EEVblog #453 - Mysteries of x1 Oscilloscope Probes Revealed

[EEVblog]

Dave cracks out the whiteboard and side cutters to explain two lesser known facts about x1 oscilloscope probes:
- Why the bandwidth is much lower in x1 mode than x10 mode
- Why oscilloscope probe cables aren't like ordinary coax cables

[BravoV]

To complement this nice video, a nice article written by Dave's mate Doug Ford ..

-> The Secret World of Oscilloscope Probes (PDF), recommended for download. 

[ConKbot]

I was going to say Ive seen a pdf on the high frequency loss transmission line bit of oscilloscope probes before, but couldnt find it, glad to see its handled already.

[Winston]

Doubleplus good video!  I never would have guessed that lossy transmission line cause.

[free_electron]

The reason for having x1 probes is simple: measuring in the millivolts range...
You dont want to atteniate a signal and then send it through an amplifier at the backend.

[nitro2k01]

Am I right in saying that the equivalent circuit for this type of coax is reminiscent of a Butterworth filter topology?

[CarlG]

Very interesting video! I knew that probe cables aren't ordinary coaxes, and that one shall treat them with care (as well as ordinary coaxes!) but I haven't bothered to find out about the details. Didn't know about the wiggling of the center lead, for example. Gonna tip my coworkers about this video

[KedasProbe]

So is it easy to attach a normal coax to a cheap probe? (to make a separate X1 probe)
And how much more bandwidth can be expected.
or maybe the 9Mohm can be replaced with an 1Mohm?

I assume there does not exist a very cheap 1X only probe with like 50Mhz bandwidth.
or maybe X2 with higher bandwidth.


Edit: this is claimed to have a bandwidth for 60Mhz (no probe just clips)
http://www.amazon.com/P1011-Alligator-Oscilloscope-Probe-Coaxial/dp/B00AK9GCQW
I bought 2 x100Mhz probes for only 5 euro each on ebay, will do some 'surgery' myself.
See if I can make a X2 probe.

Edit2: found this: X2 500Mhz (very expensive >500Euro)
TPP0502  http://www.tek.com/sites/tek.com/files/media/media/resources/TPP1000_TPP0500_TPP0502_Passive_Voltage_Probes_Datasheet_51W-26151-3.pdf

[alm]

So is it easy to attach a normal coax to a cheap probe? (to make a separate X1 probe)
And how much more bandwidth can be expected.

Dave simplified (in my opinion oversimplified) this to stay away from transmission line theory. For best high frequency performance, you want to terminate a 50 ohm coax cable in 50 ohms, otherwise you get reflections and ringing. The scope input is 1 Mohm. There is no such thing as 1 Mohm coax cable. Therefore resistive coax is necessary (not just in 10x probes!) to dampen these effects. Attaching a piece of 50 Ohm coax to a 1 Mohm scope input will have a higher bandwidth, but it will also have some nasty ringing in the pass band. The amplitude of signals at certain frequencies might go up by a factor of two! That can introduce some severe distortion. It will also present a huge capacitive load to your circuit under test.

Another important issue is the source impedance. Probes are usually specced when driven from a terminated 50 ohm source, with a 25 ohm source impedance (50 ohm source in parallel with 50 ohm termination). The 100 pF 1X probe will have an impedance of about 160 ohm at 10 MHz, so the voltage at the probe tip will be about 85% of the open circuit voltage due to probe loading. Now connect the same probe to a voltage divider consisting of two 5 kohm resistors across the same 10 MHz sine wave. The source impedance of this point is 2.5 kohm. The signal at the probe tip is now only 6% of the open circuit voltage. This is why 1x probes can be useless even for 1 MHz signals.

or maybe the 9Mohm can be replaced with an 1Mohm?

Only if you also change the rest of the compensation network.

I assume there does not exist a very cheap 1X only probe with like 50Mhz bandwidth.

Not at 1 Mohm. 1X-only probes will use the same resistive coax (it's not just there to optimize the 10X probe) and will have similar ~10 MHz / 100 pF specs. You can get more bandwidth with low-impedance probes, like the 50 ohm coax that Dave suggested. A piece of 50 ohm coax terminated into 50 ohms will have a very high bandwidth, but the input impedance is only 50 ohms.

Edit: this is claimed to have a bandwidth for 60Mhz (no probe just clips)

Terminated into 50 ohms maybe. Terminated into 1 Mohm no.

Some further reading: Tektronix Oscilloscope Probe Circuits. By far the most detailed work on scope probes. Also contains some information about 1x probes.

[EEVblog]

Dave simplified (in my opinion oversimplified) this to stay away from transmission line theory.

Yes, of course. The video was already long enough, I thought it wasn't the place for transmission line theory or simulation.

[alm]

I agree, but claiming that the resistive coax is only there to optimize 10x performance is wrong. The issues that you mention for 10x probes (like ringing in the pass band) also exist for 1x probes.

[EEVblog]

I agree, but claiming that the resistive coax is only there to optimize 10x performance is wrong.

I did not mean to imply that (sorry, can't remember my exact words off-hand), what I was saying is that the resistive coax design and the associated compensation network are carefully designed to be optimised for the x10 mode, in order to get the greatest bandwidth in x10 mode, as that is the main performance criteria for the probe. The x1 mode simply shorts out the 9M resistor and comp cap in the probe, and then you "get what you get" in terms of x1 performance. Yes, it will help dampen the signal in x1 mode too.

[alm]

A switchable probe is definitely a compromise, which is why you don't see many 500 MHz switchable scope probes. A dedicated 1x probe will have almost all of the issues you mention, though. It has the same (or at least very similar) resistive coax and parasitic capacitance. You don't generally find 1x hi-Z probes with >> 10 MHz bandwidth either.

For a future episode it might be instructive to compare a piece of Z₀ coax and a piece of resistive coax, both terminated in 1 Mohm // 15 pF. You could do either simulation, actual measurements or both. Should only take you about 5 minutes .

[JackOfVA]

The resistive loss cable probe was invented at Tektronix with the key patent filed in 1956. Well worth reading -- http://www.google.com/patents/US2883619

Interestingly, HP had a patent issued many years later for a broadband scope probe with standard lossless (or at least negligible loss) cable. http://www.google.com/patents/US5172051

[EEVblog]

A switchable probe is definitely a compromise, which is why you don't see many 500 MHz switchable scope probes.

Yokogawa do a 400Mhz switchable, with inversely terrible rated x1
Can't say I've ever seen better than that.
http://tmi.yokogawa.com/ca/discontinued-products/oscilloscopes/voltage-probes/700988-400-mhz-passive-probe/

[LapTop006]

The reason for having x1 probes is simple: measuring in the millivolts range...
You dont want to atteniate a signal and then send it through an amplifier at the backend.

Eh, as long as the signal's hot enough that's fine.

I wonder why FET probes aren't more common then, the parts shouldn't be much more then for a 10X probe, especially as the cable might be able to be simplified (that's not actually clear after some quick research).

Sure the current lot of cheap scopes don't have the aux power pins, but I wouldn't be surprised to see them in a future rev, the cost to add them is minimal for another "premium" feature.

[JackOfVA]

The reason for having x1 probes is simple: measuring in the millivolts range...
You dont want to atteniate a signal and then send it through an amplifier at the backend.

Eh, as long as the signal's hot enough that's fine.

I wonder why FET probes aren't more common then, the parts shouldn't be much more then for a 10X probe, especially as the cable might be able to be simplified (that's not actually clear after some quick research).

Sure the current lot of cheap scopes don't have the aux power pins, but I wouldn't be surprised to see them in a future rev, the cost to add them is minimal for another "premium" feature.

FET input gates are limited for maximum voltage to reasonably low values, some tens of volts at most and usually closer to +/- 10 or 20V. So long as the signal to be probed is less than maximum gate input voltage all is good. But, when the voltage exceeds the maximum gate voltage the probe will not faithfully reproduce the input voltage. And, at some level of input voltage, the protection mechanisms the probe designer included will break down leading the magic smoke contained inside the probe to be vented to the atmosphere, often accompanied by cursing from the probe owner.

To attenuate the voltage, therefore, you can't use the oscilloscope input attenuator / voltage range knob over much range (if at all), but rather have to apply attenuation directly at the probe tip. There are a few attenuated FET probes available and I seem to recall one FET probe with an optional 10x or 100x attenuator sleeve that fitted over the probe  body, but these are not all that satisfactory a solution to varying the combined probe/oscilloscope vertical gain. For the obvious practical reasons, it is both undesirable and difficult to build the attenuator control into the body of an  FET probe.

[G7PSK]

I measured my probe leads in the times one setting and found that they were only 100.1 ohms I have another od probe that was 346 ohms in the X1 My hundred times probe could not be measured with a DVM so I will have to get the megger to that.

[sonic]

For a future episode it might be instructive to compare a piece of Z₀ coax and a piece of resistive coax, both terminated in 1 Mohm // 15 pF. You could do either simulation, actual measurements or both.

I tried to remove over-/undershot using a Z₀ probe and made screenshots, but I'm having difficulties interpreting the results, since I'm not exactly knowing what I'm doing

[sean87]

So, that being said, can someone explain, why there is 1x and 10x after all? what is the particular use of each? why not having it fixed on 10x ?

[alm]

10x is the most commonly used because it has much more bandwidth and reduced circuit loading compared to 1x. The downside is that it attenuates the signal ten times, which is not a very good idea for low level signals. The 1x mode is traditionally used for power supply ripple measurement, for example. This ripple tends to be 120 Hz or so, with amplitudes in the mV.

But 10x is by far the most versatile. I use 10x probes by default and only occasionally reach for a 1x probe.

[ddavidebor]

Because you want to misure small signal of a few millivolt sometimes, and you need the lowest vertical range possible.

So if you are measuring a 8mV signal, in 1mV range with x10 probe you can't do it, you need x1 probe

[EEVblog]

10x is the most commonly used because it has much more bandwidth and reduced circuit loading compared to 1x. The downside is that it attenuates the signal ten times, which is not a very good idea for low level signals. The 1x mode is traditionally used for power supply ripple measurement, for example. This ripple tends to be 120 Hz or so, with amplitudes in the mV.

One point about that is that most PSU measurements are specified over a 20MHz bandwidth, so outside the range of usual run of the mill x1 probes.
Usually near enough, but if you want to do it properly, you'll need something else.

[alm]

Don't confuse ripple and noise. Ripple will usually have a fundamental frequency of 100 or 120 Hz for linear power supplies with full bridge rectification, and might be in the tens of kHz for switchers. Measuring ripple on power supplies is common during repair to check for dried out electrolytics.

[Salas]

You could tell a few words about the falling with frequency voltage spec on probes which many may be not aware of. A 300V at 10kHZ probe can be 100V at 1MHZ & 10V at 100MHZ...