Tektronix P6106 scope probe performance?

[dzseki]

I've got a tek P6106 and a P6105 probe along with my HP 1720A (275MHz) oscilloscope that I bought a month ago. Both probes are the 2m (6,6ft) version. I am curious if anyone have any performance detail from the first hand about the P6106 probe? I've looked up two different tektronix spec sheet about this probe, and I found it somewhat inconsistent.
First, they say it is usable with scopes up to 350MHz bandwidth -OK!
Then they say if the oscilloscope's bandwidth is greater than 255MHz, with this probe the overal performance will be 250MHz (-3dB) -This already stinks if we calculate the rise time equation of cascaded systems.
Otherwise there is absolute no information what rise time should expected... This probe was inteded to use with the Tek 485 and 475A scopes, to my understanding my HP 1720A scope is superior to both in regarding bandwidth in 1MOhm input mode.
On the other hand I've tried to take performance measurements with this probe, and the results are disappointing. I want to use this probe to measure high frequency video signals. First I put my PC's VGA output directly to the scope's 50 Ohm input with a DSUB to BNC cable, and despite the 75 Ohm / 50 Ohm impedance mismatch I've measured 1,4ns overall rise time (400MHz DAC in the VGA, 275MHz scope) so this is a good starting point.
With the P6106 probe I can't get rise time below 2ns though, even without grounding cable (so the ground barrel of the tip touched the measuring ground directly).
On the other hand my P6105 probe is very simlar to the P6106, but it has an advertised bandwidth of 100MHz, and has less adjustments for the response compensation, but I've measured eventually the same rise time (2ns) with that probe as well (though the HF compensation wasn't as good as with P6106).
Does anybody have infromation about these old probes, what should I expect? What probe should I use to get the maximum performance out from the scope?

[edavid]

I don't know why you say the P6106 stinks.  It's a good 250MHz probe.   2ns risetime seems reasonable... you would need a faster source and scope to make a better measurement.

Anyway, you might try an HP 10017A 275MHz probe.

[nctnico]

I said it before and I'll keep repeating it: as a rule of thumb you need HF probes (or direct 50 Ohm coupling) for signals over 100MHz. The input capacitance of regular probes kills HF signals.

[alm]

What is the impedance of a 10.5 pF probe at 250 MHz compared to say a 50 Ohm input? I agree that passive hi-Z probes become less useful as the frequency goes up, but that 250 MHz is still in the in-between area where passive hi-Z probes might be useful.

[nctnico]

Did you ever try to make a meaningful measurement? I tried but quickly moved to an HF passive probe to get something on the screen which makes sense. 50 Ohms loaded by approx 70 Ohms gives a lot of dampening and its frequency dependant too so a square wave looks likes anything but a square wave.

[w2aew]

The Tek 485 is a 350MHz BW scope, rated fairly conservatively. I've measured sub 1.5ns rise times with a properly compensated P6106 on this scope with a low ground inductance test fixture.

[edavid]

The Tek 485 is a 350MHz BW scope, rated fairly conservatively. I've measured sub 1.5ns rise times with a properly compensated P6106 on this scope with a low ground inductance test fixture.

The 485 is actually spec-ed at 250MHz in 1M input mode.

[marshallh]

Here is a capture demonstrating the problem with high bandwidth, high impedance probing.

Scope: Lecroy 9374C 1GHz input bandwidth
Probe: Lecroy PP006 500mhz 10x 10Mohm, 11pF with 4mm ground clip



The 250mhz, 1.8v bus clock is bandwidth limited and you only see the fundamental, and none of the harmonics necessary to see it as the square wave it is. It is also loading down the clock driver a lot.

[dzseki]

I said it before and I'll keep repeating it: as a rule of thumb you need HF probes (or direct 50 Ohm coupling) for signals over 100MHz. The input capacitance of regular probes kills HF signals.

I am heading to connect to the 50 Ohm input, as I have three gain stages (different boards) that are interconnected with 75 Ohm coaxial cable, so I can connect there directly with the scope, the problem is I need 75 Ohm / 50 Ohm matcher, but that is the smaller problem.
This is not enough yet, the last stage is a CRT driver amplifier, which puts out a 60Vpp signal that can't be terinated with 75 Ohm, actually the wise men says the CRT Cathode is mostly behaves like a single 8-10pf capacitor on the output, so for the best possible measurement I'd need to simulate this as much as possible.

[alm]

Did you ever try to make a meaningful measurement? I tried but quickly moved to an HF passive probe to get something on the screen which makes sense. 50 Ohms loaded by approx 70 Ohms gives a lot of dampening and its frequency dependant too so a square wave looks likes anything but a square wave.

As long as the source impedance is not too high it works OK. For a 25 Ohm source (terminated 50 Ohm system), the attenuation by 70 Ohm will only be about 25%, which is less than -3dB. If you compare the output of a fast signal generator (low impedance), the FET probe is clearly superior, but the passive probe is by no means unusable. If you look at the lines of a memory bus (top is a passive probe, bottom is a FET probe), the passive probe probably produces a little more ringing and slower rise times, but again its quite usable in my opinion. Passive probes are of course more sensitive to ground lead inductance, which might be an issue here.

Also note that a FET probe at its rated bandwidth is not much better. For example, a 1.5 GHz FET probe might have 1 pF of capacitance, or about 100 Ohm at 1.5 GHz. So the bottom line is that both probes suck near their rated bandwidth, it's just that FET probes generally have a wider bandwidth.

Resistive divider probes are much better in this regard (fairly constant impedance over its rated bandwidth, and higher than a FET probe width the same bandwidth near the top of the range), but the DC loading sucks and is too high for many circuits.

[macboy]

Here is a capture demonstrating the problem with high bandwidth, high impedance probing.

Scope: Lecroy 9374C 1GHz input bandwidth
Probe: Lecroy PP006 500mhz 10x 10Mohm, 11pF with 4mm ground clip



The 250mhz, 1.8v bus clock is bandwidth limited and you only see the fundamental, and none of the harmonics necessary to see it as the square wave it is. It is also loading down the clock driver a lot.

With all due respect, did you expect to see the first harmonic (3rd) at 750 MHz through a 500 MHz rated probe? Or perhaps the second one (5th) at 1250 MHz? Do you think that the 250 MHz clock signal is a nice sharp-edged square wave?

edit:
I just noticed you said you used the "4mm ground clip". I am not sure whether you mean you used the alligator clip ground lead, or the spring tip, or something else. Can you clarify?

[nctnico]

If its a scope with a proper Gaussian response then you should see some of the third harmonic if you use a proper HF probe. A sine wave is a nice indication if there is some signal but in many cases you need some of the extra harmonics to check logic levels etc. I once had a problem with a 100MHz clock. A standard 1:10 probe showed me a sine wave but with an HF probe I could see that the duty cycle wasn't good so the clock didn't meet the specs causing erratic behaviour of the circuit.

[dzseki]

Guys, my intention to measure an analog video path and set it for the best response (best rise and fall time without ringing). The thing in question is a CRT projector, I want to extend it's bandwidth if I can.
The projector is considered  pretty old by today's standard (18yrs old), yet it handles HD signals quite well, and I am curious if I can make it even better. It has the sharpness to handle the vertical resolution well, but at some point the signal path's bandwidth limits the horizontal resolution. The advertised bandwidth of the projector is 100MHz (-3dB), it has three gain stages, I have an idea how each stages are performing (based on their components), but want to measure also what they are capable. Now the first two stages should have a bandwidth near to 200MHz but there isn't much components there which could be altered -so there it's just curiousity. The last stage is the CRT driver which probably has about 100-130MHz BW, and the output signal is about 60-80Vpp. On that board there's quite a few things that I can play with, so the main task is to adjust this board for peak performance.
And I have the HP 1720A (275MHz) scope and this P6106 probe (and a few other slower one) for this problem, and the best rise time I've ever got with this probe was around 2ns only (and this wasn't easy already), mostly get around 2,5-3ns instead. This I find rather poor for my purpose, and I try to find out the causes and looking for solutions..
The scope surely has the bandwith because I was able to measure 1,4ns rise time from the consumer grade 400MHz video DAC output (Geforce 7950GT) directly, that is not bad at all.

[alm]

Ground lead inductance might still be in issue. Of course alligator clips are out, you should only use low inductance grounds (eg. the springs you put around the ground collar). Is the point in the circuit you use as ground close to the signal, or is it some random other trace that happens to be connected to ground? Is the video signal terminated in both the measurement with direct BNC connection and with the 10x probe? Otherwise (depending on the length of the traces/cable from the driver) you might get nasty reflections that mess up your signal.

As ntcnico already mentioned, test point impedance is critical. The bandwidth figure for the probe is for a 25 Ohm output impedance (a terminated 50 Ohm system). A test point with 75 Ohm output impedance will give substantially less effective bandwidth. Another reason why terminating might help, since it will lover the output impedance. Terminating in 50 Ohm is probably just fine, the impedance mismatch between 75 Ohm and 50 Ohm is tiny compared to 75 Ohm and 10 Mohm (at DC).

[marshallh]

I just noticed you said you used the "4mm ground clip". I am not sure whether you mean you used the alligator clip ground lead, or the spring tip, or something else. Can you clarify?

4mm spring tip

[nctnico]

components), but want to measure also what they are capable. Now the first two stages should have a bandwidth near to 200MHz but there isn't much components there which could be altered -so there it's just curiousity. The last stage is the CRT driver which probably has about 100-130MHz BW, and the output signal is about 60-80Vpp. On that board there's quite a few things that I can play with, so the main task is to adjust this board for peak performance.

Make a resistive divider which has a parallel impedance of 50 Ohms and attach that to the output of the drive stage. You can get away with a 1:100 divider ratio. That way you make a passive divider probe which has a very high bandwidth. Make sure you use resistors able to handle some power.

[dzseki]

components), but want to measure also what they are capable. Now the first two stages should have a bandwidth near to 200MHz but there isn't much components there which could be altered -so there it's just curiousity. The last stage is the CRT driver which probably has about 100-130MHz BW, and the output signal is about 60-80Vpp. On that board there's quite a few things that I can play with, so the main task is to adjust this board for peak performance.

Make a resistive divider which has a parallel impedance of 50 Ohms and attach that to the output of the drive stage. You can get away with a 1:100 divider ratio. That way you make a passive divider probe which has a very high bandwidth. Make sure you use resistors able to handle some power.

Not sure what do you want to suggest here, the CRT driver is not capable of driving 50 Ohm DC resistance at all.

[dzseki]

Ground lead inductance might still be in issue. Of course alligator clips are out, you should only use low inductance grounds (eg. the springs you put around the ground collar). Is the point in the circuit you use as ground close to the signal, or is it some random other trace that happens to be connected to ground? Is the video signal terminated in both the measurement with direct BNC connection and with the 10x probe? Otherwise (depending on the length of the traces/cable from the driver) you might get nasty reflections that mess up your signal.

I achieved the 2ns with the cable directly coming from the VGA card, the BNC end was terminated by a 301 Ohm and a 100 Ohm parallel (sum.: 75 Ohm) and touching with the probe's tip and ground barrel directly to the legs of the resistors. Adding a short (2 inch) ground lead to the probe, and the clamping tip made the measurement even worse.

[nctnico]

components), but want to measure also what they are capable. Now the first two stages should have a bandwidth near to 200MHz but there isn't much components there which could be altered -so there it's just curiousity. The last stage is the CRT driver which probably has about 100-130MHz BW, and the output signal is about 60-80Vpp. On that board there's quite a few things that I can play with, so the main task is to adjust this board for peak performance.

Make a resistive divider which has a parallel impedance of 50 Ohms and attach that to the output of the drive stage. You can get away with a 1:100 divider ratio. That way you make a passive divider probe which has a very high bandwidth. Make sure you use resistors able to handle some power.

Not sure what do you want to suggest here, the CRT driver is not capable of driving 50 Ohm DC resistance at all.

That is where the divider comes in. At 1:100 the CRT driver sees about 5000 Ohms. You know how to make a voltage divider with resistors?

[dzseki]

That is where the divider comes in. At 1:100 the CRT driver sees about 5000 Ohms. You know how to make a voltage divider with resistors?

I know how to do that, what I don't know if there needs a parallel trimmer cap for compenstaion?

[dfmischler]

[I know how to do that, what I don't know if there needs a parallel trimmer cap for compenstaion?

No, capacitance is what you are trying to avoid in this case.

[dzseki]

Back to the starting point....

Right after I got the scope I realized that the P6106 probe has a compensation range of 15pf to 24pf scope input capacitance, but the scope has a 11pf input capacitance. But I ignored this as I was able to adjust the compensation of the probe -almost.
Actually the R3 C2 adjustments have no visual effect on the screen (seen on attachment), otherwise the the response seems to be fine, should this be also an issue?