digital versus analogue oscilloscope

[Thunderchild]

the thing is there is not merely a bit of "static" it has a well defined frequency of around 110 KHz, it also superinposes onto all signals, the instructions mention having to filter out FM or AM - from where ? ok let me see a SMPS signal, what does it do ? hm yea you could peobably with a twisted mind come to think that a SMPS signal is FM or AM couldn't you ? what a great way of telling you in the manual how to cope with the oscilloscopes own defect whilst calling it something else

 

[bountyhunter]

I have a TEK TDS 210 60MHz digital scope that I've had for a few years. It has a 1GS/s sample rate so it displays fast transients within its bandwidth without dots.

But what is the usable bandwidth? On high frequency switching converters you can get voltage spikes on the turn-off waveform that may be 10 or 20 nano seconds wide. I remember distinctly doing a "side by side" for my boss with my analog scope and the fancy 2GIG digital to show him the dots and reading error because it misses the full peak amplitude of the spike. He wasn't real happy, but I got tired of being ragged on for having an analog scope. Digital scopes are OK for most things, not all things.

 

[bountyhunter]

I thought the 7904A was light. Compare with a 7704 (not 7704A). It had a non-switching supply and was easily twice the weight of a 7904. the 7000-series overall lightened up the Tektronix product line over the older 500-series. If you wanted a workout for the gonads, you tried lugging a 556 around the shop. It was portable because it had handles.

Dean

Maybe the 7904A was light for you, I have a blown out lower back. Seemed pretty heavy to me. I do remember the A version had a switching power supply that clicked a few times when it turned on. The old boat anchors were heavier.

BTW, the 7904A was company property so it stayed behind. The scope I use now is a 7834 storage scope I got used off ebay. Seems to work OK, it's also a very good scope.

 

[crutschow]

What bandwidth? On high frequency switching converters you can get voltage spikes on the turn-off waveform that may be 10 or 20 nano seconds wide. I remember distinctly doing a "side by side" for my boss with my analog scope and the fancy 2GIG digital to show him the dots and reading error because it misses the full peak amplitude of the spike. He wasn't real happy. Digital scopes are OK for most things, not all things.

I stated it was a 60MHz oscilloscope, so I was referring to that bandwidth.

A 10ns spike has an equivalent bandwidth of about 50MHz (assuming an approximate half sine-wave for the spike) then a 60MHz bandwidth should be adequate to reasonably accurately detect the spike with a digital scope that has a 1GS/s sample rate (10 samples during the spike interval).

Dean, the TDS 210 does indeed have a 1GS/s sample rate (see **broken link removed**).

 

[Thunderchild]

I think with digital bandwidth may be an approimation, it depends on what your looking at, a random wavefor needs as many samples as possible, a square essentialy to detect its frequency at least (not duty cycle) 2 samples per wave will do it increasing the "apparant" bandwidth, it all comes down to knowing your instrument

 

[bountyhunter]

I stated it was a 60MHz oscilloscope, so I was referring to that bandwidth.

A 10ns spike has an equivalent bandwidth of about 50MHz (assuming an approximate half sine-wave for the spike)

That's a wrong assumption. A spike has a very steep vertical edge then falling edge with some ringing after. The total thing may take 10 - 20ns, the edges are a lot faster. I have seen turn off ringing transients where the fundamental frequency of the ring was 100+MHz, but the edge spike was very steep compared to that.

 

[bountyhunter]

I stated it was a 60MHz oscilloscope, so I was referring to that bandwidth.

A 10ns spike has an equivalent bandwidth of about 50MHz (assuming an approximate half sine-wave for the spike) then a 60MHz bandwidth should be adequate to reasonably accurately detect the spike with a digital scope that has a 1GS/s sample rate (10 samples during the spike interval).

Dean, the TDS 210 does indeed have a 1GS/s sample rate (see **broken link removed**).

I may be slow, but if something samples at 1GIG (10 EXP 9) times per second, that means in 10 ns it has taken 10 samples. Ten data points for displaying the whole transient waveform which is an upspike, falling spike, then a ringing portion. hence the dots and missed data, hence the sour look on my boss' face when he saw the scopes side by side..

 

[bountyhunter]

I think with digital bandwidth may be an approimation, it depends on what your looking at, a random wavefor needs as many samples as possible, a square essentialy to detect its frequency at least (not duty cycle) 2 samples per wave will do it increasing the "apparant" bandwidth, it all comes down to knowing your instrument

Right. A square wave for example may have a fundamental frequency of 1MHz, but if it has fast rising edges, a 50 MHz bandwidth scope can distort those edges very badly (drag them out and round the corners). The fourier content of any waveform with a steep edged waveform goes out nearly to infinty, depending on how steep the edge is. Spikes are like the edges of square waves.

 

[crutschow]

That's a wrong assumption. A spike has a very steep vertical edge and following edge with some ringing after. The total thing may take 10ns, the edges are a lot faster. I have seen turn off ringing transients where the fundamental frequency of the ring was >100MHz, but the edge spike was very steep compared to that.

My assumption was correct based upon your statement of it being a 10 or 20ns spike. If you want to change your straw-man description of the spike differently with more detail, than obviously the assumption needs to change. If you want to detect a 10ns spike with 100MHz ringing, than a digital scope of at least 100MHz bandwidth and higher sample rate is required.

Some of the newer digital scopes have special modes to detect short, random spikes, even at slow sweep speeds, which an analog scope does not do well.

The is also a question of, how much detail do you really need to look at a spike? Generally the presence of a spike and it's approximate amplitude is enough to tell you, whether you have a problem. The precise shape of the spike is generally of less interest (but certainly can be useful in some situations).

 

[bountyhunter]

My assumption was correct based upon your statement of it being a 10 or 20ns spike. If you want to change your straw-man description of the spike differently

Whatever. You're the one who looked at the word "spike" and immediately turned it into a half sine wave to prop up the meager sampling abilities of a digital scope. Even if you try to claim I meant the entire duration of the spike would be 10ns, a spike has vertical edges, not sinusoidal, so you can't use sinewave bandwidth to analyze the waveform.

The is also a question of, how much detail do you really need to look at a spike?

The correct peak positive and negative amplitude would be a minimum, which you don't get if the sample doesn't happen to fall right on it.

Like I said, and what I said was correct: digital scopes suck swamp water trying to represent waveforms that happen in a very short time because they are sample limited. Period, case closed.

 

[crutschow]

Whatever. You're the one who looked at the word "spike" and immediately turned it into a half sine wave to prop up the meager sampling abilities of a digital scope. Even if you try to claim I meant the entire duration of the spike would be 10ns, a spike has vertical edges, not sinusoidal, so you can't use sinewave bandwidth to analyze the waveform.




The correct peak positive and negative amplitude would be a minimum, which you don't get if the sample doesn't happen to fall right on it.

Like I said, and what I said was correct: digital scopes suck swamp water trying to represent waveforms that happen in a very short time because they are sample limited. Period, case closed.

Whatever.

That's not true for scopes with sufficient sample rate. But since you say the case is closed, and your dismissive attitude suggests so is your mind, I'm signing off this fruitless discussion. You are more than welcome to stay with your archaic analog scopes. I used them for 30 odd years and more than happy if I never see one again (along with vinyl records, film cameras, and analog TV).