Response of cheap RF Active Probe from Ebay.

[Insatman]

Greetings...I bought one of those cheap RF active probes on Ebay and received it recently from Kiev, Ukraine.  "RF Active Probe 0.1-1500 MHz-1.5 GHz analyzer oscilloscope, RF cable included"   It comes with a short SMA cable about 20-30cm.   I made some simple mods before testing it.   None should have affected frequency response significantly.   I added a ground pogo plunger about 6mm away from the main probe tip.   I added a 22uF filter capacitor on the open pads near the DC supply input.  I added an LED set for about 1mA using a 9V battery as the supply.   This is to remind me to shut off the battery when finished using the probe.  The battery was attached using about 1 meter of small silicon twisted wires.   The SMA cable was changed out for a 1 meter cable of the same type.  A simple enclosure was 3D printed in PLA to protect the probe and make it easier to use. 

Tests were done on a spectrum analyzer / tracking generator from 100kHz to 1Ghz.   It gets bad after 1 GHz, so I didn't keep the data.

Pulse response was testing using one of Leo's 35ps Trise pulsers and displaying probe output on a TEK784D in sampling mode.

The unit does not meet the promised spec of 100kHz to 1.5Ghz -20dB  +/-2.5dB.   Not even close.  The unit is supposed to be -20dB nominal but varies from -4.5 to -16.7dB over the frequencies tested.
It overshoots badly on the step-pulse test, but is fairly flat after 2ns or so.   

It is still useful however and for $18 including shipping I'm not complaining really.   

I suppose the longer cable and other mods could have affected the >1Ghz response

[rhb]

Thanks for posting this. 

Have you considered adding a "giimick" capacitor?  A short piece of coax across the input trimmed to knock down the hump/overshoot?  Glue it down once it's trimmed to length.  I've seen this in some posts and videos about cheap RF directional couplers.  The better ones had a gimmick added to improve the top end.

For $18 it's worth a bit of fiddle.  I've ordered one of the Ukrainian directional couplers I think by  the same outfit.  Now I'll need to order something else.

[Mechatrommer]

Pulse response was testing using one of Leo's 35ps Trise pulsers and displaying probe output on a TEK784D in sampling mode.

you are testing it with 14GHz BW pulse there its 2cm wavelength your pogo pin probe is about that length already try probing at the very fet tip and see if the ringing improved.

[edpalmer42]

I looked at those probes, but they didn't make sense to me.  How can you talk about a GHz probe that doesn't have any type of ground pin??    That suggested that the designer/vendor doesn't really understand what he's doing - or I don't.

It also doesn't inspire confidence when the website that's printed on the board doesn't exist.

Ed

[Gyro]

I was browsing last night.... All three listings (there are three different PCBs from Ukraine) indicate that the user needs to add an RF ground probe and has a bare area of ground copper on the back of the PCB for attaching it.

[edpalmer42]

I was browsing last night.... All three listings (there are three different PCBs from Ukraine) indicate that the user needs to add an RF ground probe and has a bare area of ground copper on the back of the PCB for attaching it.

Now that you point that out, I can't remember which type of probe I looked at.  They've been on ebay for a while now.  But I see that there's another style shown in the completed auctions that also has no ground pin, doesn't have an obvious place to connect one, and doesn't mention the need for one.

Does anyone know whether these probes pick up external RF interference since they're not shielded?

Ed

[rhb]

[
Does anyone know whether these probes pick up external RF interference since they're not shielded?

That's what friends who smoke cigars are for.  Nice aluminum tubes for DIY probes.

[2N3055]

Copper tubes are good for this..

[Insatman]

Further testing on the RF probe.  Based on one comment I thought it best to use a slower rise pulse for my tests, in this case ~600ps rather than the 30ps previously used.  The pulser selected was an avalanche/charge-line pulser (homebrew).  The output was attenuated to ~4.5V and fed into a BNC "T", one side of which was routed to the scope directly through a short cable and the other was used for the probe input.  See pictures below.   You can still see some peaking on the probe output waveform even though the risetime is lower than the direct pulse.   Longer timescales revel that the direct pulse has a downward slope while the probe shows and upward slope.  Note this slope is caused by the probe and/or setup because normally the pulser has a very flat-top.  Another picture shows how the input pulse is loaded when power is removed from the RF probe.  Last waveform is a short ~3ns pulse where I attempt to overlay the RF probe and direct waveform using voltage scaling and time-shift.   I tried loading the probe with 1pf to capacitance to stop the peaking but this impacted the direct waveform too much to be considered.    During these tests the delicate tip of the RF probe cracked and I replaced it with a Pogo spring-loaded contact.  No real change in RF probe performance was observed after the retrofit.

 

[Mechatrommer]

You can still see some peaking on the probe output waveform even though the risetime is lower than the direct pulse. Longer timescales revel that the direct pulse has a downward slope...

maybe that explained it because now this one...


and this one...


look somewhat tally, ie peaking respond at 670MHz. another possibly explanation of your downward pulse is due to downward freq respond of the probe at lower freq. i dont quite like the probe because it has non flat respond there, ymmv...

[Insatman]

You can still see some peaking on the probe output waveform even though the risetime is lower than the direct pulse. Longer timescales revel that the direct pulse has a downward slope...

look somewhat tally, ie peaking respond at 670MHz. another possibly explanation of your downward pulse is due to downward freq respond of the probe at lower freq. i dont quite like the probe because it has non flat respond there, ymmv...

This probe is definitely not ideal.  I was/am looking for a cheap acceptable active probe.  As you know such a probe from Keysight or Tek will cost more than my scope is worth!   Maybe someone here has had better luck with some other cheap probe from another vendor?   Anyway, next experiment is to build one of those passive RF probes and see what I get.

[medical-nerd]

Hiya

Many thanks for this - I was seriously looking at these but do not have the equipment to characterise them so was relying on them to work at the higher frequencies.

Cheers

[rhb]

I'm a big fan of step tests as it tells a lot with very little equipment.  I bought one of Leo Bodnar's 40 pS pulsers and love it.

But I'm not sure that the shape of the pulse response matters that much for an RF probe.  Obviously there are some issues with measurement accuracy, but my notion of use cases for an RF probe are more in the line of probing sinusoidal signals in  the front end of a receiver.

I'd be quite interested to know of  a pulse use case for an RF probe.

[jjoonathan]

Even tektronix has been known to juice the risetimes of their probes with some extra gain towards the end of their passband 

[Insatman]

I'm a big fan of step tests as it tells a lot with very little equipment.  I bought one of Leo Bodnar's 40 pS pulsers and love it.

But I'm not sure that the shape of the pulse response matters that much for an RF probe.  Obviously there are some issues with measurement accuracy, but my notion of use cases for an RF probe are more in the line of probing sinusoidal signals in  the front end of a receiver.

I'd be quite interested to know of  a pulse use case for an RF probe.

I have one of Leo's pulsers as well.  Very nice piece of engineering.  Only a volt or so output however.   

As for the case of pulse use for an RF probe.   I have always worked with pulsed circuits and I wanted a probe capable of driving 50 ohms and hopefully with a bandwidth superior to my existing 500Mhz 1M probes.  Preferably a probe with very low capacitance and high resistance...all of which this Ebay seller promised.   Basically I'm looking for a probe that can record a pulse with a risetime well under 1ns and not ring too much on the flat-top.   

[rhb]

Very interesting.  Got a design for a PA for Leo's pulser?  That would be really nice for checking the individual  attenuator steps.

I'm trying to come up to speed on FPGAs so that I can reprogram a COTS Zynq based DSO to suit my requirements.  I'm still waiting on kit to arrive, but have been reading a lot of Xilinx documentation while I wait.  I'm supposed to get a bunch of stuff in the post today.

One thing I think would be extremely valuable would be a  digital filter to correct a  probe response using Leo's pulser or similar as the reference input,  computing the filter using the ARM and then loading that into the FPGA and routing the ADC output through the filter.  The step responses on my two Instek DSOs are  pretty wretched judged against a well calibrated Tek.

I spent my career in reflection seismology where  correcting for non-ideal source signatures  and input amplifier and anti-alias filter impulse responses is SOP.  The only problem with doing this in a COTS DSO is having enough real estate left in the FPGA.

[Insatman]

Very interesting.  Got a design for a PA for Leo's pulser?  That would be really nice for checking the individual  attenuator steps.

I'm trying to come up to speed on FPGAs so that I can reprogram a COTS Zynq based DSO to suit my requirements.  I'm still waiting on kit to arrive, but have been reading a lot of Xilinx documentation while I wait.  I'm supposed to get a bunch of stuff in the post today.

One thing I think would be extremely valuable would be a  digital filter to correct a  probe response using Leo's pulser or similar as the reference input,  computing the filter using the ARM and then loading that into the FPGA and routing the ADC output through the filter.  The step responses on my two Instek DSOs are  pretty wretched judged against a well calibrated Tek.

I spent my career in reflection seismology where  correcting for non-ideal source signatures  and input amplifier and anti-alias filter impulse responses is SOP.  The only problem with doing this in a COTS DSO is having enough real estate left in the FPGA.

Interesting concept...I'd like to see you try to build this kind of compensation device.  As for a PA for Leo's pulser...I think it would simply be better to build a higher voltage step generator than to try and build some kind of fast linear amplifier.   I've built both avalanche transistor and reed switched chargeline pulsers for years.   The best reed switch designs can get you 100's of volts with risetimes down to 400ps or less.  But they are not very reproducible pulse to pulse and suffer low rep-rates and lifetime.   The avalanche pulsers are just a bit slower and lower voltage but are very reproducible and have long lifetime.  The one I used in the post above outputs 45V into 50 ohms with rep-rate of up to 1kHz.   It has a risetime around 600ps.    Properly engineered you can probably get ~200ps or less out of a avalanche transistor circuit at 10's of volts.   

[rhb]

Implementing this is at the top of my list right after being able to transfer the ADC output to DRAM and from DRAM to the display. The filter is just a tapped delay line for short filters.  Longer filters are often better implemented in the frequency domain via an FFT.  But an FPGA should be able to pipeline the filter so aside from real estate the only cost would be 20-30 nS of latency at 1 GS/S which is easy to correct.

If you can live with downloading a trace to a PC and running Octave/MATLAB I can put something together.  It's been a long time since I implemented this sort of thing.  I've been using canned routines for 20+ years, but it is very simple DSP 101 stuff.  Doing it in an FPGA is more exotic. I know what to do, just not how.  I'm building out a dev environment with a Zybo Z7-20 and a BeagleBoard X15 to develop code for my Instek GDS-2072E.  So I'm in hardware mode at the moment.  Lots of bits and pieces in transit.

Computing the spectral response of my scopes is on my "To Do" list and computing a correction filter is very little more effort beyond remembering how to use Octave.

As a general outline, set Leo's pulser output to an appropriate level and record the output of the RF probe.  If you  have a scope that samples at 10 GS/S or more, record the pulser output directly, otherwise, assume that it is a perfect step.

a= pulser step in time domain
b= probe step response in time domain

A = FFT(a)
B = FFT(b)

A/B = correction filter in frequency domain
c = IFFT(A/B), the correction filter in the time domain

B * A/B = corrected step response which is b*c in the time domain  "*"  indicating a convolution in time which is a bunch of multiply & adds.  Something an FPGA is well suited to.

There are an assortment of details regarding phase, frequencies in B with coefficients near zero, etc. A lot depends upon exactly what you want to do.  In seismic processing we tend to make things symmetric aka "zero phase".  But we work entirely in recorded time and it makes interpreting the results easier.  Reality is minimum phase.

The classical way to calculate the filter  is  the Wiener-Levinson prediction error filter which operates in the time domain.  That produces a minimum phase output.

I thought that perhaps among the vast variety of MMICs  there might be a suitable device.  I guess I should go read the Jim Williams paper.

[rhb]

A common joke.

An engineer, a mathematician and a geophysicist went to a job interview.  During the interview, the engineer was asked how much 2 + 2 was to which he replied, "4".  The mathematician was next.  When he was asked, he said, "Under certain conditions it can be 3 or 5, but in most instances it's 4."  When the geophysicist was asked the same question, he looked around, nudged the door shut and said, "What would you like the answer to be?"

[Alex Eisenhut]

A common joke.

An engineer, a mathematician and a geophysicist went to a job interview.  During the interview, the engineer was asked how much 2 + 2 was to which he replied, "4".  The mathematician was next.  When he was asked, he said, "Under certain conditions it can be 3 or 5, but in most instances it's 4."  When the geophysicist was asked the same question, he looked around, nudged the door shut and said, "What would you like the answer to be?"

I think you meant "economist"...

[Insatman]

A common joke.

An engineer, a mathematician and a geophysicist went to a job interview.  During the interview, the engineer was asked how much 2 + 2 was to which he replied, "4".  The mathematician was next.  When he was asked, he said, "Under certain conditions it can be 3 or 5, but in most instances it's 4."  When the geophysicist was asked the same question, he looked around, nudged the door shut and said, "What would you like the answer to be?"

I think you meant "economist"...

I would say "department manager"

[Insatman]

Hello Everyone,
I wrote the Ebay vendor and told him I tested his probe and it didn't meet the promised specifications (see attached pic from his ad).  His response speaks for itself
===============
New message from: 60dbmcom (764Purple Star)
Hi

i tested many times too
I got about 7dB attenuation in the range from 0.1 to 500 MHz
20 db attenuation i got at 1.5 Ghz. My result nearly the same as yours
I think it is very good result for usd8 board
Anyway, if you don't like my probe and according ebay rules, you can send me back and i give you money back


Best regards
Paul
==================

So he flat-out knew it wasn't even remotely close to the specifications claimed.   Plus he also knows that with the cost of international postage returning the probe is pointless.   Note I paid $12 plus $6 shipping, not $8.   

I'm not wildly upset because I'm out less than $20 and I have a probe that has some uses...but this just goes to show you how some vendors will claim anything to get a sale.

[rhb]

A common joke.

An engineer, a mathematician and a geophysicist went to a job interview.  During the interview, the engineer was asked how much 2 + 2 was to which he replied, "4".  The mathematician was next.  When he was asked, he said, "Under certain conditions it can be 3 or 5, but in most instances it's 4."  When the geophysicist was asked the same question, he looked around, nudged the door shut and said, "What would you like the answer to be?"

I think you meant "economist"...

Back in the late 70's a friend of mine published a paper.  He  sent the same seismic data to a dozen processing shops for processing. None of the processed sections looked even remotely the same. Large seismic  processing jobs cost millions of dollars and are used to make billion dollar decisions.  It's a bit embarrassing, but no one has ever found a solution.

[KD0CAC John]

Old accounting saying - Figures do not lie , but lier's can figure .
And as long as we at it , lawyer's are figuring

Oh and good thread topic - the one about the probe .   

[metrologist]

Copper tubes are good for this..

Do you have plans for this? Thanks