zero capacitance probe

[electrolust]

Do I understand correctly, that the 50r input of a scope has no capacitance in parallel, at the scope input?

The Tek P6150 probe claims .15pF with the x10 barrel and lists no capacitance with the x1 barrel.  How is .15pF, much less zero, capacitance possible?  I take it from the spec that the cable itself has no capacitance and the x10 attenuator adds the .15pF.

I'm also looking at AoE 3rd ed, p. 809, and for their homemade lo-z cable they use RG-178, which seems to have 29.4pF/ft capacitance.  How can this probe/cable capture a 10MHz square wave with so much capacitance?

[w2aew]

The 50 ohm coax for the P6150 probe certainly does have capacitance. It is evenly distributed along the length of the transmission line, just as the inductance is distributed along the length of the line as well. The dimensions of the coax and the selection of the materials are designed to make the combination of the distributed inductance and capacitance appear like a 50 ohms. The scope input appears like a broadband 50 ohm termination. So, when used without the attenuator tip, the "probe" looks like a broadband 50 ohm load.

The used with the attenuator tip (450 ohm series resistor), the input looks like a 500 ohm load in parallel with the 0.15pF capacitance. 

[electrolust]

I think I get it.  Would it be correct to say, this is getting into transmission line theory?

[Zero999]

I think I get it.  Would it be correct to say, this is getting into transmission line theory?

Yes.

As mentioned above, the 50 Ohm cable looks like a perfectly resistive 50R load, when loaded with a 50R resistor.

Below the resonant frequency of the cable, if it's terminated with >50R then it'll appear capacitive and when it's terminated <50R it'll appear inductive. This is why if you connect a capacitance meter to a piece of cable with an open circuit on the other end, it will measure the capacitance.

[w2aew]

I think I get it.  Would it be correct to say, this is getting into transmission line theory?

Yes.

As mentioned above, the 50 Ohm cable looks like a perfectly resistive 50R load, when loaded with a 50R resistor.

Below the resonant frequency of the cable, if it's terminated with >50R then it'll appear capacitive and when it's terminated <50R it'll appear inductive. This is why if you connect a capacitance meter to a piece of cable with an open circuit on the other end, it will measure the capacitance.

Another way to think about this.  For a cable that is open circuit at the far end, it will look capacitive or inductive depending on the frequency.  For frequency where the cable is less than a 1/4 wavelength long, it will appear capacitive.  When it is between 1/4 and 1/2 wavelength long, it will appear inductive.  Then, from 1/2 to 3/4 wavelength, it's capacitive again, and so on.  So, for a fixed length of open circuited coax / transmission line, the input impedance will vary as a function of test frequency. 

Some useful features here are:
- at odd quarter wavelengths (1/4, 3/4, etc.) an open circuited line will appear as a short, and a shorted line will appear as an open
- at even quarter wavelengths (1/2, 1, etc.) the input impedance will exactly match the load connected at the far end, even if mis-terminated.

[David Hess]

I agree with all of the above.

One other thing which complicates the matter is that most oscilloscopes which have switched 50 ohm input terminations on their high impedance inputs do not present a 50 ohm load either.  They just place the 50 ohm termination in parallel with the high impedance input so they present a 50 ohm load shunted by their typical input capacitance of between 10 and 20 picofarads.  (1)

Oscilloscopes with dedicated 50 ohm inputs lack this extra shunt capacitance and since they also do not use the FET buffer stage necessary for a high impedance input, they are also lower noise.

(1) The only oscilloscope I know of which actually switched between dedicated high impedance and 50 ohm inputs is the Tektronix 485.  Maybe someone knows of some others which did this.

[Alex Eisenhut]

necessary for a high impedance input, they are also lower noise.

(1) The only oscilloscope I know of which actually switched between dedicated high impedance and 50 ohm inputs is the Tektronix 485.  Maybe someone knows of some others which did this.

HP1741 has a manual switch on each input volts/div switch.



You can see it goes ac-gnd-dc-50R.

[tggzzz]

necessary for a high impedance input, they are also lower noise.

(1) The only oscilloscope I know of which actually switched between dedicated high impedance and 50 ohm inputs is the Tektronix 485.  Maybe someone knows of some others which did this.

HP1741 has a manual switch on each input volts/div switch.

You can see it goes ac-gnd-dc-50R.

The 1740A just uses a switch, S1C, to connect a 50ohm resistor (A3R1) across the input to ground. The attenuator and amplifier is unchanged. If you overload the 50ohm input, the resistor fries.

The Tek485 has two parallel attenuator sections, one for 50ohm impedance throughout, one high impedance. The amplifier is unchanged, but takes its input from either the 50ohm attenuator output or high impedance attenuator output. If you overload the 50ohm input, a relay pops to protect the resistor.

I have both scopes, like them both, but the 485's risetime is noticeably faster!

[David Hess]

The Tek485 has two parallel attenuator sections, one for 50ohm impedance throughout, one high impedance. The amplifier is unchanged, but takes its input from either the 50ohm attenuator output or high impedance attenuator output. If you overload the 50ohm input, a relay pops to protect the resistor.

On the 485, it is not just the high impedance attenuator which is swapped for a low impedance attenuator.  The input goes to an RF relay which directs it to either a 2 section low impedance attenuator or a 2 section high impedance attenuator followed by the high impedance buffer.  Then an RF switch directs the now 50 ohm signal from the appropriate section into a 2 section low impedance attenuator.  There is never a 50 ohm termination resistor on the input in 50 ohm mode; it is either a 50 ohm transmission line or a 50 ohm attenuator.

I have both scopes, like them both, but the 485's risetime is noticeably faster!

In high impedance mode the 485 has the same bandwidth as the 250 MHz 475A which was released 5 years later but in low impedance mode, the 485 is 350 MHz which was quite an accomplishment in a portable oscilloscope for the time.  The 485 was released in 1972 and it was not until 1984 that Tektronix released the 2465 to replace it and the 2465 was only 300 MHz although in both 50 ohm and 1 megohm modes.

I usually recommend against the 485 because of its extra complexity with the dual input sections but people who have them seem to really like them.  I bet some of that is because the higher acceleration voltage of the 350 MHz CRT makes it extra sharp and bright.  I like my 7904 for the same reason.

Anyway, I would be interested in hearing about any other switchable 50 ohm and 1 megohm input oscilloscopes that do it without just shunting the high impedance input with a 50 ohm termination.  I double checked the 1741 schematics and it just uses a switch 50 ohm termination on its high impedance input like most oscilloscopes.

[tggzzz]

I usually recommend against the 485 because of its extra complexity with the dual input sections but people who have them seem to really like them.  I bet some of that is because the higher acceleration voltage of the 350 MHz CRT makes it extra sharp and bright.  I like my 7904 for the same reason.

Yes, plus the 485's triggering. ISTR getting mine to trigger on a 1GHz signal; I suspect I could have gone higher, but the amplitude was so low

But I also like my 1740A; I'm going to sell my 465s in preference to selling that.

[nfmax]

I think I get it.  Would it be correct to say, this is getting into transmission line theory?

Yes.

As mentioned above, the 50 Ohm cable looks like a perfectly resistive 50R load, when loaded with a 50R resistor.

Below the resonant frequency of the cable, if it's terminated with >50R then it'll appear capacitive and when it's terminated <50R it'll appear inductive. This is why if you connect a capacitance meter to a piece of cable with an open circuit on the other end, it will measure the capacitance.

Another way to think about it - a 50 ohm transmission line, no matter what impedance it is terminated with, will 'look' lie a 50 ohm resistor until the signal has time to travel to the far end, discover that it's not terminated in 50 ohms, and travel back to the input. At that moment the impedance suddenly changes. Or to put it another way, the impedance is a function of the reflected signal in the line.

[mzacharias]

I usually recommend against the 485 because of its extra complexity with the dual input sections but people who have them seem to really like them.  I bet some of that is because the higher acceleration voltage of the 350 MHz CRT makes it extra sharp and bright.  I like my 7904 for the same reason.

Yes, plus the 485's triggering. ISTR getting mine to trigger on a 1GHz signal; I suspect I could have gone higher, but the amplitude was so low

But I also like my 1740A; I'm going to sell my 465s in preference to selling that.

You like the 1740A better than the 465's? Any particular reason or just personal preference? I've never used a 1740A - I love my 465 but they aren't the most reliable things, and I've had to fix mine a couple times - I always dread that I might have to do a major dis-assembly on it. The drawings look scary. I've seen pics of the insides of a 1740A and ISTR they look somewhat easier from a service standpoint.

Are their crt's as good as the 465? That's the one thing I always loved about those - always thought Tek crt's were the best.

[tggzzz]

I usually recommend against the 485 because of its extra complexity with the dual input sections but people who have them seem to really like them.  I bet some of that is because the higher acceleration voltage of the 350 MHz CRT makes it extra sharp and bright.  I like my 7904 for the same reason.

Yes, plus the 485's triggering. ISTR getting mine to trigger on a 1GHz signal; I suspect I could have gone higher, but the amplitude was so low

But I also like my 1740A; I'm going to sell my 465s in preference to selling that.

You like the 1740A better than the 465's? Any particular reason or just personal preference? I've never used a 1740A - I love my 465 but they aren't the most reliable things, and I've had to fix mine a couple times - I always dread that I might have to do a major dis-assembly on it. The drawings look scary. I've seen pics of the insides of a 1740A and ISTR they look somewhat easier from a service standpoint.

Are their crt's as good as the 465? That's the one thing I always loved about those - always thought Tek crt's were the best.

Any discussion like this has to be based in personal preference; if that's not acknowledged, then there is a degree of self-deception!

Having said that, I've repaired Tek 464s, 465s, 475s, 485s, 2445s, 2465s, 1502s, and my 1740. I've had fewer problems on my 1740, but I wouldn't read too much into that.

The construction of the Tek scopes improved considerably over the years; the 465s and 475s are pretty dreadful, the 1502 and 485 are much better, as are the 24x5s (with the exception of the cal setting memories).

Tek definitely unnecessarily pushed some of their capacitors too hard (e.g. 15V tant beads on 13V lines). OTOH the Tek timebase/attenuator switches are better than the HP equivalents (e.g. tracks are eroded by the switch rotor).

The HP1740A's screen is at least as good as the 465/475, plus it has a (unprotected) 50ohm termination. The 465/475 is smaller; the 485 is more or less the same size.

But I can get more money selling 465s than 1740s; shame!

[T3sl4co1l]

A perhaps more useful but more mind-bending view is this:

Transmission lines are real, and there is no such thing as a resistor, capacitor or inductor.

What we call a "resistor", or etc., is a component which exhibits that characteristic over a usefully wide frequency range.

Resistors are resistive from DC to some frequency limit, where they become inductive or capacitive.  (In particular, small value resistors become inductive, and large value resistors become capacitive.)

Capacitors are capacitive over a frequency range: at DC, they have leakage resistance; at high frequency, they have ESR and ESL (and other more complicated responses).

Inductors are inductive over a frequency range: at DC, they have DC resistance; at high frequency, they have EPR and parasitic capacitance (and other more complicated responses, especially when multi-winding transformers are included among inductors).

Minding that the whole process is recursive, so as a component drifts from one characteristic to the next, that other characteristic is only valid over a range as well, and so on!

A transmission line is a different paradigm.  An infinite, ideal (lossless, non-dispersive) transmission line has a constant impedance, over all frequencies: DC to light.  That's called the characteristic impedance, Zo.  (The impedance is real, i.e., a resistance, but we call it impedance to remind ourselves we're talking about impedance at AC frequencies.)

When transmission lines, of different lengths and Zo's, are connected together, we observe reactance.  How much depends on the ratios of impedances, while the frequency behavior is determined by the lengths (because of standing waves).

If we have a straight length of transmission line, and we measure its end-to-end impedance at DC (zero frequency), we get the resistance of the line: zero ohms for an ideal line, but nonzero for a real line, of course.  At low (but nonzero) frequencies, however, we measure an impedance (even if it's lossless).  This impedance is proportional to length, Zo and frequency.  What else is proportional to those figures?  Inductance!

This is how a transmission line can be said to have inductance: it is an approximation, which is only valid at frequencies much less than the electrical length of the transmission line.

The same goes equally well for capacitance, of course: if you have an unterminated transmission line, and measure its impedance at one end, you get infinity at DC, but finite impedance at nonzero frequency.  The impedance is inversely proportional to length, Zo and frequency: a capacitance!

When a transmission line is terminated by a matched resistance, it looks infinite: a true resistor load (terminator) is indistinguishable from an infinite lossless transmission line.  The wave seems to go on forever, except you've tricked it into a trap that turns it into waste heat.

In practice, real resistors can be made very good indeed (resistive up to ~GHz).  As long as it's a resistor at frequencies well beyond what you're using them at (including harmonics of a signal -- you can't cheat the system by, say, dropping your clock frequency to 1kHz!), then it's ideal enough not to care!

So, going back to the probe: if we have a resistor that's close enough to a true resistance, at any frequency in the range we're worried about (i.e., some ~GHz), we can make a voltage divider into a "true 50 ohm" transmission line.  The resistor should be 450 ohms, so we get a calibrated 1/10 ratio into 50 ohms.

In practice, some compensation will be needed, to account for the transmission line being lossy (it attenuates high frequencies more than low frequencies), or we can make the line short enough not to care (but that might be impractically short?), or we just accept that it happens, and mentally adjust the measurements accordingly.

Tim