Hi Z probe for 50 ohm spectrum analyzer

[joeqsmith]

If it were the 50 ohm, you could still switch it out.   Seems like an odd failure for the switch.  Guessing it's not these but upstream.  Maybe C33 or a bad cable.  Maybe your meter or your connectors.   Should be a few minutes to hunt it down.   

[Kosmic]

It has 1nF input capacitor connected to what? To mosfet gate and loaded with 10M resistor

Input capacitance of that probe is 1 nF input capacitor in series with gate capacitance of that mosfet ( Which is typically around 2 pF), that is in parallel with parasitic capacitance of that 10M resistor .

So how much of that capacitive load is presented to DUT?
1nF or something about  2pF?

From the article:

Capacitor C1 has a value of about
0.5 pF, and is made by patches of copper on each side of the board. The gain
of the buffer itself is a little less than
one, but because of the voltage divider
action of C1 and the input capacitance
of T1, the overall loss of the probe is
approximately 20 dB

So no, I don't think substituting the 0.5pF cap by 2 x 2nF cap will really give the same input capacitance of 0.75pF. Unfortunately I'm not equipped to measure the input capacitance of the probe so I can't validate.

I recently acquired a nanovna, so I did more test with the cheap active probe. I removed all my modifications and tried to measure the impedance of the probe in the state it was received.



the input capacitance is around 3.9pF at low frequency but get quickly to 6.5pF at 543MHz and 17.8pF at 726MHz. We are really far from 0.5pF.

It's a shame since the author of that probe could had made it better just by closely following the Elektor article.

[mnementh]

Maybe this is suitable for your application.  It is somewhere between DIY and a packaged solution.
"RF Active Probe 0.1-1500 MHz-1.5 GHz analyzer oscilloscope" from ebay seller 60dbmco in Ukraine
https://www.ebay.com/itm/332393676082

Nice find.  Have you tried one to see how well it performs?

I have one of those. Look like the seller tried to implement the design published in Elektor magazine a while ago. On his ebay listing he even use some schematic from the article.

Most of the problem come from C1. For some reason (limit attenuation between 0-500Mhz  ), he decided to use 2 x 2nF in series. Was supposed to be a 0.75pF PCB cap created with 2 planes on each side of the PCB.

   This is what the original probe look like.      This is the frequency response from 50 to 1500Mhz.   

Now a bodge to try to fix the probe. I relocated the pogopin on the other side of the board. It was soldered on a piece of copper tape. I added some super glue to make sure the copper tape and pogopin will stay in place. The Kapton tape was just added to isolate the ground pin.

       Now the frequency response of the fixed probe.   

For reference, the Elektor article about the active probe is in attachment: https://www.eevblog.com/forum/testgear/hi-z-probe-for-50-ohm-spectrum-analyzer/?action=dlattach;attach=1060862

[   https://www.eevblog.com/forum/testgear/test-equipment-anonymous-(tea)-group-therapy-thread/msg3621622/#msg3621622

For those who care... 60dbm has updated his design with a proper gimmick cap on the input as in the original Elektor project. I've designed a shell for it which includes a holder for A23 car remote battery; the ~20mA draw sucks it down pretty quick, but still quite usable and mucho convenient for occasional use. https://a360.co/37CE8El



I'm waiting on pogo pins for the (GND) side then I'll try and get some data from my nanoVNA for y'all.

mnem
 

[mnementh]

      

This is what I've got so far; it told me what I needed, which is that it would present reasonably low loading on a CPU crystal oscillator I wanted to probe for signs of life. I'm still working between bouts of packing and looking at properties to move; I'll be back with more data ASAP.

One thing I like aboot 60dbm's new probe is the big pads for soldering pins to; you can easily reconfigure to match almost any needed pin spacing. Of course that is probably also why ~1.1 puff loading instead of 0.5, but hey... usability factor is important too; that's why I went faffing aboot with a little car remote battery.

mnem
 

[Kean]

I've designed a shell for it which includes a holder for A23 car remote battery; the ~20mA draw sucks it down pretty quick, but still quite usable and mucho convenient for occasional use. https://a360.co/37CE8El

Hey mnem,

Did you intend for the F360 link to allow downloads of your model?  Pretty please 

Since I bought the ebay probe a few years ago, I have since acquired a pair of used & working HP 54701A 2.5GHz Active RF Probes (100k 0.6pF input).
I may never use the ebay probe again, but it would be nice to have a case protecting it so if ever someone wants to borrow one of my RF probes, I can offer that one.

Kean

[Kean]

For anyone coming here later, mnemeth has just posted the STL and Fusion design files in another thread
https://www.eevblog.com/forum/3d-printing/3d-printer-yet/msg3623898/#msg3623898

[mnementh]

LOL.... you done stole muh thunder. 

Just be aware that this fits the new, blue version of this probe/PCB as shown above pretty snugly. PCB dimensions are 9mm W x 1.6mm thick exactly; approx 57mm long not including the SMA connector. I don't know if it will fit the old green version; the cutouts for that cap and diode in the battery compartment are specific to the blue one.

mnem
Direct Link to ZIP file:   https://www.eevblog.com/forum/3d-printing/3d-printer-yet/?action=dlattach;attach=1243488
 

[Kosmic]

Since mnem got a new model with the gimmick capacitor on the input, I decided to order another one to compare.

The 2 probes. I replaced the original pogopin and used shorter one I had.





I changed my setup a little bit to be able to measure S21 on top of S11. The VNA has been previously calibrated with the small test PCB and the probe is connected near the load.



Old probe (yellow) VS the New one (blue):


New probe (yellow) VS the Old one (blue):


Overall, the new one is an improvement. Perform relatively well up to 1G and input capacitance is a lot lower (1.2 to 1.5pF compared to 3.5 to 8.5pF for the old one). The gain is unfortunately not linear and starting at -25dB (and not -20dB as mentioned in the Elektor article).

[harrimansat]

Can  you please take S21 curve without probe ground connection?

Thanks

Regards

Miguel

[Kosmic]

Can  you please take S21 curve without probe ground connection?

Thanks

Regards

Miguel

In this case, the source and the receiver are in the same VNA so they share a common ground. Also, even if the probe is battery powered, the ground is shared (connected by the SMA connector).

This is a sweep with the ground pin floating. The reference in blue is a normal sweep with ground. In yellow there is no ground.


You always need a good and short ground to avoid problem like that

[harrimansat]

My DIY fet probe

[Joel_Dunsmore]

Here's one I help out with as a youngster.  Maybe not quite the price point but 30 years later it still has quite good performance, and we still actively sell it:
https://www.keysight.com/us/en/product/85024A/high-frequency-probe-300-khz-3-ghz.html
and tech overview: https://www.keysight.com/us/en/assets/7018-06843/technical-overviews/5968-2101.pdf

No ground pin as the  barrel held in your hand provided a low enough impedance to ground that no ground clip was needed.  AC only of course. 
0.7 pF input  impedance, flat to past 3 GHz.
You could DYI it if you had your own custom GaAs fab. 

[Kosmic]

Brand new it might be a bit much but I can always try my luck with ebay. Added to my watch list