Oscilloscope feature: logarithmic timebase sweep

[Tomorokoshi]

I've encountered a case where it would be convenient to have a logarithmic timebase sweep on an oscilloscope. For instance, at the trigger time 1 ms/div would be nice, increasing with the sweep to end at perhaps 1000 ms/div. This would get a lot of detail at the start where all the action is, while also showing the final values in the same view. In my case an external x-axis sweep generator wouldn't be practical.

[LazyJack]

Well, that is callef dual timebase run B after A mode. All good oldschool analog scopes had this.
Edit: oh, i see you want slower sweep at the end. That is less common, usually sweep B is faster that sweep A, although I don't see any reason why this is a must.

[thinkfat]

Isn't that the same as Zoom mode in DSOs?

[tggzzz]

Generate the log of the relevant control signal, and apply it to the x axis in XY mode.

[DaJMasta]

Do you have to see the signal right after the trigger?

Through SCPI you may be able to wait until your initial trigger condition, then sent a series of commands to switch timebase and record a single shot.

Otherwise, I think the conventional way with a modern DSO is to rely on the long memory depth and just brute force it.  Zoom out as far as you can with seeing the required detail level of the trigger, move that as early as possible in the memory capture, and hope you have enough memory to see the end of your signal.  Another option could be using a scope on a fast setting as your trigger device, then running a trigger output from it into a much slower timebase scope or other digitizer, so you use the trigger scope to find the condition and then just turn on the other recorder with the signal.  This may be possible in some architectures, but it's effectively like a dual timebase solution and I don't think it's that commonly supported in DSOs.

[nctnico]

I've encountered a case where it would be convenient to have a logarithmic timebase sweep on an oscilloscope. For instance, at the trigger time 1 ms/div would be nice, increasing with the sweep to end at perhaps 1000 ms/div. This would get a lot of detail at the start where all the action is, while also showing the final values in the same view. In my case an external x-axis sweep generator wouldn't be practical.

Some DSOs allow to use an external sample clock. I think several Lecroy models have this feature. For example: my Wavepro 7200A accepts a 30MHz to 2GHz external clock. You can use a sweep generator to generate the variable clock frequency (in theory; I have not used this feature let alone try to sweep the sample clock).

[tautech]

I've encountered a case where it would be convenient to have a logarithmic timebase sweep on an oscilloscope. For instance, at the trigger time 1 ms/div would be nice, increasing with the sweep to end at perhaps 1000 ms/div. This would get a lot of detail at the start where all the action is, while also showing the final values in the same view. In my case an external x-axis sweep generator wouldn't be practical.

What sort of scope ?
The modern DSO with deep memory and History/Sequence and Search modes should have no issues with returning the info you seek.

[Mechatrommer]

I've encountered a case where it would be convenient to have a logarithmic timebase sweep on an oscilloscope. For instance, at the trigger time 1 ms/div would be nice, increasing with the sweep to end at perhaps 1000 ms/div. This would get a lot of detail at the start where all the action is, while also showing the final values in the same view. In my case an external x-axis sweep generator wouldn't be practical.

Some DSOs allow to use an external sample clock. I think several Lecroy models have this feature. For example: my Wavepro 7200A accepts a 30MHz to 2GHz external clock. You can use a sweep generator to generate the variable clock frequency (in theory; I have not used this feature let alone try to sweep the sample clock).

and then the Ref/Ext Clock will go to PLL to generate 10GHz for internal usage. i dont think sweeping a PLL is wise.

[Andreas]

The idea is not bad, but very strange looking on a digital scope.

different views here with X-Axis

a) time
b) sqrt(time)
c) log(100000 * time)

with best regards

Andreas

[T3sl4co1l]

It makes sense for certain applications, in the same way that log freq makes sense for spectrum analyzers and sweep generators.  I think the use cases are fewer.  One I can think of is testing an unknown step response; this way you get the entirety of even a very long transient, on one screen, without having to jump around.

Tim

[nctnico]

I've encountered a case where it would be convenient to have a logarithmic timebase sweep on an oscilloscope. For instance, at the trigger time 1 ms/div would be nice, increasing with the sweep to end at perhaps 1000 ms/div. This would get a lot of detail at the start where all the action is, while also showing the final values in the same view. In my case an external x-axis sweep generator wouldn't be practical.

Some DSOs allow to use an external sample clock. I think several Lecroy models have this feature. For example: my Wavepro 7200A accepts a 30MHz to 2GHz external clock. You can use a sweep generator to generate the variable clock frequency (in theory; I have not used this feature let alone try to sweep the sample clock).

and then the Ref/Ext Clock will go to PLL to generate 10GHz for internal usage. i dont think sweeping a PLL is wise.

No. I'm not talking about the external reference clock. I'm talking about the external sample clock!

[Tomorokoshi]

It makes sense for certain applications, in the same way that log freq makes sense for spectrum analyzers and sweep generators.  I think the use cases are fewer.  One I can think of is testing an unknown step response; this way you get the entirety of even a very long transient, on one screen, without having to jump around.

Tim

That's pretty much the application. Response of a PID with fast initial change for a few milliseconds easing into the final drive value after a couple seconds. A qualitative view as opposed to a quantitative view.

[mag_therm]

Search " Bode Plotter"

Any older ( I mean really old)  oscilloscope that can draw a Lissajous curve should do it.

I have a 1960's vintage Eico TR410 10 MHz  that can do it although I never tried a Bode Plot.

The Ext sync ( 2 banana sockets ) will accept an analog drive in the X axis,
 when the time base selector is turned right around CW to the X : "EXT" position.

But you have to provide a saw tooth that periodically  pulls the trace back to the left, along with your log curve, and in sync with your Y data.

I recall that the old analog computers could output both the Y and X for a Bode Plot  to paper plotters although its been a long time since they went extinct.

[Andreas]

That's pretty much the application. Response of a PID with fast initial change for a few milliseconds easing into the final drive value after a couple seconds.

Although my scope is able to do this: nowadays I would rather use a scope with long enough memory and a good zoom function.

with best regards

Andreas

[Mechatrommer]

I've encountered a case where it would be convenient to have a logarithmic timebase sweep on an oscilloscope. For instance, at the trigger time 1 ms/div would be nice, increasing with the sweep to end at perhaps 1000 ms/div. This would get a lot of detail at the start where all the action is, while also showing the final values in the same view. In my case an external x-axis sweep generator wouldn't be practical.

Some DSOs allow to use an external sample clock. I think several Lecroy models have this feature. For example: my Wavepro 7200A accepts a 30MHz to 2GHz external clock. You can use a sweep generator to generate the variable clock frequency (in theory; I have not used this feature let alone try to sweep the sample clock).

and then the Ref/Ext Clock will go to PLL to generate 10GHz for internal usage. i dont think sweeping a PLL is wise.

No. I'm not talking about the external reference clock. I'm talking about the external sample clock!

maybe you are refering to AUX IN input at the front configured as trigger channel. any normal channel can be used as log trigger this way. WP7000 is like any other WP/WM in the era afaik, only accept single ext 100MHz clock at the back and goes to 10GHz synthesizer/pll+distributor... you may want to check/service that pll because my used/old SDA6000 and DDA3003 never locked when they arrived. they still can work without locking but i guess more jitter or out of spec sampling rate. https://www.ebay.com/itm/LeCroy-WAVEPRO-7200A-LeCroy-2GHz-4-Channel-Digital-Oscilloscope-USED/254747092233?hash=item3b501c2509:g:3GYAAOSwB8ZfhNQ9

[David Hess]

Well, that is callef dual timebase run B after A mode. All good oldschool analog scopes had this.
Edit: oh, i see you want slower sweep at the end. That is less common, usually sweep B is faster that sweep A, although I don't see any reason why this is a must.

On a dual timebase oscilloscope, "mixed sweep" displays sweep A up until the delay time, and then continues with sweep B so the same sweep has two time/div scales.  This was used for exactly the purpose Tomorokoshi is asking about and commonly with video signals.

Delayed sweep displays only the B sweep after a delay determined by A sweep.  Alternate delayed sweep alternately displays the A and delayed B sweeps.  Alternate delta delayed sweep alternately displays the A sweep, and two B sweeps with separate delays.

[Ground_Loop]

Save off the trace data to a file and replot  with the necessary time axis math.

[Circlotron]

If you don't need a realtime display you could save the data and import it into a spreadsheet and re-plot it on whatever graph you like.

Edit ^^ beat me by 28 seconds. LOL

[2N3055]

That's pretty much the application. Response of a PID with fast initial change for a few milliseconds easing into the final drive value after a couple seconds.

Although my scope is able to do this: nowadays I would rather use a scope with long enough memory and a good zoom function.

with best regards

Andreas

I agree. No need to go log time probably.. On my Picoscope 3000D I can grab 416 MSamples at 1 sec/div with 41MS/sec sample rate.... 24 ns sample interval.. That gives full 10 sec with 24 ns time res...