How does an oscilloscope input do what it does?

[FriedMule]

So your scope has to accept 500V, 1.5GHZ, "no noise", "linear", true to the input,  with 50 Ohm termination and low burden Voltage.
How is that done?

I mean if you let 500V go through your scope, do you suddenly own a box with an amazing amount of magic smoke but if you divide the Voltage to i.e. 5V do you have the noise floor to make it all unusable.
You could maybe try to counter that by some elaborate circuit but then does your input signal get mangled and does in no way look like what is going in.
In short, how do they manage it? :-)

[ataradov]

There are reverse engineered front-end schematics. Including one by Dave for some Rigol scope.

All those characteristics are true,  but not all at the same time. There is no way a scope will accept 500 V on 50 Ohm termination. Typically it would be something like 5 V RMS max.

[FriedMule]

Okay, thanks. 5V does not sound as much, I mean what about 9V or even 12V, or what about a tube amplifier at 700V, you can use a 10:1 probe, but does that not change what you are able to detect?
I have tried to search for scope circuits but I did not find much, do you know a link or the right search terms? :-)

[ataradov]

Don't probe 9V or 12V on 50 Ohm termination. 50 Ohm is designed for a very specific use case. And you can apply pretty high voltage to 1 MOhm input, since the current is low.

Here is a topic and a video https://www.eevblog.com/forum/blog/eevblog-675-how-to-reverse-engineer-a-rigol-ds1054z/

[FriedMule]

I am wondering, how do a noob protect the gear against mistakes?

Probing 120V because the scope says 500V max
Not using 50 Ohm as default because the coaxial is 50 Ohm
Zapping the scope by choosing two different grounds with Voltage différance
And so on?

[ataradov]

They don't unless they get familiar with the equipment they are used. Once they fry a channel or two, they stop being noobs.

[FriedMule]

LOL! :-)
Could one not build a "noob-filter" I mean voltage limiter or maybe a simple probe-between thingy? :-)

[ataradov]

You can build all sort of devices, but they all would lead to performance degradation. Scope manufactures have no real incentive to do that, as absolute majority of customers know what they are doing.

And if you are talking about an external device, then how would a noob know they need one?

And most entry level scopes have no 50 Ohm termination anyway. So that's you protection.

[tautech]

I am wondering, how do a noob protect the gear against mistakes?

RTFM

Probing 120V because the scope says 500V max
Not using 50 Ohm as default because the coaxial is 50 Ohm
Zapping the scope by choosing two different grounds with Voltage différance
And so on?

Engage the grey matter between the ears. 

At 14 I first used a scope after reading up on how to in magazines....back in the 70's .....and not again until the 90's and many many since, not one ever damaged and used for a wide range of uses some several hundred volts but always with thought and planning beforehand.
Know your scope and its capabilities....RTFM then RTFM again, forwards backwards until you know it all.

[brabus]

LOL! :-)
Could one not build a "noob-filter" I mean voltage limiter or maybe a simple probe-between thingy? :-)

A very famous theorem says: no matter what protection you put on an oscilloscope input, there will always be someone able to fry it. NO. MATTER. WHAT.

Did I just invent this theorem? Yes. Does it work? Hell yeah.

[David Hess]

For 1 megohm inputs, a high impedance is placed in series with the input to limit the current into low leakage diode clamps.  The high impedance is bypassed with enough capacitance to prevent the input capacitance from attenuating the signal.

For 50 ohm inputs, something else has to be done.  What works is to place a current driven diode bridge in series with the input.  The effect is to limit the input current without compromising frequency response or input impedance.  The input voltage protection is then limited by the reverse breakdown voltage of the diodes and the voltage compliance of the current sources.  The limit in this case is that high frequency diodes have relatively low breakdown voltage but that is somewhat alleviated by being able to use several in series for each element, which has the benefit of dividing the capacitance improving the frequency response anyway.  I will not say that you could get to 500 volts of protection at GHz frequencies, but 10s of volts and maybe at least 100 volts is possible, with protection up to 500 volts feasible at some lower frequency where diodes with a higher breakdown voltage can be used.

Besides low impedance inputs, this scheme can also be used to protect low impedance outputs like the output of a function generator or any digital output.  I have thought about using it to protect the force outputs of an LCR meter which are usually easily damaged.

Tektronix used this method on their 350 MHz 485 oscilloscope and their 1 GHz 7A29 vertical amplifier.

[tggzzz]

Tektronix used this method on their 350 MHz 485 oscilloscope and their 1 GHz 7A29 vertical amplifier.

The 485 has the advantage of separate 1Mohm and 50ohm attenuators, and a mechanical relay to switch between the two. Overloading the 50ohm input causes the relay to switch it back to 1Mohm.

[PartialDischarge]

I have tried to search for scope circuits but I did not find much, do you know a link or the right search terms? :-)

Then you should visit w140.com and enjoy with all the scope schematics you want starting from the 60s

[radiolistener]

Okay, thanks. 5V does not sound as much, I mean what about 9V or even 12V, or what about a tube amplifier at 700V, you can use a 10:1 probe, but does that not change what you are able to detect?

The voltage limit specified for the oscilloscope input refers to DC current.

For AC current, the voltage limit drops very quickly with increasing frequency.

Usually oscilloscope has 5 Vrms max limit for AC current at 50 Ω input and about 40 Vrms at 1 MΩ input. Exact limit value depends on specific oscilloscope model.


Regarding 10:1 probe, it is the same, it's voltage rating specified for DC current. The voltage limit of 10:1 probe for AC is about 25-50 V.

If you apply 700 V AC to 10:1 probe it will be burned out and can damage your oscilloscope! Usually it release magic smoke at much lower voltage, for example 100 V AC is enough to burn out your 10:1 probe.

Also, 700 V AC is very dangerous. 700 V at 50 Ω load equals to 4900 W power. This is much more than microwave oven.

[David Hess]

Tektronix used this method on their 350 MHz 485 oscilloscope and their 1 GHz 7A29 vertical amplifier.

The 485 has the advantage of separate 1Mohm and 50ohm attenuators, and a mechanical relay to switch between the two. Overloading the 50ohm input causes the relay to switch it back to 1Mohm.

The 7A29 includes an RF relay in series with the input controlled by the same overload protection IC used in the 485.  In both cases, the current bridge protects against overload until the relay can be activated.

[FriedMule]

I do really appreciate all your answers and yes RTFM is important. And I have read a lot to avoid some of the worst:-)


But imagine this normal situation:

A noob wants to find out how to reduce the ripples in his working DIY PSU, it's a 17V 500mA.
To him does 10:1 attenuation sound a bit stupid because he does not need the higher frequency and do in fact think he needs the higher resolution.
He has learned from EEV, among others that he has to remember if he wants to get the right voltage value does he also have to think of the 50 Ohm termination, he knows his probe is 50 Ohm and you can maybe set the scope at 50 Ohm and 1:1.

Now does he put on the probes and his scoop is damaged?

[tggzzz]

Noob needs to read the spec, think, understand the spec -- and why the spec is the way it is.

Noob needs to relate the spec to the construction of the components.

Noob realises there is no short cut to wisdom!

[drussell]

He has learned from EEV, among others that he has to remember if he wants to get the right voltage value does he also have to think of the 50 Ohm termination, he knows his probe is 50 Ohm and you can maybe set the scope at 50 Ohm and 1:1.

Why do you keep insisting that scope inputs and probes are 50Ω ?!   

The normal input impedance of most oscilloscopes is 1,000,000Ω and your normal x10 probe is set up to work with that. 

For most people, the only times you need to worry about 50 ohm is essentially when you're connecting a BNC signal cable to your scope to do specialized measurements where you will be using a T-connector with a 50Ω terminator on it (or 75Ω or whatever you need to match your source and/or line's impedance.)  Some scopes have a mode with the ability to turn on an internal 50Ω termination instead of using a T-connector and terminating resistance but that is NOT the normal operating mode and has nothing to do with measurements you're making with the standard type probes.

[tautech]

I do really appreciate all your answers and yes RTFM is important. And I have read a lot to avoid some of the worst:-)


But imagine this normal situation:

A noob wants to find out how to reduce the ripples in his working DIY PSU, it's a 17V 500mA.
To him does 10:1 attenuation sound a bit stupid because he does not need the higher frequency and do in fact think he needs the higher resolution.
He has learned from EEV, among others that he has to remember if he wants to get the right voltage value does he also have to think of the 50 Ohm termination, he knows his probe is 50 Ohm and you can maybe set the scope at 50 Ohm and 1:1.

Now does he put on the probes and his scoop is damaged?

The ripple is the AC component therefore AC input coupling is engaged to remove the DC component the AC is riding on top of.
If/when higher resolution is required rather than the normal 10:1 probe you should mostly use then use a 1:1 probe so to be able to use the sensitivity of your scope instead of having the signal attenuated by a factor of 10 as it is with a 10:1 probe.
If using a DSO remember to change the channel input setting to match the probe attenuation in use.
For this sort of basic probing needing to provide proper attenuation for a passive probe is not required so we use a 1M ohm input setting where the full input rating of the scope is available instead of the limitations of a 50 ohm input.

Simple basic stuff............

[nfmax]

LOL! :-)
Could one not build a "noob-filter" I mean voltage limiter or maybe a simple probe-between thingy? :-)

A very famous theorem says: no matter what protection you put on an oscilloscope input, there will always be someone able to fry it. NO. MATTER. WHAT.

Did I just invent this theorem? Yes. Does it work? Hell yeah.

I once worked with a future FRS* who fried the 50 ohm input of a 500MHz plug-in, when such things were rare and expensive beasts, attempting to measure the mains power input to a thermostatically controlled furnace.

* FRS = Fellow of the Royal Society

[BrokenYugo]

There is a noob filter for the most part, but as the saying goes they'll always build a better idiot.

I find it's best to start with something crude and simple to gain best understanding, say my crappy early 70s 5 MHz Heathkit scope.

http://www.schematicsforfree.com/archive/file/Test%20Equipment,%20Meters%20&%20Measurement/Oscilloscopes/Products/Heathkit%20Io-102%20Oscilloscope.gif

In this case the vertical amp is protected by D1 and D2 (transistors being used as a sort of zenerish diode because zeners were expensive 50 years ago, google Esaki effect for more info) which will conduct at at around 9-12 volts to protect the FET input, R1 limits the current so you don't smoke the protection transistors on every minor overload, C1 just improves high frequency response. You'll note this same structure is present on all user accessible inputs.

[David Hess]

A noob wants to find out how to reduce the ripples in his working DIY PSU, it's a 17V 500mA.
To him does 10:1 attenuation sound a bit stupid because he does not need the higher frequency and do in fact think he needs the higher resolution.
He has learned from EEV, among others that he has to remember if he wants to get the right voltage value does he also have to think of the 50 Ohm termination, he knows his probe is 50 Ohm and you can maybe set the scope at 50 Ohm and 1:1.

Now does he put on the probes and his scoop is damaged?

High impedance 1:1 probes work with a 1 megohm input and not a 50 ohm input.

But for the sake of argument let's say the noob uses a 1:1 low-z probe with a 50 ohm input.  For lower bandwidth oscilloscopes which a noob is likely to have access to, up to 300 to 500 MHz, the 50 ohm input is actually the 1 megohm high impedance input with an internal 50 ohm feed-through termination switched into place.  The oscilloscope will detect the overload and disconnect the 50 ohm termination leaving the 1 megohm input which can tolerate hundreds of volts with out damage.

For the measurement you are suggesting, there is actually no reason to use a 50 ohm input.  A 1:1 high impedance probe is completely adequate and even has some advantages.

[radiolistener]

A 1:1 high impedance probe is completely adequate and even has some advantages.

When you deal with MHz frequency, even 1:10 probe is not adequate to measure 100 V AC signal. Such 1:10 probe just release magic smoke and turns into garbage. If you try to use 1:1 probe, a magic smoke will be released from the oscilloscope instead of probe

Here is voltage-frequency curve for a typical high impedance input limit.
As you can see, at 10-100 MHz even 50 V can destroy the equipment easily. Despite the fact that such equipment is rated for 300 V. Because 300V rating is relevant for DC.

[David Hess]

The original poster's application involved power supply noise and ripple measurements where probe derating with frequency does not apply.

In the case of high frequency derating, either a better probe is used or a sampler, which is not to be confused with sampling.  A sampler divides the input while maintaining a high voltage rating.