Recommendations for active differential scope probe for small signals

[euzer]

After recommendations for a 'reasonable' active differential scope probe targeted for small signals, like USB, lower cost alternative to Agilent (N2750A, N2818A, 1130B). Not really interested in the probes aimed at high voltages.

Requirements: 200MHz, small test clip set, preferably integrated tweezer type.
Attenuation in the range2:1, 5:1, 10:1

I don't expect any third party probes to be compatible with the additional connections of the Agilent 3000x probe to make it self powered, so will presumably be battery powered.

Thanks.

[LaurentR]

Cal Test CT3688
http://www.caltestelectronics.com/ctitem/21-probe-oscilloscope-differential/CT3688

That's a rebranded version of Sapphire SI-200, which is so far their fastest and lowest-signal design (they are the main high voltage diff probe vendor, including for Keysight).

It seems to be similar to the Keysight N2792A and N2818A (~$2k), which are probably the same Sapphire design customized for Keysight.

I got one recently for that purpose. I have a post here:
https://www.eevblog.com/forum/testgear/eevblog-hvp70-differential-probe/msg1503736/#msg1503736
There are a couple of other posts about it on the forum.

It's officially $989 but for an unexplained reason, I got mine on Amazon US (shipped and sold by Amazon) for $560.

If you are not in the US, you might find it under some other brand like the Testec TT-SI 200 in Europe.

[georges80]

TEKTRONIX P6243 is a decent probe (1GHz) and can be found for a couple of hundred on ebay. Get one that has the full probe tip 'bag of bits' to give you various probing options. Not differential though.

Add an tektronix 1103 'power supply' and you can use the probe system with any scope.

Total would be in the $400 - $500 range depending on your luck.

Tektronix P6246 is a differential probe (I have one, but scored it for free), more $ but good performance.

This will give you a probe system that has minimal capacitance to ensure you don't upset your high freq measurements.

cheers,
george.

[Mechatrommer]

$212 used...
Agilent 1141A 200MHz Differential Probe
Agilent 1142A Power Module

another $60 if needed...
10X Attenuator for Agilent 1141A

the lost pins can easily be made. if you want cheaper you just have to be patient... i got mine at less than $200 some years back, i only need the 10X attenuator to complete the collection, not really that important i just need the right moment, $60 is still a little bit too expensive for me. i never tried it with USB signal though, its just basically decoration in the box so far... you may find cheaper deal shipped to your place. the link above is only what shipable to my country, ymmv...

[LaurentR]

Nice info on the Tek and Agilent probes, thanks. These are bulky with the power supplies but used, are very decently priced for the specs and quality.

Anyone knows the noise characteristics of these?
Input equivalent noise of the CT3688 at 10X is 3mVrms quoted (2.6mV measured). I looked up other likely-Sapphire-based Keysight probes like the N2792/N2793 and they quote 6mVrms and 4.7mVrms, which is not as good (but close).

I searched for that info for the Agilent 1141A and didn't find much except for that one line: "Displayed noises: ≤ 50 µVrms". Not sure what that corresponds to. If it's output noise, that's much lower than the other ones at 1X or even at 10X.

The Tek P6246 says "<50 nV/squareroot Hz". Not sure how to read this. Does that mean 50nV*20k (@400MHz) = 1mV (output I assume)?

[mk_]

After recommendations for a 'reasonable' active differential scope probe targeted for small signals, like USB, lower cost alternative to Agilent (N2750A, N2818A, 1130B). Not really interested in the probes aimed at high voltages.

Requirements: 200MHz, small test clip set, preferably integrated tweezer type.
Attenuation in the range2:1, 5:1, 10:1

I don't expect any third party probes to be compatible with the additional connections of the Agilent 3000x probe to make it self powered, so will presumably be battery powered.

Thanks.

DIY:

https://www.mikrocontroller.net/topic/382504 (well, the schmatics are readable even if you don`t know german very much)

final version start at https://www.mikrocontroller.net/topic/382504#4442662 with some diagrams some posts below

[Mechatrommer]

Nice info on the Tek and Agilent probes, thanks. These are bulky with the power supplies but used, are very decently priced for the specs and quality.
Anyone knows the noise characteristics of these?

if you are talking about 1141. attached are the noise figures for my 1141 at different 1X, 10X, 100X (i dont have 10X attn but i bet my bottom dollar on that), is it within spec or outside? i'm not sure i bought it damaged (well not really, the output coax is just disconnected when i got it, all i did is resolder it).

±10mVpp @ 1X, ±100mVpp @ 10X, ±1Vpp @ 100X. you see, its just multiplication of each other. noise figure at no (1X) attenuation is just = 1/10X noise figure at 10X attenuation, and noise figure at 100X attenuation is just = 100X of noise figure at no attenuation etc. because simply the DSO just scale it in FW according to probe attenuation setting in the DSO. where the diff probe is outputting the same noise figure regardless of attenuation used. the 4th attachment is i tried change the scale based on your test in the another thread at no attenuation of 1141.. i can say 1141 is a winner if i want to (below is comparison i added my 1411 result on top) but we cant because its not apple to apple comparison... (my test is not in a clean room so there sometime larger or lower noise going on, and shorting the probe doesnt make any difference, so i left the probe disconnected)




to make apple to apple comparison we need to see noise figure at the same attenuation level (or precisely at the same max diff input, or the same common mode) imho, because different probe is for different purpose. although CT3688 is the least noise figure, but it cannot do mains work (±100V - ±300V++ common mode), it might be suitable for the OP though since only probing USB.

for other probe like EEVBlog, we can use for mains work but at X100 attenuation, meaning 10X the noise figure in your chart (±0.25Vpp from eyeballing it). the same applied to DP-25. both probe will be suitable for large differential signal probing on mains. and it also applicable to the OP, if all he want is differential signal of the USB (±5V), noise will be lowered since he will set the DSO at high V/div anyway. but to probe low diff signal at high (mains) common mode such as high side BJT's Vbe in range of 0.3V, all three will not do the work nicely. CT3688 will exceed its rating, EEVBlog and DP-25 signal will be swarmed by larger noise figure esp Pintek DP-25. 1141 will be nicer in this regards (at 10X attenuation, ±3V max differential input ±500V common mode, noise of 1141 is ±0.1Vpp), but for OP purpose, he has to use 100X attenuator with 1141 to see the ±5V signal, but at larger noise ±1Vpp. noise will be noticable on the DSO but averaging may help, its just when the intended diff signal very close to noise floor, its more difficult even with averaging. otoh if you need larger than ±30V differential input, than 1141 is out of the league. ymmv.

edit: sorry i forgot to mention the spec for 1141 (refer to 5th pic for input voltage spec page 55 from https://cdn.testequity.com/documents/pdf/1142A-ug.pdf):

no attenuation: max diff signal is ±0.3V, max common mode Vdc is ±200V
10X attenuation: max diff signal is ±3V, max common mode Vdc is ±500V
100X attenuation: max diff signal is ±30V, max common mode Vdc is ±500V

and another think... other spec like input impedance also need to be considered, CT3688 has the lowest among all, so less ideal for high impedance node probing. i think why CT3688 can get lowest noise is because it doesnt have adjustable attenuator, so the designer can strive for lowest noise signal possible from input to output, other probe may have exposed big switch, so difficult to get the noise lower, ymmv. fwiw...

[David Hess]

The Tek P6246 says "<50 nV/squareroot Hz". Not sure how to read this. Does that mean 50nV*20k (@400MHz) = 1mV (output I assume)?

That is exactly it; total noise is proportional to the square root of the bandwidth.  To compare the probes, the total noise should be corrected for bandwidth.

The CT3688 specifications say 0.3mVrms over 200MHz so 21nV/(Hz^-2) which is consistent with a low voltage differential probe that lacks input attenuation and comparable to the Tektronix P6246.