Minimum Oscilloscope Bandwidth?

[mc1822]

james-s,
I have accumulated some of this equipment over time in case I decide to go further in my electronics hobby. I do have some electronic knowledge.I know what capacitors, resistors, diodes, transistor, etc. are and what they do. I also am very experienced in soldering. I have read and am reading various books on electronics in general and how to diagnose and repair audio and radio equipment. I am also viewing videos on Youtube related to these subjects. My hope is that by joining various electronic forums I will be able to learn from more experienced posters! Besides, don't we all like "toys"? 

[Fungus]

james-s,
I have accumulated some of this equipment over time in case I decide to go further in my electronics hobby. I do have some electronic knowledge.I know what capacitors, resistors, diodes, transistor, etc. are and what they do. I also am very experienced in soldering. I have read and am reading various books on electronics in general and how to diagnose and repair audio and radio equipment. I am also viewing videos on Youtube related to these subjects. My hope is that by joining various electronic forums I will be able to learn from more experienced posters! 

Yep, definitely a candidate for an Analog Discovery...

[thinkfat]

This means you get 3 dots/points on screen  per sine wave cycle which is crap to say the least.

Nope. Look up sin x/x reconstruction and sampling theory in general. Usually digital scopes work well up to fs=2.5 * bandwidth. So 100Ms/s is good for 100/2.5=40Mhz. The theoretical limit is Nyquist but for practical purposes you'll need some headroom for the anti-aliasing filter.

If you know your input is a sine wave (or generally, within the bandwidth limit), it's OK, yes.

Er, no. Arbitrary waveforms are fine, provided that the signal's highest component frequency is within the sampling limit.

It is, of course, necessary to be clear about the definition of "signal". For example a Tek 1502 has ~25kS/s and a 4GHz front end; the hp54100a has 40MS/s and a 1GHz front end.

That's what I meant when I said "within the bandwidth limit". I should have said "sampling limit".

The TEK1502 is a TDR, fidelity in waveform reconstruction is not key here, I'd say. The HP54100a is a sampling oscilloscope that explicitly exploits the aliasing caused by subsampling the input signal. Works for repetitive waveforms only.

[tggzzz]

This means you get 3 dots/points on screen  per sine wave cycle which is crap to say the least.

Nope. Look up sin x/x reconstruction and sampling theory in general. Usually digital scopes work well up to fs=2.5 * bandwidth. So 100Ms/s is good for 100/2.5=40Mhz. The theoretical limit is Nyquist but for practical purposes you'll need some headroom for the anti-aliasing filter.

If you know your input is a sine wave (or generally, within the bandwidth limit), it's OK, yes.

Er, no. Arbitrary waveforms are fine, provided that the signal's highest component frequency is within the sampling limit.

It is, of course, necessary to be clear about the definition of "signal". For example a Tek 1502 has ~25kS/s and a 4GHz front end; the hp54100a has 40MS/s and a 1GHz front end.

That's what I meant when I said "within the bandwidth limit". I should have said "sampling limit".

The TEK1502 is a TDR, fidelity in waveform reconstruction is not key here, I'd say.

Waveform fidelity is key when determining the location and value of impedance, e.g. where in a connector the impedance is 49ohms rather than 50ohms, or to what extent a cable has a regular impedance variation.

The HP54100a is a sampling oscilloscope that explicitly exploits the aliasing caused by subsampling the input signal. Works for repetitive waveforms only.

Of course. That's the point! A point that many people think violates the sampling theorem.

[vk6zgo]

Thanks to everyone that replied! I already have a digital multimeter, function generator, frequency counter and variable dc power supplies. I am curious if a audio generator or signal tracer would be of use?

A lot of people have dived into esoterica, way beyond what you would use.

bd139 has a more practical view, but has described an Oscilloscope as a "jack of all trades" instrument.
With respect, that is its strength.

A cheap, second or third, or nth hand analog 'scope can do the job of :-

(a)A Signal tracer.
(b)A large scale ac voltmeter.
(c)A large scale centre zero DC voltmeter.

In (a), if the radio is showing signs of life, but no audio output.you can look for the presence of the local oscillator, & if that is present, see if there is any IF output from the mixer, then check through the IF stages to the detector.
If there is detected audio, you can then look at the audio output amp.

In (b), if you are aligning the radio the "simple" way by just looking at the increase in audio output, you can observe the audio waveform.
If the level increases, you are "peaking" the IF or whatever, just like an analog ac voltmeter.

In (c), if you need to look at DC operating voltages, it is often easiest to "free-run" your scope ("auto trigger"), then watch the deflection of the trace in the positive or negative direction.
Of course, you have to use your brain, as it doesn't "spoon feed" you like a DMM.
If, instead of just looking at the audio output when aligning a Rx, you are monitoring the AGC line, you can again use the 'scope in this mode.

"Short wave " radios suggests a maximum frequency requirement of around 30MHz for conventional superhets, but for Wadley Loop & synthesised radios, you may be better with a 100MHz instrument.

If you are lucky enough to pick up one of those two or more channel analog 'scopes which provide an output of one vertical channel, you can use the 'scope as a "buffer" or a preamplifier, allowing the use of a frequency counter, in places where the counter's low input impedance or lack of sensitivity may be problematical, if used alone.

Another trick that can be used with most such 'scopes, is to use two vertical channels "in cascade" to obtain extra sensitivity.
This is not usually a very accurate method, but sometimes, either the presence , or shape of a signal is more important than its absolute amplitude.

DSOs can do most of these things, obviously excepting the last two, but usually have a reasonably good frequency counter built in.

[David Hess]

Regarding fixing radios, I don't think I've ever used my oscilloscope for that. Most radio faults are things like bad capacitors and cracked solder joints, dirty pots and drifted resistors. You should be able to accomplish most repairs with a multimeter and your eyeballs. Audio and RF test oscillators may occasionally be useful.

That is my experience also when working on radio equipment.  However I *have* used my oscilloscope as an impromptu RF voltmeter to verify signal levels.  But mostly it gets used to trace audio circuits and verify proper operation of power supplies.

In my most recent case, I used my oscilloscope to measure the transient response of the repaired regulated power supply which was for the RF power amplifier.  I wanted to make sure that it was never applying excessive voltage to the output transistors.  This turned up some flaws in the original design which Kenwood solved by adding a zener diode clamp to the output which is what had shorted out blowing up the regulator's pass transistors.

I also used my DSO to measure the startup time of the various local oscillators which gives some indication of their design margin.

[thinkfat]

The HP54100a is a sampling oscilloscope that explicitly exploits the aliasing caused by subsampling the input signal. Works for repetitive waveforms only.

Of course. That's the point! A point that many people think violates the sampling theorem.

Well, it does "violate" the sampling theorem. If that was possible. But it isn't. The sampling theorem is not a law, it describes an effect. The sampling theorem just tells you where in the frequency domain the signal you're sampling is going to show up

[Fungus]

The HP54100a is a sampling oscilloscope that explicitly exploits the aliasing caused by subsampling the input signal. Works for repetitive waveforms only.

Of course. That's the point! A point that many people think violates the sampling theorem.

"Folding" is a well documented part of sampling theory:

https://en.wikipedia.org/wiki/Aliasing#Folding

It's why the Nyquist "limit" is also known as the "folding frequency":

https://en.wikipedia.org/wiki/Nyquist_frequency

[nigelwright7557]

This means you get 3 dots/points on screen  per sine wave cycle which is crap to say the least.

Nope. Look up sin x/x reconstruction and sampling theory in general. Usually digital scopes work well up to fs=2.5 * bandwidth. So 100Ms/s is good for 100/2.5=40Mhz. The theoretical limit is Nyquist but for practical purposes you'll need some headroom for the anti-aliasing filter.

Anyway, for repairing receiver I'd also look for a signal source (RF generator) and a DMM.

Explain to me how 3 points on the screen represents a 40MHz sine wave ?
I would suggest you need at least 30-40 points to give a half decent sine wave.

[Fungus]

Explain to me how 3 points on the screen represents a 40MHz sine wave ?
I would suggest you need at least 30-40 points to give a half decent sine wave.

Nope. Sin(x)/x reconstruction only needs "more then two" points (less than three still works, eg. five points every two cycles is still plenty).

The explanation: There's only one possible sine curve that fits through those points, it can be extrapolated from them.

[bd139]

Taylor series.

Nothing wrong with repetitive signal sampling either. I rather like a 54600 scope. 20MS/s single shot. Aliasing? Just don’t be a dumbass and it’s fine.

[thinkfat]

This means you get 3 dots/points on screen  per sine wave cycle which is crap to say the least.

Nope. Look up sin x/x reconstruction and sampling theory in general. Usually digital scopes work well up to fs=2.5 * bandwidth. So 100Ms/s is good for 100/2.5=40Mhz. The theoretical limit is Nyquist but for practical purposes you'll need some headroom for the anti-aliasing filter.

Anyway, for repairing receiver I'd also look for a signal source (RF generator) and a DMM.

Explain to me how 3 points on the screen represents a 40MHz sine wave ?
I would suggest you need at least 30-40 points to give a half decent sine wave.

If your reconstruction filter was just a "connect the dots" interpolation, then, yes, you'd need a lot more points than 3.
Of course that is not what is done. I've found the attached paper (an EETimes article) by some Agilent guy who explains how the interpolation works. Basically, you run the samples through a rectangular shaped filter with a stop frequency of F_s/2, or practically, through a FIR filter with enough coefficients to be a close approximation of this rectangular shape. This can be shown to exactly reconstruct the original waveform provided it  is properly bandlimited, i.e. has no frequency components above F_s/2.

[tggzzz]

The HP54100a is a sampling oscilloscope that explicitly exploits the aliasing caused by subsampling the input signal. Works for repetitive waveforms only.

Of course. That's the point! A point that many people think violates the sampling theorem.

Well, it does "violate" the sampling theorem.

No it doesn't.

If that was possible. But it isn't. The sampling theorem is not a law, it describes an effect.

The sampling theorem  is a fundamental mathematical relationship that cannot be broken.

The sampling theorem just tells you where in the frequency domain the signal you're sampling is going to show up

For a stationary unchanging signal i.e. with a  bandwidth of ~0Hz, you can take as long as you want to take samples and then reconstruct it.

To do that you need a trigger and sampler that operates sufficiently fast to resolve edges, but the samples can be taken arbitrarily infrequently. 

For another example, understand how Tayloe mixers work, as seen in SDR dongles and elsewhere.

Key search term: subsampling.

[Wallace Gasiewicz]

I don't know what type of radio you want to fix or experiment with, Shortwave? Ham? FM?
For repair and other fun, I use an HP 8935 Service monitor. It combines spectrum analyzer with tracking or separate RF , AF
generators and power meter. It is really a cell tower tester.
It also has a scope , but is it really an audio scope and not really a good one. There is also a lot of cell phone stuff that is of no use to me.
You can generate a signal ,modulate it with AM or FM and trace it. You can look for signal purity on the spec analyzer. This thing is more expensive, like about 1K USD, (I just sold a really good unit with a beat up case for $600) but it does lots of stuff that are quite helpful in fixing all sorts of radios.
It is also a pretty good AM, SSB and FM receiver up to 1 GHz.

If you are testing ham radios you can generate an audio signal to put into the mic input and see what you get in the ant output.
http://www.amtronix.com/usequip1.htm has a chart that lists some service monitors and what they do. Their prices are high in my opinion, but they stand behind their units.

If you want a scope, may I suggest an old analog one. You can pick up an old analog scope at hamfests for cheap. The new digital scopes are getting much better and have caused the price on old analog scopes to come down quite a bit. You can get a nice analog one and learn what you want and need in a scope. In the meantime, the digital scopes will get better and better and more used ones will show up cheap.
About bandwith... if you are working on old radios remember that some shortwave radios have IF above 50 MHz, even if they only receive up to 30 MHz. FM radios are higher.
General rule with bandwith on old O scopes is that the MHz stated is where the signal that is put on the screen is 80% of what is should be in amplitude.
Another thought: audiophools seem to love really old even vacuum tube scopes for their low frequency capabilities.
As far as my O scopes go... I use an ancient 100 MHz HP scope. A HP 1980, sold for about  $18K new, I bought it for $50. It is usable up to about 150 MHz. It is partially "digital"
The thing that most folks do not take in consideration when using a scope is that the probes are at least as important as the scope at RF frequencies. You will need a 1X and 10X probe, probably two of each.
I really have not used new digital scopes and am not a good source of info on them except they seem to be getting better and better for less and less money. I have used old digital scopes in the past but the technology is quite different.

Wally KC9INK
 

[vinlove]

A couple year ago I had a few different makes of old analogue o scopes, and all of them had some problems such as one channel not working, display problems etc. Most of them had some problems with dodgy contact on the rotary switches too.

Like everything else, and all other electronic devices, with old age their limited lives were keep ending.

I sold them all for parts, and got a brand new Rigol DSO and a couple of handheld o scopes.

If you are lucky, you will get good perfect working old scopes, but also there are possibilties that you might be buying a bag of problems and restoration projects with little prospect of finding the replacement parts. Chances are more likely latter.

[james_s]

Vacuum tube scopes are great for working with radio transmitters, tesla coils, other vacuum tube electronics and anything else where high voltages can be present. A tube scope has a much better chance of surviving abuse that would fry expensive semiconductors in a solid state scope. Other than that there is not much reason to want one beyond collecting an antique.

[bd139]

Unless the front end is a nuvistor   

[David Hess]

Nuvistor's are also tougher than semiconductors.

[vk6zgo]

Vacuum tube scopes are great for working with radio transmitters, tesla coils, other vacuum tube electronics and anything else where high voltages can be present. A tube scope has a much better chance of surviving abuse that would fry expensive semiconductors in a solid state scope. Other than that there is not much reason to want one beyond collecting an antique.

Semiconductor based 'scopes have been used for 40+ years on vacuum tube & other high voltage equipment.(even a solid state Analog TV Horizontal output stage has some serious voltages)
I've never seen one "fried", in a career of a similar duration.

The number of times you have to connect a 'scope directly to the really high voltage parts of an energised transmitter are really vanishingly small.

[0culus]

Someone blew up one channel of our very expensive Keysight scope at work last year. Never found out who did it; Keysight repaired it under warranty.


If you use common sense and the right tools you won't blow it up.

[vk6zgo]

A couple year ago I had a few different makes of old analogue o scopes, and all of them had some problems such as one channel not working, display problems etc. Most of them had some problems with dodgy contact on the rotary switches too.

So one channel doesn't work?
You've still got another one that does.

I've used 'scopes with one faulty channel at work, when, as often, they were the only one left.
If I'd complained to the boss that it only had one channel, he would have replied: "Well how many channels are you using?"

At home, I successfully fault found an ignition fault on a 1988 Ford, using an old BWD 'scope with faulty triggering.
I just let it "free run" & watched the pulses from the EMM pop up & drift by.

Faulty switches ---wiggle 'em!

[nigelwright7557]

Nope. Sin(x)/x reconstruction only needs "more then two" points (less than three still works, eg. five points every two cycles is still plenty).

The explanation: There's only one possible sine curve that fits through those points, it can be extrapolated from them.

What happens if its not a sine wave ?

[tautech]

It gets it wrong and really noticable in Single shot mode.

Ideally the more data points to reconstruct a waveform the better and scopes with a good amount can be used entirely in Dot mode where any interpolation results don't matter as damn as it's not used.

[David Hess]

Nope. Sin(x)/x reconstruction only needs "more then two" points (less than three still works, eg. five points every two cycles is still plenty).

The explanation: There's only one possible sine curve that fits through those points, it can be extrapolated from them.

What happens if its not a sine wave ?

If no aliasing is present, then there is only one solution and the sin(x)/x interpolation reproduces the input.  If aliasing is present, then there are multiple solutions and interpolation does not produce an accurate result.  Of course the anti-aliasing filter if present distorts the input anyway so it may come down to a case of choosing the lessor evil unless the sample rate can be increased.

[Fungus]

If you use common sense and the right tools you won't blow it up.

And a fixed 10x (or 100x) probe - none of that 1x rubbish.