Analyse complex waveform with analog Oscilloscope

[DDTech]

Dave and all the others,

I could use a hand on analyzing a complex waveform on my Tektronix 2245A analog Oscilloscope.

I have two signals CLK and SDA with a portion of ~ 440µs of data followed by 175ms of "nothing" and I have no success in getting the data portion flicker free on the screen so that I can actually analyse the data.
All the demos and tutorials I found deal with more homogenous data or waveforms from a signal generator and there things always seem to be easy.

Using the "norm" setting, I can trigger the data part and fix it horiontally, but the trigger LED keeps on flashing and so does the display. My data portion is only about 1/400 of the total signal length and I assume that is too seldom for the trigger to produce a steady display.
"Trigger holdoff" did not help me in this case.

I ran all the diagnostics in the service menu and all of them passed - so the scope itself seems to be OK.

Can this be done with a scope like that (pure analog) or is digital storage necessary?

Thanks for any help

Frank

 

[Alex Eisenhut]

The trigger *is* producing a steady display, it can't trigger on what isn't there, your repetition rate is simply too low. 5 Hz ...

Wrong tool IMO, a cheap and cheerful logic analyzer will do that job very well.

Or if you have the time and resources, build a I2C monitor that gobbles the I2C data and tosses it out a UART to a PC.

Years ago I built one for another serial protocol and everyone used it after. Mostly because dedicated I2C/SPI sniffers are stupid expensive.

[TimFox]

The traditional way to handle problems like this with an analog CRO is to use the "delayed sweep" function, where the horizontal waits a preset time after trigger before starting a faster sweep.  After the delay time, you have a choice of starting the sweep immediately or waiting for a trigger on the second time base.  The manual should explain how to do this:  small Tek scopes require that you set the A time base for delay, pull out the knob, then set the B time base for delayed sweep.

[DDTech]

Alex,

The trigger *is* producing a steady display, it can't trigger on what isn't there, your repetition rate is simply too low. 5 Hz ...

Thanks for the quick reply. That's what I was assuming.
Yesterday I tried to get a better result by adding waves from my veeery ooooold HP square wave generator trying to use those as a trigger source and 5hz was the value I was playing with and with that low frequency, yes You get flicker that way or the other.

Wrong tool IMO, a cheap and cheerful logic analyzer will do that job very well.

That actually was my first attempt. I got a Hantek 6022 USB scope with integrated logic analyzer, thinking that it would be enough for my entry level needs, but i was wrong.
the Oscilloscope software is either crashing or showing so much noise that it's hard to tell the difference between wave and noise.
But the true piece of crap is the logic analyzer software. The guy who made it should be sentenced to use it himself for the rest of his lifetime.
There is nothing You can set except printer settings - printing is highly important when analyzing data signals... No, sorry, You can do more. You can choose between blue style, black style, silver style and aqua.
The need for renaming channels seems to be completely overrated as well as the idea of matching colors to the used cables or upper and lower threasholds.
But for the price, I could live with that if only I could properly analyze my signals. Unfortunately You can only select zoom all, zoom out and zoom in which perform zooming in steps. You can't select an area by rubberband and when You zoom in, the view jumps almost to the beginning of the measurement. However there are no scrollbars. You have to navigate by swiping.
To make it short, after literally swiping for 15 minutes in a view of 10µs devisions without ever getting to the first data portion, I gave up and returned the Scope.

Or if you have the time and resources, build a I2C monitor that gobbles the I2C data and tosses it out a UART to a PC.
Years ago I built one for another serial protocol and everyone used it after. Mostly because dedicated I2C/SPI sniffers are stupid expensive.

yes I guess it'l bee something like that. From the Tectronix I get enough information about the timings. So I should be able to write a small program with an arduino, that collects the data and writes it out every now and then.

Thanks

[DDTech]

Tim,

The traditional way to handle problems like this with an analog CRO is to use the "delayed sweep" function...

thanks, yes that's the "trigger holdoff" on the Tektronix, but it did not help in this case.

[TimFox]

"Trigger hold off" and "delayed sweep" are different functions, both of which are included in high-end analog scopes.
The trigger hold off function ignores any triggers of the time base (presumably the A time base for dual-base scopes) for a time controlled by the hold-off pot.
Delayed sweep is a more powerful function, where one can specify either the time where the fast B-channel sweep starts or a delay before a trigger of the B time base is accepted.
To avoid flicker on the display, one would have to trigger the sweep at rates similar to video or cine.  Usual values for flicker threshold are 15 Hz to 60 Hz, depending on several variables, for human visual systems.  The persistence time of the typical P31 phosphor is < 1 msec.

[tggzzz]

Tim,

The traditional way to handle problems like this with an analog CRO is to use the "delayed sweep" function...

thanks, yes that's the "trigger holdoff" on the Tektronix, but it did not help in this case.

As Tim Fox points out,  trigger holdoff and delayed sweep are very different.

"Delayed sweep" is effectively "zooming in" on a small part of the trace.

I'm not familiar with your scope, but look for timebase controls labelled something like "main" and "delayed" switches, "A" and "B" rotary knobs. When "main" but A != B you will see part of the trace is brighter; switching to "delayed" will zoom in on the brighter part.  Since it is analogue, zooming in will further decrease the brightness of the trace.

[DDTech]

"Trigger hold off" and "delayed sweep" are different functions, both of which are included in high-end analog scopes....

sorry, then I got you wrong.
I was playing with the delay on the B signal. But, if not by accident, I did not try to trigger on that. Good point. I'll give that a try.

Thanks

Frank
 

[DDTech]

As Tim Fox points out,  trigger holdoff and delayed sweep are very different.

"Delayed sweep" is effectively "zooming in" on a small part of the trace.

I'm not familiar with your scope, but look for timebase controls labelled something like "

yes, sorry got it. On the Tektronix this is called "Mode" and "B" or "Alt" are doing just that. in B-Mode the cursor knobs are responsible for delay and position within the signal.

Thx

Frank

[Alex Eisenhut]

Wrong tool IMO, a cheap and cheerful logic analyzer will do that job very well.

That actually was my first attempt. I got a Hantek 6022 USB scope with integrated logic analyzer, thinking that it would be enough for my entry level needs, but i was wrong.

So cheap, but not very cheerful then?

Or if you have the time and resources, build a I2C monitor that gobbles the I2C data and tosses it out a UART to a PC.
Years ago I built one for another serial protocol and everyone used it after. Mostly because dedicated I2C/SPI sniffers are stupid expensive.

yes I guess it'l bee something like that. From the Tectronix I get enough information about the timings. So I should be able to write a small program with an arduino, that collects the data and writes it out every now and then.

Thanks

I2C is used everywhere, you might find someone else already did that.

[AndyC_772]

This is one of the applications for which an analogue scope isn't well suited.

The signal is only active for 440us out of every 175ms, so for most of the time, the scope is waiting for a trigger, and the beam has nothing to show on screen. I'm guessing the trace is very dim, and this is why.

If you're not getting a stable trigger, it may be because the scope is in Auto rather than Normal trigger mode. You don't want it triggering in between packets, even though there's a long delay between them.

An analogue scope is also only good at showing repetitive waveforms. If the data is changing between repeats, you'll see that it's changing OK, but you've no chance of decoding any one data packet.

My first scope was an analogue scope not unlike yours. It worked very well for me until I wanted to use it to probe individual reads and writes to an EEPROM, at which point I traded it for a digital storage scope. The DSO is a much better tool for this type of work, and even inexpensive ones will do the serial decoding for you too (though it's usually an extra cost option).

[free_electron]

take digital photocamera, point at scope screen , make room dark , open shutter of digital camera , do 1 sweep , close shutter of digital camera , take memeory card out or connect to computer , send picture to printer.

or get a digital scope.

[DDTech]

So cheap, but not very cheerful then?

Not cheerful at all.

If You expect little and don't even get that, it's pure disappointment. Especially when you first spend a lot of time thinking: "it can't be" and assume you don't see the solution due to a lack of knowledge and experience.

I2C is used everywhere, you might find someone else already did that.

most likely, but that arduino is the one that later should become the one that has to deal with the data. So, most of the code has to be written anyway.

[DDTech]

take digital photocamera, point at scope screen , make room dark , open shutter of digital camera , do 1 sweep , close shutter of digital camera , take memeory card out or connect to computer , send picture to printer. or get a digital scope.

Well, the actual question behind it was "is the tool not (well) suited for the job or am I not properly using the tool"?

But to get the job done, that truely would be another option. I already took several pictures with that in mind, but "open shutter / 1 sweep" is a good advice.  thx

[DDTech]

This is one of the applications for which an analogue scope isn't well suited.
The signal is only active for 440us out of every 175ms, so for most of the time, the scope is waiting for a trigger, and the beam has nothing to show on screen. I'm guessing the trace is very dim, and this is why.
...

That was my guessing. Problems like this and dealing with them helps to truely understand the tool and what is going on.

If you're not getting a stable trigger, it may be because the scope is in Auto rather than Normal trigger mode. You don't want it triggering in between packets, even though there's a long delay between them.

no, I'm only working in normal mode. auto in this case does not bring up anything useful at all.

An analogue scope is also only good at showing repetitive waveforms. If the data is changing between repeats, you'll see that it's changing OK, but you've no chance of decoding any one data packet.
My first scope was an analogue scope not unlike yours. It worked very well for me until I wanted to use it to probe individual reads and writes to an EEPROM, at which point I traded it for a digital storage scope. The DSO is a much better tool for this type of work, and even inexpensive ones will do the serial decoding for you too (though it's usually an extra cost option).

I already have one picked out, I was just hoping I could get away (a little longer) without spending the money.
It probably would be more easy if I had an analog scope with digital storage, but as I don't I'll go for the DSO.

Although I'm not able to easily analyze each bit of the data packet, the scope still helped a lot on solving the problem. I found that one of my level shifting transistors was somewhat dodgy and it directed me to a missing resistor. Now the signal "looks good" and I should be able to properly read and analyze it in code.

[DDTech]

Tim,

Delayed sweep is a more powerful function, where one can specify either the time where the fast B-channel sweep starts or a delay before a trigger of the B time base is accepted....

I still might have to look into the docs, but it's a lot better now. The trigger signal is steady now. The display still flickers, but as the data is comming around only every 200ms (~5Hz) I'm almost sure this can't be fully avoided.

Thanks for all the responses.

Frank

[mrkev]

Analog scope is really not good for this sort of stuff.
Where does the data come from? The only way I can think of is to make the device send the same data more often. That is if you even can change the way how that data is send. If it's a micro, you can even generate the square signal to trigger the scope.


Btw. In the old times, they would use camera with a long exp. time (according to the time base) with the auto release  hooked to the trigger. That way, you can sort of capture one frame.
My friend did that as a project like 10 years ago (back in high school) and he was able to come up with some good (not great ) results. He used old digital camera and synchronizing circuit wired directly to the release switch.

[tggzzz]

Although I'm not able to easily analyze each bit of the data packet, the scope still helped a lot on solving the problem. I found that one of my level shifting transistors was somewhat dodgy and it directed me to a missing resistor. Now the signal "looks good" and I should be able to properly read and analyze it in code.

And that is appropriate use of the tools!

Use a digital or analogue scope to ensure signal integrity. Then, once you have seen that the analogue signal can be interpreted as the correct digital signal, switch to debugging in the digital domain. That might be using code, it might be using a logic analyser?

[DDTech]

Where does the data come from? The only way I can think of is to make the device send the same data more often. That is if you even can change the way how that data is send. If it's a micro, you can even generate the square signal to trigger the scope.

It's a digital micrometer, so I don't have influence on what is coming when.
I actually do have several resources and even pieces of ready code. So I could probably run that and be fine. But for me in this case it's more than taking someone elses code in order to get the job done.
I prefer to know/understand what I'm doing over tiptoeing in a minefield.
Therefore I'm trying to solve this as if I did not have that much information, so that next time in a similar situation (but without access to resources), I'll not be completely lost.
I think they call it "learning".

Btw. In the old times, they would use camera with a long exp. time ....

I was wondering how they actually did it professionally in those days. Digital storage scopes were available even then, but probably muuuch more expensive.

[DDTech]

And that is appropriate use of the tools!...

seems I'm on the right track... thanks

[TimFox]

As an older person, I prefer using an analog scope when I don't know yet what the problem is.  To measure the result (time, voltage, etc.) or store a single trace, I switch to a digital scope.

[mrkev]

Where does the data come from? The only way I can think of is to make the device send the same data more often. That is if you even can change the way how that data is send. If it's a micro, you can even generate the square signal to trigger the scope.

It's a digital micrometer, so I don't have influence on what is coming when.
I actually do have several resources and even pieces of ready code. So I could probably run that and be fine. But for me in this case it's more than taking someone elses code in order to get the job done.
I prefer to know/understand what I'm doing over tiptoeing in a minefield.
Therefore I'm trying to solve this as if I did not have that much information, so that next time in a similar situation (but without access to resources), I'll not be completely lost.
I think they call it "learning".

Yes. That is the best way how to approach learning  ...

If it is a micrometer, it probably sends the same data (measured value) anyway. So you could use delayed sweep in this case
(if your scope even has that option).

Btw. In the old times, they would use camera with a long exp. time ....

I was wondering how they actually did it professionally in those days. Digital storage scopes were available even then, but probably muuuch more expensive.

Digital storage scopes were not available before something like 40 years at all (and in some countries not even in late 80's). They used analog storage scopes. But that was completely different beast. It had tube that could "store" the trace after single release. Of course, it didn't have any memory that could be copied or stored for longer time (the screen would just glow much longer after being hit due to the special construction).
But in the production, when you wanted to store and show the data from your scope, you had to use photography (and darkroom + all the equipment).

[tggzzz]

Digital storage scopes were not available before something like 40 years at all (and in some countries not even in late 80's). They used analog storage scopes. But that was completely different beast. It had tube that could "store" the trace after single release. Of course, it didn't have any memory that could be copied or stored for longer time (the screen would just glow much longer after being hit due to the special construction).
But in the production, when you wanted to store and show the data from your scope, you had to use photography (and darkroom + all the equipment).

We use Polaroid instant camera. Much easier.

And don't forget that Tektronix had large screen graphical terminals that used storage scope technology; the image lasted hours unless erased. Great for getting a computer to draw a smith chart; not so good for multi-screen program listings!

[mrkev]

We use Polaroid instant camera. Much easier.

And don't forget that Tektronix had large screen graphical terminals that used storage scope technology; the image lasted hours unless erased. Great for getting a computer to draw a smith chart; not so good for multi-screen program listings!

Sure, but that wasn't something that was accessible in every country... F.e. I have never seen Polaroid camera (sure, I am not that old but still), because it wasn't common technology in the former soviet part of Europe, neither was the Tektronix.
But they actually produced this camera rig, that was build to fit the exact scope model screen and could be easily synchronized with the trigger...