¿ªÔÆÌåÓý

  1. HP-Agilent-Keysight-Equipment
  2. Topics

Testing Scope Probes

 

Hi all,
There's some seller on Ebay offering Agilent 10073C oscilloscope probes at a knock-down price and naturally I bought one like a complete idiot would. Anyway, this is supposed to be a 500Mhz passive probe so I decided to compare it against my existing selection of passive probes just to get some idea if that bandwidth really was as advertised. So I set up my Marconi 5.4Ghz RF signal generator for 50-500Mhz output and tested the probes against it one by one. Obviously the lower BW probes I have couldn't make the grade and fell flat on their faces well before the half-gig mark and some before 100Mhz even. That was of course expected, however, the Ebay one's performance was no better and possibly a bit worse than an old 150Mhz Tek probe I had lying around and it simply could not cut it even at 300Mhz, let alone 500.
However, before I shoot my mouth off and complain about this seller sending me a fake Agilent probe, I thought I'd best check with the Panel here to make sure I haven't forgotten to allow for anything in my testing. The scope I am using for this test is a 350Mhz Tek analog one and it can 'see' waveforms perfectly well at half gig when connected with a high bandwidth probe (my 3.5Ghz one) so I know the scope's not at fault.
Any suggestions?
 

What is bandwidth?

The classic definition is when the measured voltage is at the
half power point, which is 0.707 x V (to be measured).

So, your 350MHz scope should show at least 0.707V for a 1V signal
at 350MHz. Manufacturers typically try to do a little better than
what they specify...

500MHz is nearly an octave above the scope's rated bandwidth. What
you measure there could be zero, and it would not be out of spec.
More probably, it will have peaks and valleys all over the map above
the rated bandwidth. But probably would be about 0.25V, with most
of the scopes I have used.

The probe is 10M, and it a real work of art to get such a high impedance
to pass anything at 500MHz. It can't be done without treating the
1M input of the scope and the 10M probe as a unit as far as adjusting
the scope goes. Typically, any probe above 100MHz has to be designed
and adjusted to work with the host scope.

Clearly these probes were made to some other standard than your
350MHz scope.

My rule for scope measurements is don't use a 10M 10x probe for
any measurements above 50MHz, and then only if the probe is grounded
right at the circuit with a short and direct connection to the sleeve
at the probe tip... Never with the ground clip lead!

-Chuck Harris

On Tue, 09 Apr 2024 11:44:41 -0700 "Jinxie" <paul666@...>
wrote:
Hi all,
There's some seller on Ebay offering Agilent 10073C oscilloscope
probes at a knock-down price and naturally I bought one like a
complete idiot would. Anyway, this is supposed to be a 500Mhz passive
probe so I decided to compare it against my existing selection of
passive probes just to get some idea if that bandwidth really was as
advertised. So I set up my Marconi 5.4Ghz RF signal generator for
50-500Mhz output and tested the probes against it one by one.
Obviously the lower BW probes I have couldn't make the grade and fell
flat on their faces well before the half-gig mark and some before
100Mhz even. That was of course expected, however, the Ebay one's
performance was no better and possibly a bit worse than an old 150Mhz
Tek probe I had lying around and it simply could not cut it even at
300Mhz, let alone 500. However, before I shoot my mouth off and
complain about this seller sending me a fake Agilent probe, I thought
I'd best check with the Panel here to make sure I haven't forgotten
to allow for anything in my testing. The scope I am using for this
test is a 350Mhz Tek analog one and it can 'see' waveforms perfectly
well at half gig when connected with a high bandwidth probe (my
3.5Ghz one) so I know the scope's not at fault. Any suggestions?
 

On Tue, Apr 9, 2024 at 08:44 PM, Jinxie wrote:
The scope I am using for this test is a 350Mhz Tek analog one and it can 'see' waveforms perfectly well at half gig when connected with a high bandwidth probe (my 3.5Ghz one) so I know the scope's not at fault.
Any suggestions?
1. As Chuck says, you don't really know the behavior of your "500MHz" 10073C on your 350MHz analog 'scope (much) above that 350MHz. Surely, the falloff of the combination isn't Gaussian, as it's supposed to (more or less) be from below 350MHz (with some bumps and valleys) to 350MHz.

2. The input impedance spec. of your "10MOhm" 10073C is around 12pF. So, at 500 MHz, it presents a load of around 26Ohm to the generator! Your generator would need to have an output impedance far below its spec'ed (?) 50Ohm if you don't want its output signal to be significantly reduced by the load from the probe. ISTR that Hi-Z probes are spec'ed assuming a nominal 20Ohm driving impedance.

Your 3.5GHz probe either is a lo-Z or a FET probe. If the latter, its input capacitance would be a few pF, significantly lower that the passive 10073C. That's the very "raison d' ¨ºtre" for a FET probe...

Raymond
 

If it helps, I've been using the 10073 series probes for many years. The 10073C scope probe isn't a 10Meg probe, it's more like 2.2Meg. Try measuring it (with a DMM set to ohms) with it connected to a regular 1Meg scope input.

I think you will see 2.2Meg and not 10Meg.

Also, try compensating the probe and have a look at the shape of the leading edge on the scope display as you adjust the compensation. On the scope display a?10073C will look very different to a run of the mill x10 scope probe when adjusting the compensation. The leading edge of the waveform should be much, MUCH 'tighter' with the 10073C. This is another signature for a genuine 10073C.

You can also try measuring the rise-time of the probe?
 

On Tue, Apr 9, 2024 at 10:09 PM, jmr wrote:
The 10073C scope probe isn't a 10Meg probe, it's more like 2.2Meg.
Yes, according to the datasheet it's 2.2MOhm (DC). Made completely insignificant against its 500MHz impedance of less than 30Ohm.
Its frequency compensation range is specified as 6 - 15pF, as is usual for the higher-BW HP 'scopes (11pF ISTR). The adjustment range makes these probes incompatible with most other general purpose 'scopes. As an example, Tek 'scopes are in the range 15 - 20pF (ISTR) at hi-Z.

Raymond
 

At 500MHz, it's not completely insignificant that the 10073C is a 2.2M probe. By designing for a 2.2Meg input divider (and not a 10Meg divider)? it makes it much easier to achieve good compensation across a 500MHz bandwidth. I assumed it was common knowledge that these are not 10M probes.?

So the first simple test the OP should be doing is to see if it s a 2.2Meg probe.
 

On Tue, Apr 9, 2024 at 10:42 PM, jmr wrote:
At 500MHz, it's not completely insignificant that the 10073C is a 2.2M probe. By designing for a 2.2Meg input divider (and not a 10Meg divider)? it makes it much easier to achieve good compensation across a 500MHz bandwidth. I assumed it was common knowledge that these are not 10M probes.?

So the first simple test the OP should be doing is to see if it s a 2.2Meg probe.
Except that at 500MHz, a source sees only the 26Ohm, not 2.2MOhm (nor 10MOhm).

What would the benefit to the OP be of testing that the 10073C's input resistance is according to its DC spec?

Raymond
 

I recommend measuring it to see if it is a fake probe? A fake version might have a 10M input. If it is 10M then it's safe to assume it's a fake without doing any more tests.?

Hopefully, Jinxie already knew the 10073C is not a 10M scope probe so maybe this test has already been done? I assumed anyone with experience of HP scopes/probes would know these classic probes are not 10M probes.
 

On Tue, Apr 9, 2024 at 11:11 PM, jmr wrote:
I recommend measuring it to see if it is a fake probe? A fake version might have a 10M input. If it is 10M then it's safe to assume it's a fake without doing any more tests.?
Oh yes, didn't take that into account, of course.

Raymond
 

The other thing to do is to make sure you are compensating it correctly. Also, you should see quite a difference in the response of the compensation if it is a genuine 10073C compared to a regular 10M probe. The 10073C compensation should appear tighter as it is adjusted as it is compensating for a lower Z divider in the probe.

The compensation is much easier for a 2.2Meg probe and therefore it's much easier to get a consistent frequency response with a genuine 10073C probe.
A regular 10Meg probe will be fairly awful in comparison.?
 

I have no access to the probe until tomorrow morning. I have not as yet tested to check if it's 2.2M ohms, which sounds like a very simple quick and dirty test and would eliminate it as real or fake instantly.
It's a shame, as I have a 3Ghz VNA which would be a great tool for testing the bandwidth, but the probe's impedance is nowhere near 50 ohms and that's what the analyser is expecting!
I've got a TDR board somewhere too with a 30ps rise time (IIRC).
All the comments are very helpful in giving me ideas as to how to approach the testing; keep 'em coming; thanks, guys.
 

Also, when testing with your sig gen you need to terminate the sig gen with a good 50R load. I'd recommend using a short test PCB with some 50R microstrip on it with a good 50R termination at the far end. Then fit this directly at the sig gen output and probe directly with the 10073C using the tip ground and not the clip ground.?

It's a while since I used a Mi 2032/42 but the sig gen should be OK when doing a test like this. I'm assuming that your scope is a Tek 2465 or similar? I think these old scopes only have 300MHz - 400MHz bandwidth but they should be OK with a 10073C probe as long as you compensate it correctly. I'm not sure how much bandwidth you will see with the 10073C with that setup though.
With a modern 500MHz or 1GHz BW scope I'd expect to see good performance from a genuine 10073C.

If it helps, I've got four genuine 10073C probes here if you want to see any close up images? I can do some tests with them but this will just replicate what the datasheet already shows in terms of bandwidth and rise-time etc.
 

It's worth adding a 10dB pad at the sig gen output to make sure the source Z is close to 50R and then the effective source impedance driving the probe will be about 25 ohms with the output termination included. This should give reasonable results with a 500MHz or 1GHz scope and a 10073C probe. I'd expect to see some bandwidth degradation with the 350MHz BW scope (Tek 2465 or similar?).
 

Thanks for the offer, JMR, I'll see how I get on with some other tests before imposing on you.
I have asked a colleague to measure the probe impedance at DC and it's a whisker over 2.2M ohms, so that's good.
I'm just wondering from what's been said in the comments and the literature for the probe whether this type is just incompatible with the scope? The scope's Zin is 15pf || 10M ohms so perhaps that's not what this probe is expecting to 'see' as it were. I'm a bit behind the times and was under the impression that any passive probe would work with any analog scope. If that's not the case, what kind scope would this probe be intended for use with I wonder and could this incompatibility be the cause of this? issue?
 

Yes, it's a Tek 2465A; the highest bandwidth scope I have.
 

On 4/9/24 16:29, Jinxie wrote:
what kind scope would this probe be intended for use with I wonder and could this incompatibility be the cause of this? issue?
It's made for HP scopes and logic analyzers like the 16500C 16700B... Could be...
 

The 10073C should have come with a BNC adaptor sleeve. So should your "good" probe. To eliminate all loading and source imedance issues connect a BNC T to the signal generator output and directly connect each probe (with BNC sleeve) to the outputs of the T. Ideally place an attenuator (3 to 10dB) between generator and T to improve matching of the generator. Be wary of "thru" BNC terminators many are only good to relatively low frequencies. Connect each probe to a separate input of the 'scope. You can then directly compare the responses of the two probes. As a double check swap the probes between the two channels. Don't forget to check the compensation.
The 10073C will be near the end of it's adjustment range but should be OK the 2465A. Just make sure you can adjust past the optimum point so haven't run out of range.

Robert.
 

I'll try that, Robert, thanks. It'll probably not be today owing to time pressures but I'll report back here in due course...
 

Sorry - need a bit of clarification here. "BNC adaptor sleeve" - what is that? Does it fit over the probe tip or something?
 

The problem with trying to do this with two probes in parallel (using a BNC T piece) is that you will end up with double the loading so I'd expect this test method to be doomed to failure if the aim is to see the expected bandwidth of the scope + probe. Also, feeding the BNC T directly from a 50R source will just make it even worse. Ideally, the source impedance needs to be quite low. 50R is too high for testing a 500MHz probe.? You could end up seeing a BW of 150MHz rather than the expected BW if you use Robert's method as described.?

1 - 20 of 126