-
Probing at high speeds. Will the 10x probe overload the circuit?
Posted by ballsystemlord on 08 May, 2024 05:05 -
Hello,
So when I first learned about probing, I learned that you should use the 10x setting of the probe whenever possible so that you don't load down your circuit very much.
Recently, I came upon a video by Mr. Ganssle.
https://youtu.be/aJsJibDNg9M
He shows that a 10pF capacitor goes from over 300ohms of capacitive reactance to just 159ohms of capacitive reactance.
What happened to the 9Mohm resistor in the probe?
Like, it should still be a 10 to 1 input impedance, right?
Formula: sqrt((1/(PI×2×100 000 000×0.000 000 000 01))^2+(9 000 000^2))
He goes on to discuss using just a piece of coax and 1K resistor.
Wouldn't that need some sort of capacitive compensation?
And if this really is *the answer* to high speed probing, why are we using 10x probes at all? Why not just a resistor hanging off of a piece of coax with maybe an attached pogo-pin and alligator clip on the ground terminal?
Thanks!
-
Hello,
The capacitance is effectively in parallel with the 10M
So when I first learned about probing, I learned that you should use the 10x setting of the probe whenever possible so that you don't load down your circuit very much.
Recently, I came upon a video by Mr. Ganssle.
https://youtu.be/aJsJibDNg9M
He shows that a 10pF capacitor goes from over 300ohms of capacitive reactance to just 159ohms of capacitive reactance.
What happened to the 9Mohm resistor in the probe?
Like, it should still be a 10 to 1 input impedance, right?
Formula: sqrt((1/(PI×2×100 000 000×0.000 000 000 01))^2+(9 000 000^2))
input resistance so the capacitive reactance will dominate and the input impedance will be ever so slightly less than 159 capacitive reactance. The formula you have used would apply for the series connection of a capacitor and resistor, which is not what a probe looks like.Quote
Such probes are really only useful for higher frequency signals e.g. >100Mhz and the lower impedance circuits typically found with high speed signals. There are commercial probes available, known as Z0 probes, that have a series 450
He goes on to discuss using just a piece of coax and 1K resistor.
Wouldn't that need some sort of capacitive compensation?
And if this really is *the answer* to high speed probing, why are we using 10x probes at all? Why not just a resistor hanging off of a piece of coax with maybe an attached pogo-pin and alligator clip on the ground terminal? resistor and have a x10 attenuation when coupled to a scope with a 50 input. Since the cable is terminated at the scope with the cable characteristic impedance it still looks like 50 at the probe end of the cable and doesn't have the excess cable capacitance that is seen with cable terminated by the 1M input of a scope as for a 10M probe. Without the excess capacitance there is no need for compensation.
However such probe present an excessive loading, especially DC loading, on lower frequency and higher impedance circuits so are not generally suitable for most circuit probing needs. -
Like, it should still be a 10 to 1 input impedance, right?
It will still be 10:1 division ratio, where the voltage is measured at the probe tip.
But note that a 600ohm output, for example, driving a *10 probe will have a reduced voltage at the probe tip.QuoteHe goes on to discuss using just a piece of coax and 1K resistor.
Wouldn't that need some sort of capacitive compensation?
And if this really is *the answer* to high speed probing, why are we using 10x probes at all? Why not just a resistor hanging off of a piece of coax with maybe an attached pogo-pin and alligator clip on the ground terminal?
There is no single answer to probing, since all probes become part of the circuit and all alter the operation/measurement in one way or another. You need to match the class of probe to the circuit and measurement.
Even if using a standard *10 "high" impedance probe, a 6" ground lead makes it unsuitable for high speed circuits. With every probe, a 150nH lead inductance will resonates with the tip impedance; for a 6" lead plus 15pF, that is about 100MHz. For an example, see
https://entertaininghacks.wordpress.com/2015/04/23/scope-probe-accessory-improves-signal-fidelity/
https://entertaininghacks.wordpress.com/2016/09/17/scope-probe-accessory-higher-frequency-results/
The different classes of probe - and why they need to exist - are discussed in the references at https://entertaininghacks.wordpress.com/library-2/scope-probe-reference-material/ -
Such probes are really only useful for higher frequency signals e.g. >100Mhz and the lower impedance circuits typically found with high speed signals. There are commercial probes available, known as Z0 probes, that have a series 450
resistor and have a x10 attenuation when coupled to a scope with a 50 input. Since the cable is terminated at the scope with the cable characteristic impedance it still looks like 50 at the probe end of the cable and doesn't have the excess cable capacitance that is seen with cable terminated by the 1M input of a scope as for a 10M probe. Without the excess capacitance there is no need for compensation.
However such probe present an excessive loading, especially DC loading, on lower frequency and higher impedance circuits so are not generally suitable for most circuit probing needs.
Whether the loading is "excessive" depends entirely on the circuit. Some resistive divider probes have a 5kohm tip resistance.
Z0 resistive divider probes present a more constant loading, which is often more important that the absolute value.
You need to match the probe class to the circuit and measurement; no surprises there!
-
In addition to the signal and the circuit (impedance), which probe to use also depends on what you want to measure. I remember an old Tektronix training video that can probably be found on YouTube saying that for best amplitude accuracy, a FET probe with its high impedance over most of the frequency band is best (less loading that reduces amplitude). But for best rise time accuracy, a Z0 probe with its very flat impedance over frequency performs better.
-
Hello,
...
He goes on to discuss using just a piece of coax and 1K resistor.
Wouldn't that need some sort of capacitive compensation?
And if this really is *the answer* to high speed probing, why are we using 10x probes at all? Why not just a resistor hanging off of a piece of coax with maybe an attached pogo-pin and alligator clip on the ground terminal?
Thanks!
When I think of high speed, I am thinking digital. And yes the resistive divider you mention is common. The following thread may give you some insight:
https://www.eevblog.com/forum/testgear/12-ghz-active-probe-project/msg4973839/#msg4973839
-
When I think of high speed, I am thinking digital.
Strictly speaking it doesn't matter. If you are looking at a waveform with a scope you are looking at an analogue waveform that something will interpret as a digital signal.
Exceptions: photon counting and femtoamp circuits
Relevance: all the probing considerations apply unchanged to both "digital" circuits and "analogue" circuits. -
When I think of high speed, I am thinking digital.
Strictly speaking it doesn't matter. If you are looking at a waveform with a scope you are looking at an analogue waveform that something will interpret as a digital signal.
Exceptions: photon counting and femtoamp circuits
Relevance: all the probing considerations apply unchanged to both "digital" circuits and "analogue" circuits.
In the context of loading and drive, how much it can tolerate, it most certainly matters. -
When I think of high speed, I am thinking digital.
Strictly speaking it doesn't matter. If you are looking at a waveform with a scope you are looking at an analogue waveform that something will interpret as a digital signal.
Exceptions: photon counting and femtoamp circuits
Relevance: all the probing considerations apply unchanged to both "digital" circuits and "analogue" circuits.
In the context of loading and drive, how much it can tolerate, it most certainly matters.
What do you think "most certainly matters"?
The only significant parameters are signal impedance (usually 50 or ~100 ohms) and energy as a function of frequency (closely related to slew rate).
Binary digital signals are 0 or 1, ternary digital signals are 0,1,2 (or -1,0,1 if you prefer), quaternary... They are all represented by a wide variety of analogue voltages/currents, and baud rates (and separately, bit rates). Oscilloscope show the analogue voltages. -
What do you think "most certainly matters"?
Anything that effects the SI. -
What do you think "most certainly matters"?
Anything that effects the SI.
Precisely. SI is an analogue phenomenon not digital, and the same issues also occur in analogue circuits. -
Relevance: all the probing considerations apply unchanged to both "digital" circuits and "analogue" circuits.
Do you really think rise time and amplitude are equally important in analog and (binary) digital signals? Is amplitude a critical parameter for an accurate eye diagram? Is rise time important in what's generally a more narrow band analog signal like RF? -
Relevance: all the probing considerations apply unchanged to both "digital" circuits and "analogue" circuits.
Do you really think rise time and amplitude are equally important in analog and (binary) digital signals? Is amplitude a critical parameter for an accurate eye diagram? Is rise time important in what's generally a more narrow band analog signal like RF?
Firstly that is a separate discussion.
Secondly yes they are important.
Thirdly, not all RF signals are narrowband.
Fourthly, the distinction between RF and digital is an artificial construct derived from historic engineering disciplines, not physics. It is as unreal as the distinction between hardware and software.
Fithly, the correspondence between the time and frequency domains is valid without reference to digital, RF, DSP, or any job title.
And I could continue, but I doubt that discussion would help the OP understand the different classes of probes and how they affect the circuit they are part of. -
What do you think "most certainly matters"?
Anything that effects the SI.
Precisely. SI is an analogue phenomenon not digital, and the same issues also occur in analogue circuits.
As I stated, it just provides context which I thought would have been obvious. Sure we can become pendant but it doesn't add anything to the discussion. -
What do you think "most certainly matters"?
Anything that effects the SI.
Precisely. SI is an analogue phenomenon not digital, and the same issues also occur in analogue circuits.
As I stated, it just provides context which I thought would have been obvious. Sure we can become pendant but it doesn't add anything to the discussion.
The key distinction between digital signals (0,1,etc) and analogue signals (volts, amps, frequency, etc) is not pedantry.
False distinctions between analogue waveforms and RF/microwave etc is not pedantry.
Failing to grok those is the source of much bafflement, many incorrect statements (especially w.r.t. sampling rate), and many problems seen all too often in circuits (especially SI).
Hence it is important, not mere pedantry. -
When you consider details, digital is analog. I use the term digital to provide context about what we are probing. Again not 0's and 1's but loading, drive strengths, transition levels, overshooot.... As you previously posted:What do you think "most certainly matters"?
Anything that effects the SI.
Precisely. SI is an analogue phenomenon not digital, and the same issues also occur in analogue circuits.
As I stated, it just provides context which I thought would have been obvious. Sure we can become pendant but it doesn't add anything to the discussion.
The key distinction between digital signals (0,1,etc) and analogue signals (volts, amps, frequency, etc) is not pedantry.
False distinctions between analogue waveforms and RF/microwave etc is not pedantry.
Failing to grok those is the source of much bafflement, many incorrect statements (especially w.r.t. sampling rate), and many problems seen all too often in circuits (especially SI).
Hence it is important, not mere pedantry.QuoteStrictly speaking it doesn't matter. If you are looking at a waveform with a scope you are looking at an analogue waveform that something will interpret as a digital signal.
Much like my own post so far, I suspect these comments are too high level and obvious. They add little to no value to the discussion. -
When you consider details, digital is analog. I use the term digital to provide context about what we are probing. Again not 0's and 1's but loading, drive strengths, transition levels, overshooot.... As you previously posted:What do you think "most certainly matters"?
Anything that effects the SI.
Precisely. SI is an analogue phenomenon not digital, and the same issues also occur in analogue circuits.
As I stated, it just provides context which I thought would have been obvious. Sure we can become pendant but it doesn't add anything to the discussion.
The key distinction between digital signals (0,1,etc) and analogue signals (volts, amps, frequency, etc) is not pedantry.
False distinctions between analogue waveforms and RF/microwave etc is not pedantry.
Failing to grok those is the source of much bafflement, many incorrect statements (especially w.r.t. sampling rate), and many problems seen all too often in circuits (especially SI).
Hence it is important, not mere pedantry.QuoteStrictly speaking it doesn't matter. If you are looking at a waveform with a scope you are looking at an analogue waveform that something will interpret as a digital signal.
Much like my own post so far, I suspect these comments are too high level and obvious. They add little to no value to the discussion.
Lewis Carroll referred humourously to similar issues 170 years ago, via his Humpty Dumpty character
The drive strength at the far (rx) end of a 50/100/etc ohm transmission line is the same whatever is at the near (tx) end.
Digits are captured in digital tools such as logic analysers, or protocol analysers, or printf() statements. Scopes capture, ahem, analogues.
A key principal is to use an (analogue) scope to observe (analogue) signal integrity, then flip to using a digital tool to observe digital signals. But you know that.
If beginners were taught that and understood it, fewer questions would need to arise - because they sorted out many problems on their own. Their remaining questions would be more interesting and less repetitive That's why I think it is important.
-
Suggest you read the Tektronix Circuits Concepts books:
Oscilloscope Probe Circuits
062-1146-00.pdf
Oscilloscope Probe measurements
https://w140.com/tekwiki/images/1/19/062-1120-00.pdf
also:
https://download.tek.com/document/51W_30013_0_MR_Letter.pdf
We use TEK P6136 Zo, P6137 10X and DIY Zo just a 50 Ohm coax and series small axial R 450 or 4950 Ohm at the tip.
Enjoy
Jon
-
Hello,
Typically it does need a bit of compensation but it greatly depends on what kind of accuracy you are after. Most high speed signals you encounter nowadays are driven from a source impedance between 40 to 60 Ohm. Loading that with a resistor of 1k Ohm and some capacitance can easely lead up to a difference in amplitude of 10% between the original (no probe attachted) signal and with a low-Z probe attached. But still, a low-Z probe is very suitable for use on high-speed signals. I'm using these quite a lot myself.
So when I first learned about probing, I learned that you should use the 10x setting of the probe whenever possible so that you don't load down your circuit very much.
Recently, I came upon a video by Mr. Ganssle.
https://youtu.be/aJsJibDNg9M
He shows that a 10pF capacitor goes from over 300ohms of capacitive reactance to just 159ohms of capacitive reactance.
What happened to the 9Mohm resistor in the probe?
Like, it should still be a 10 to 1 input impedance, right?
Formula: sqrt((1/(PI×2×100 000 000×0.000 000 000 01))^2+(9 000 000^2))
He goes on to discuss using just a piece of coax and 1K resistor.
Wouldn't that need some sort of capacitive compensation?
This page has some info on how high speed low-Z probes (basically a piece of coax with a resistor attached to it) are constructed including a DIY example:
https://emcesd.com/1ghzprob.htm
-
When ballsystemlord titled Probing at high speeds, they never defined what they were attempting to probe. I provided what I considered the topic to be about based on their mention of resistive probes, thinking the OP would provide more details. I really have no idea what that were actually after and assumed after reading some of the posts, lost interest. A shame really.
-
When ballsystemlord titled Probing at high speeds, they never defined what they were attempting to probe. I provided what I considered the topic to be about based on their mention of resistive probes, thinking the OP would provide more details. I really have no idea what that were actually after and assumed after reading some of the posts, lost interest. A shame really.
Agreed.
Quite a few people tried to extract information from him, to no effect.
Doesn't look good, but perhaps he has been "unavoidably detained in a real life situation". -
Note that up at RF frequencies, the loading effect of a typical 10Meg x10 scope probe isn't as simple as 10Meg in parallel with about 12-15pF.
The 10Meg Rp will fall quite steadily with increasing frequency. By 10MHz it could fall to about 20k ohm and by several hundred MHz it can fall to maybe 100 ohms Rp or less.
By contrast, the parallel capacitance Cp seen at the tip will start to fall through VHF and into UHF. It might be 12pF at 1MHz but this could easily fall to 8pF or even less up at 200MHz.
So once you get well into the VHF region, the Rp loading effect can also become significant. The quality of the compensation also matters as this can affect how a fast pulse appears on the scope display.
Generally speaking, x10 scope probes don't perform very well for critical stuff like this. It is possible to buy x10 scope probes that use 2Meg resistive dividers rather than 10Meg and this can help improve the high frequency response as flatter compensation should be easier to achieve. -
Note that up at RF frequencies, the loading effect of a typical 10Meg x10 scope probe isn't as simple as 10Meg in parallel with about 12-15pF.
...
It's not just the probes performance, the user also plays a part. I couldn't tell you how many times I have seen someone using that long ground clip lead and hooked tip. -
There was an interesting and long thread on https://groups.io/g/HP-Agilent-Keysight-equipment
This message shows the measured tip R and C for an HP10073 probe - a quite decent *10 "high" impedance probe.
https://groups.io/g/HP-Agilent-Keysight-equipment/message/143362
And this shows the same for an old Tek probe https://groups.io/g/HP-Agilent-Keysight-equipment/message/143360
There are more measurements later in that thread, too. -
I couldn't tell you how many times I have seen someone using that long ground clip lead and hooked tip.
Oh yes
It is always worth asking for a photo of the experimental setup, to rule out a likely cause being long leads and solderless breadboards.