anywhere near that. As they looked similar to a couple of other probe types I have, I thought I'd see how they compared quantitatively.

I did two tests on each probe.

The first test was with a Jim Williams pulse generator with a 50cm coax pulse stretcher, terminated with a 50 ohm SMA 20dB pad, and the probing was done with the probe tip and ground spring directly on the SMA pad.

The second test was with an HP 8656B RF signal generator, with the frequency adjusted for the 3dB point. The probing was done with the ground spring and probe tip directly on an inline 50 ohm BNC terminator at the end of a 6' 50 ohm patch cable.

All probes were compensated on the same scope channel. The scope used is an Agilent 54831D (600MHz bandwidth).



Pulse results


50 ohm internal termination


50 ohm external termination


Agilent 1165A


Texas TX5230 - ~$20, purchased in Akihabara shop, same as this: http://hkaco.en.ec21.com/Texas_Passive_Oscilloscope_Probe--3475682_3476276.html


Barsoom P6300 - ~$40, http://www.amazon.co.uk/gp/product/B00DRABRKM?psc=1&redirect=true&ref_=oh_aui_detailpage_o08_s00


TP6100 - ~$15, http://www.amazon.co.uk/gp/product/B00GHIHW0K?psc=1&redirect=true&ref_=oh_aui_detailpage_o08_s01


HQRP T5100 - ~$30, http://www.amazon.co.uk/gp/product/B005DQ2F1I?psc=1&redirect=true&ref_=oh_aui_detailpage_o00_s00


LIHUA P6500 - ~$30, http://www.tmart.co.uk/LIHUA-P6500-High-Sensitivity-Oscilloscope-Probe-500MHz-Gray_p284939.html


Agilent 10073C


Maplin HP-2060 - ~$20 http://www.maplin.co.uk/p/60mhz-oscilloscope-probe-aq58n

So the biggest disappointment is the LIHUA P6500 that I originally suspected. It looks suspiciously identical to the TP6100 both physically and with regards performance: just look at the pulse profile. Physically, those two also look identical to the HQRP T5100 that performed really pretty well. I'll let others draw their own conclusions, noting that the Barsoom P6300, HQRP T5100 and Maplin HP-2060 all experienced resonance and therefore peaking to some degree between DC and the 3dB point.

[tggzzz]

Nice article, which I've bookmarked and passed around to friends that might not realise how crappy suboptimal probes can send you off on wild goose chases.

[nctnico]

There is something off in the table. If I calculate the bandwidth based on the risetimes I get different bandwidths. For example the Agilent 1165A only has a bandwidth of 0.35/951ps=368MHz.

Anyway, a much better rule of thumb is not to use a high impedance passive probe on frequencies over 100MHz because the capacitance screws things up badly. The bandwidth of such a probe is specified using a 25 Ohm source. Any other source (IOW a real life circuit) will give anything but the right results. Measuring high frequency signals calls for low-Z passive divider probes or active FET probes.

[Howardlong]

I would take the risetimes calculated by the scope with a punch of salt, that is why I presented the screen shots too.

The point of this was not tell everyone to go out and buy $1,000 fet probes, by the way!

[nctnico]

If you do the math (loading a high frequency signal with 10pf) you realize you have to spend the cash or use a DIY solution: http://www.emcesd.com/1ghzprob.htm

[Howardlong]

Yes, I have indeed already built two of those fet probes, and a resistive one as in your link, but thank you anyway.

To be clear, the point of the test was to make folks aware of the trash probes that are around.

The first test was done with a 50 ohm source, the second, if you read the description of my test,, presents 25 ohms.

[Pjotr]

To be clear, the point of the test was to make folks aware of the trash probes that are around.

Luckily they do not cost you a fortune

Going over 100MHz requires decent quality probes and preferable fixed 1:10 ones (not switch-able). These are never cheap but you do not need to dig up Tek money. Good high impedance HF probes have complex compensating networks with 2 or 3 adjustment points.

[Howardlong]

To be clear, the point of the test was to make folks aware of the trash probes that are around.

Luckily they do not cost you a fortune

Going over 100MHz requires decent quality probes and preferable fixed 1:10 ones (not switch-able). These are never cheap but you do not need to dig up Tek money. Good high impedance HF probes have complex compensating networks with 2 or 3 adjustment points.

I'd agree with that, the various Tek probes I have that are rated 400MHz and above (not in this test) have several adjustment points but some are covered up by a self adhesive label. However, the half dozen Agilent probes I have just have the standard compensation adjustment. They are all fixed 10:1 probes.

The cheapo 300MHz Texas probes I have have readily available adjustment points. I got lucky on those, I happened to notice them in a store in Akihabara about six weeks ago, bought four, should've bought more, about $20 each, including ground springs and BNC probe adapters. Maybe time to go back again ;-)

[Pjotr]

That's really a bargain if they perform as well as the $200 Tek probes. My point with the cheap probes, apart from electrical performance, is they fall apart after half a year of use.

I myself use the 300MHz Testec TT-HF212 probes. Costing €45 here in Europe these are of reasonable quality and have a separate HF adjuster.

[Howardlong]

Yeah, but I probably blew about $50 of roaming charges looking them up while I was in the store!

[markce]

+1 for the 300MHz Testec TT-HF212 probes.
3 adjustments, but nice pulse response. Fixed 1:10.
Nice good handling and convincing build quality.

[Howardlong]

My point with the cheap probes, apart from electrical performance, is they fall apart after half a year of use.

I think any probes will if not treated well: I have a good collection of Tek probes in various states of disassembly that I've accumulated over the years from various scope purchases. I guess because I work on my own stuff with my own equipment, I take care of them. All of the probes I have here that I've had from new are since I ever owned my first scope twenty or so years ago, all are perfectly functional, with all the accessories. I've had a couple of minor glitches, for example a ground clip getting mangled in a drawer and a dual lead coax adapter suffering a similar fate, but I either repair them myself or buy a replacement right away. As a matter of fact, both of those were Agilent probes, but they are the ones I use most.

I can see why a scope that's in a university or used in a factory environment that's continually brought in and out of stores might suffer a worse fate though. If they ain't on your account, then less care is going to be taken, simple as that.

[nctnico]

In my experience probes start to suffer from bad contact between the ground clip and broken conductors inside the cable at some point. Even if you are careful with your probes the cable does get a lot of bending. When I suspect a probe is flaky it goes into the bin. Using a bad probe is a real nuisance.

[macboy]

It is my experience that all probes need proper high frequency compensation to be performed in order to match them to the particular scope. With a random probe+scope combination, you can expect neither flat frequency response, nor flat pulse response.

[Pjotr]

That is only true for a limited range of scopes/probes. Probes are specified over a range of input capacitance, usually 10pf - 30pf and some up tot 50 pF. If your scope input capacitance is within that range there is no problem with interchangeability. Problem can be sometimes dedicated signalling pins around the scope's input BNC.

[macboy]

That is only true for a limited range of scopes/probes. Probes are specified over a range of input capacitance, usually 10pf - 30pf and some up tot 50 pF. If your scope input capacitance is within that range there is no problem with interchangeability. Problem can be sometimes dedicated signalling pins around the scope's input BNC.

Within that range, you will be able to adjust the LF compensation properly, but this does not account for high freqency effects. All scope inputs have different parasitics, and the high frequency compensation adjustments on good HF probes are there to adjust the response of the scope+probe combination.

On my cheaper probes (Velleman probe250) these adjustments are easily accessible through holes on the compensation box at the BNC connector. On my Tek probes, I usually need to physically dismantle the compensation box to access these trimmers. Some images I found to illustrate:



The basic idea is to feed a fast, flat top pulse (500 ps or faster for 500 MHz scope/probes systems) into the probe and adjust the trimmers for fast rise and minimal overshoot. This adjustment is not the same as the LF compensation that everyone is accustomed to. But just like the LF compensation, a probe that is tweaked perfectly for one scope will be off when plugged into another one. Sometimes the difference between two channels on the same scope is enough to warrant having separate probes tweaked specifically for each channel. Manuals for passive probes from Tek, HP/Agilent/Keysight, Lecroy, etc., all describe these adjustments.

[Pjotr]

Oh, in that way. Looks standard practice to me to adjust your probes to the particular scope used with it. Or even the channel on that scope. When you wrote "a random probe+scope combination", I did read probes and scopes of different brands.

[Dexter2]

Howardlong made an excellent post. A shame I did not see it sooner. I needed some cheap probes and bought the P6500 Chinese probes. Build quality is quite good as they are a copy from the Tektronix 6139B probes, but performance it terrible. Practically the probe has the -3db point below 20Mhz so it represents a square wave signal nicely only to a few 10Khz.

Image of the P6500 Probe:


I measured the frequency response of the probe and this is the result. While it is possible to resolve signals up to 500Mhz the frequencies over 20Mhz are attenuated heavily (compared to quality 500Mhz pasive Tektronix probes). No wonder Tek probes cost so much, even used ones... Freqency response was measured using a IFR 2025 2.5Ghz signal generator set to +13db and a MDO3054 scope (Y-axis are Vrms values).


"Perfect" 10Mhz signal measured:


Compared to Tektronix TPP0500:


Just a reminder to anyone not to make the same mistake as i did.

[jjoonathan]

Are you sure about your setup? I get different results.

If you are sure about your setup, we might be dealing with bad clones of good clones. Lol.

[Kosmic]

Did the same test with mine (P6500 probe) but with a noise source from DC to 500Mhz. The noise source was 50Ohms terminated and the probe was connected with a probe adapter to the bnc connector.

DC to 500Mhz, 50Mhz per division. In white, the noise source and in green, the frequency response of the probe (P6500).


Look like it start dropping at around 450Mhz.

Also tested the rise time with the calibration output of my lecroy oscilloscope. <= 1ns Rise, 25ns Width, 1V amplitude.



I had to adjust compensation to get those results though. It's really not obvious but 4 trimmers are hidden in the plastic shell of the base of the probe.

[bitseeker]

Looks good. Maybe there are bad clones of good P6500 clones.

[jjoonathan]

I think I accidentally solved the mystery: the probe -> BNC adapter is very easy to partially unseat, even after seating it as well as I can. A tiny bump is enough and the partially unseated state isn't visually obvious. 

Video: https://photos.app.goo.gl/H3d7Kci8NdBfjQW18

[jjoonathan]

Here's a PMK 500MHz probe for comparison. The differences in risetime and loading are unremarkable, but the PMK BNC adapter is very difficult to unseat in a visually non-obvious way, even when I actually try.

Video: https://photos.app.goo.gl/avtGw9bkPqDUBUNw9

[bitseeker]

Good catch.

[jonpaul]

Hi there, I am a retired EE, have been collecting, studying and using probes since 1970s, including FET, differential, current and high voltage probes.

I think that TEK probes are far superior to HP, Lecroy, etc. The Yokogawa are a close second.
The cheap Chinese knockoffs are a waste of time and money!

Most abberations can be traced to a long  ground return path. Bayonet and Micro coax adaptateurs help that issue.

For Best BW and pulse response on 50 ohm scopes and plug-ins, use  a Zo probe. Very low capacitance 1...1.5 pF and 500 ohm résistance in a 10X. BW on Zo probes can be 1-6 GHz!
We use tek P6156, but be careful in buying off ebay as they must be complet and are easy to damage!

https://cal.equipment/doc/Tektronix/TEK%20P6156%20Instruction.pdf

The active probes FET are of limited use as they need power, are easy to damage and bulky.

For everyday use on 400 MHz scopes and 1 Mohm input,  eg 2467B, we use the TEK recommended P6137, 10x.

https://download.tek.com/manual/070643203.pdf

I humbly suggest to new probers, see the Tek concept book series of 1980s
Oscilloscope Probe Circuits, a really excellent overview.

https://www.davmar.org/TE/TekConcepts/TekProbeCircuits.pdf

Finally, Here's a modern Tek note on probes
https://download.tek.com/document/60W_14232_9_MR_Letter.pdf

Happy Probing!

Bon journée

Jon (in Paris)