RF Voltage Measurements

It is not a common problem and I suspect something is broken on those scope probes or the scope. I regularly use 250MHz-rated scope probes for most of my work and it's all nice and flat with a gentle roll-off at the upper limit. If it weren't so then any measured waveform would have to be suspect.

Whoah, you're not supposed to terminate the scope channel in 50 ohms if the probe on that channel is a X10 voltage probe. Leave the channel in 1M input mode. It is a 1M probe.

Ahh, I didn't terminate the v-probe channel at 50ohms that was left at the default 1MEG, just the reference channel was terminated at 50ohms to match the amplifier output.

As for something being broken, this is a possibility but I've tried three Tektronix probes, the 500MHz x10 P6139A, a new P5100 250MHz x100, and a P6015A 70MHz x1000. None of them give me a linear response... Thinking along the same lines as Joerg I managed to borrow a 1GHz scope, yet the results are the same so I discounted a fault with the scope.

I must be missing something as my experiences would suggest that anything above 20MHz is un-measurable, which is obviously not the case! I've been struggling with this for some time now and the only way I've found to circumvent it is to calculate the percentage error at a given frequency and use that to calibrate my results, this is a really poor and time consuming method!

I started to write the same thing - then reread the OP and realised it should be OK, he is just using the 50 input as a terminator ("dummy load").

RF ==========================T=============50 ohm scope input \\ ---

As a test, take the fastest function generator you have and probe its output at 1MHz, 10MHz and 100MHz. With the scope probes. If you see undistorted results except for slight rounding at 100MHz then something wasn't right with the connections in your RF setup. If there is lots of ringing and distortion at 10MHz your scope probes or scope are faulty.

Hi John / Joerg.

That's exactly the setup I used, and the probe was definitely terminated at 1Meg.

I've looked at the output of the signal generator from 1-100MHz and it never becomes distorted, it's just the amplitude that changes. The waveform always looks fine; I even tried a square wave which also appeared ok. I'll have another look in the morning and try post some waveforms and photos.

Thanks for the help.

That would be good since in this case it can only be the measurement setup that has a problem. Probes and scope are fine. And thoroughly check your ground connections. Stuff like this happens a lot if a ground becomes loose.

Are you using the same ground clip? Maybe it's bad.

Are you using a ground clip at all?!!

John

How long are your ground clips on the probes? A long ground clip has significant XL at VHF frequencies. Keep your ground connections as short as possible. Best ground is to use a short stub of stiff wire wound around the probe shield sleeve right at the tip, maybe even soldered to a good ground point close to the point being probed.

point

the

Yep. And since he's using a T (presumably a BNC T), he could just use a probe-to-BNC adapter to get seriously good grounding at the probe tip.

OTOH, if the spectral response to a swept sine is not flat, then the time response to a square (or triangle) should be distorted as well.

Cheers, Tom

(Living proof about the judgment thing, too...sometimes have to repeat the bad judgment a few times before the good starts to sink in.)

point

the

judgment.

More on this: I just ran a test, set up the same way. I get two or three percent variation if I ground the probe right between the sleeve around the tip and the BNC shell. I get about four times as much variation if I use a standard short (about 4 inch) ground lead that attaches to the probe nearer its "handle", to the BNC shell. Things are pretty terrible if I don't connect the ground at all, of course. This is from low freq. up to about 100MHz. With the tip grounded tightly to the shell, the measurement technique is useful up to 500MHz (the limit of this particular scope)--roughly 10% amplitude variation over that span. In any event, this is NOT a precision RF amplitude measurement technique!

Cheers, Tom

..

at

ow

ort as

nd the

ound point

ers in the

judgment.

Hi All,

After reading through your comments I tried removing the ground lead of the probe and making a much shorter connection. The situation is now greatly improved. From 1 to 50MHz the error between the reference and measured signals is well below 1%, from 50MHz-100MHz the error gradually increases to 8% (still not ideal but I can cope with it!). Using the 5 inch ground lead provided with the Tek probe I find an 11% error at 50MHz and a massive 66% error at 100MHz! I=92ve well and truly learnt the value of a good ground connection!

Thanks for all the advice.

Jim.

Hi All,

After reading through your comments I tried removing the ground lead of the probe and making a much shorter connection. The situation is now greatly improved. From 1 to 50MHz the error between the reference and measured signals is well below 1%, from 50MHz-100MHz the error gradually increases to 8% (still not ideal but I can cope with it!). Using the 5 inch ground lead provided with the Tek probe I find an 11% error at 50MHz and a massive 66% error at 100MHz! I?ve well and truly learnt the value of a good ground connection! Jim.

I have posted a jpeg on alt.binaries.schematics.electronic that shows a Curly Q I use to eliminate trash on waveforms. Simple probe holder made from 18 gauge wire. Mike

Mike I meant to add the subject line is "RF Voltage Measurements Curly Q"

point

in the

judgment.

Just a hint for future measurements: Good scope manufacturers offer "semi-disposable" jacks you can solder into test circuits. They are like a shielded coax connector into which the probe tip slides, providing center and GND connection right there the tip.

at

ow

hort as

und the

round point

ters in the

judgment.

Yes, we sometimes design them into production boards to enable proper and consistent probing of certain critical signals. Also, you can often get adapters to fit probes to at least some of the common coax standards--BNC probably most commonly, since the ones like SMA designed for higher frequencies try to avoid T connections (which introduce stubs) anyway.

Cheers, Tom

Cheers, Tom

as

the

ground point

in the

judgment.

I usually try to convince clients to go to, gasp, RCA jacks. A 500ohm to

Would that you all would find the wonder of a good

SNIP

SNIP

Dammit.

Jim

-- "It is the mark of an educated mind to be able to entertain a thought without accepting it." --Aristotle

After reading through your comments I tried removing the ground lead of the probe and making a much shorter connection.

The accuracy of scope measurements is in the +- few-percent range even on a good day, and with a really fast digital scope (e.g. my Tek TDS7704B) it's worse than that, regardless of what the manufacturer may say.

Cheers,

Phil Hobbs