Active RF probe - scope termination matching

place a small 6 dB 50Ohm pad at the scope end of the xmission line, this will reduce the impact of any mis-mismatch between th eline and the scope

Mark

That would work I guess. Only my goal of a 1:1 probe would then be more difficult. But I could add a MMIC at the scope end as well.

Joop

The input capacitance should be well defined. In most 'scopes that I'm familiar with there is an adjustment for Cin.

If you add a series resistance into the 'scope, you can synthesize a constant resistance looking into the combined R-L,R-C circuit. Of course you lose bandwidth compared to the R-L circuit alone. Alternatively you can use a T-coil to get ~2.6x bandwidth and about 2x improvement in resistance flatness.

Better yet, use a 'scope that has a real 50ohm input.

-frank

I must say I do not own a very high frequency scope (yet). I am simulating what I may expect from a homemade probe. You mention real scopes compensate Cin. What does that mean? They boost the frequencies at the high end? I guess it cannot mean they compensate for reflections of non-pure ohmic termination.

This is a bit confusing for me. Can you refer me to a drawing?

Right. It seems present on all VHF scopes I have seen. But stil they have a Cin to deal with.

Joop

In the simulation a 6dB attenuator smoothens the response above 200MHz really nicely. 3dB would already have an effect, but 6dB makes the curve definately monotone. So if it is safe to assume a scope compensates somewhat for the expected drop at these frequencies, then that would work great.

Joop

You're joking or just simulating ? First of all, what MMIC, what oscilloscope, and what frequency/amplitude range?

Vasile

The BF998 may be fine for a low-noise low-capacitance input, but take care with your "buffer" output stage. MMIC are meant for 50-ohm termination. Just use that, don't fool around.

Do you have any isea what the BF998 dc Vgs drift might be? I like the 0-volt Vgs operation (Id=10mA, Vds=10V, Vgs2=4V), but how tempco stable is it?

I didn't say "compensate", I said "adjust". Essentially there is an _additional_ shunt C used to bring the capacitance up to some predefined value. Obviously this doesn't improve reflections -- but it does make the design of a matching network easier.

A long time ago I experimented with an improved input network for making a Tek 475 have a flatter 50ohm Zin. I used the T-coil method. It worked, but not enough to make the effort worthwhile.

However this could still be a very worthwhile learning exercise. But don't (IHMO) just use circuit simulations -- dig into the equations if you're more interested in theory, or do some real circuit tests if you're more interested in practical details. Either could be done quite cheaply and you'd come out with much more than you went in.

-frank

The device I had in mind is a BGA616 for which there is a SPICE model. Like you said, it is meant to be loaded with 50 ohm, however measuring the output impedance shows the impedance of the BGA616 itself varies quite a bit with bias current. So the right attentuation/impedance match before the cable might help surpress reflection but then at the probe side. More attenuation, hmm...

I have no idea. The datasheet only mentions all the *typical* values at 25 degrees. Did you have a bad experience on this point?

Also I can imagine a simple probe will not be optimal on all fronts. In that case I prefer low loading of the circuit to be measured. And temperature effect on gain would be secondary. As long as there is enough gain left to reasonably measure small signals.