Old analogue scopes beat new digital scopes, for bit-bashed digital outputs

[Fungus]

It is a shame people ignore the multiple explicit caveats in the first post.

And an even bigger shame when they can't see how finely tuned your code was to produce those exact screenshots.

[tggzzz]

It is a shame people ignore the multiple explicit caveats in the first post.

And an even bigger shame when they can't see how finely tuned your code was to produce those exact screenshots.

As has already been pointed out to you a pulse train with a low duty cycle scarcely constitutes "finely tuning", and is a common waveform in hard realtime systems. For example, on one of my other projects with this device the input period varies from 60ns to 6s.

[Fungus]

As has already been pointed out to you a pulse train with a low duty cycle scarcely constitutes "finely tuning", and is a common waveform in hard realtime systems.

Get back under your bridge.

No "beginner" or non-troll would ever type this:

Code: [Select]

while (1) {
        outPort <: 1;
        outPort <: 1;
        outPort <: 1;
        outPort <: 1;
        outPort <: 0;
...


[MK14]

I think both the OP's oscilloscopes, are beaten by this:



It looks much cooler, can tell the time as well, and a situation can be carefully engineered, to make it look better than the original two scopes.

[Gyro]

That brings a whole new dimension to ground lifting! 

[David Hess]

This is a simple 10kHz waveform output.

The old analogue scope display is clear and simple - and correct.

But the (single shot) digitising scope display is woefully incorrect. When not in single shot mode, the trace twinkles like a demented Christmas decoration!

The DSO needs peak detection (1) to display this waveform correctly which the Rigol DS1000E series does not despite what its lying manual says.  When I first evaluated Rigol's DSOs (before the DS1000Z series became available), I asked their customer service about it and just got more lies.  "Of course they support peak detection, it is right there in the manual!"  Having Rigol double down was very frustrating.

My 27 year old 2232 DSO would have no trouble displaying this and it has done so.  My slightly older 2230 might have problems because its peak detection only goes down to 100ns.

I have no idea why the burst pulse is shown incorrectly on the DS1000E; that is just broken.  The bandwidth is not high enough to reproduce the edges but the sample rate and bandwidth should be plenty for the peaks and where is that trailing edge trash coming from?  That is not caused by low bandwidth or insufficient sample rate.  I call shenanigans; that DSO is broken by design unless there is a problem with the probe.

(EDIT added for clarification. That is a very common waveform in hard realtime systems, when you are blipping a bit to see how fast/slow a loop is occurring, or  the latency between two events, typically a stimulus and a response)

I have often made this measurement in real time systems.  With a good oscilloscope you can measure the distribution of response time or latency and capture long latency glitches which completely destroy real time performance.  If the distribution is gaussian, then you can even measure it quantitatively with a delta delayed sweep on an analog oscilloscope.

Clue: Toggle the pin instead of blipping it then you can use the DSO's pulse width measurement to get the elapsed time instead of trying to count fractions of squares on a a green screen.

That is a good work around for an oscilloscope which has trouble making this measurement but instead of toggling, I set and reset so the pulse polarity is always consistent.

(You do have pulse width measurement on the Analog 'scope, right?)

Mine does.

BTW that Rigol isn't new but very old. Turning on peak detect will help a lot to see more.

The Rigol DS1000E series does not support peak detection; it supports envelope detection over many acquisitions.  Envelope detection should work for this and would even be preferable in some cases.  From the user manual:

Peak Detect Acquisition: Peak Detect mode captures the maximum and minimum values of a signal. Finds highest and lowest record points over many acquisitions.

BTW that Rigol isn't new but very old. Turning on peak detect will help a lot to see more.

Of course. As I said, I know which 5 buttons to press to change from the "normal" (that's what is it called) setting to "peak detect". But beginners might not even notice the choice exists.

Then they'll learn an important lesson: RTFM!

And then they may learn a more important lesson: Rigol's manuals lie!

(1) Or a very long record length which will compromise performance.  On the other hand, long record lengths are inexpensive and getting more so all the time.

[2N3055]

The Rigol DS1000E series does not support peak detection; it supports envelope detection over many acquisitions.  Envelope detection should work for this and would even be preferable in some cases.  From the user manual:

Peak Detect Acquisition: Peak Detect mode captures the maximum and minimum values of a signal. Finds highest and lowest record points over many acquisitions.

And then they may learn a more important lesson: Rigol's manuals lie!

David,
I don't want you to think I defend Rigol. I don't.

But.... Technically, you are wrong.. If THEIR User manual defines "Peak Detect mode" the way they do, they dont lie.

They decided to define term "Peak Detect mode" differently than Tektronix or other US manufacturers.. It is sleazy (if deliberate), stupid, it is misleading at the first glance.. But not a lie.
In documentation for the instrument they make it clear that they decided to call "Peak Detect mode" same thing that other manufacturers call "Envelope detection".

And I agree with you, I hate when people don't stick with the standards, reinventing terminology, deliberately or not...  They might decide to call Volt or Ohm something else too.. 

Best regards,

Sinisa

[nctnico]

Peak-detect and envelope mode are not well defined so you have to read the manual to see what it actually does. Things get interesting when both peak detect and envelope mode are supported.

[David Hess]

But.... Technically, you are wrong.. If THEIR User manual defines "Peak Detect mode" the way they do, they dont lie.

They decided to define term "Peak Detect mode" differently than Tektronix or other US manufacturers.. It is sleazy (if deliberate), stupid, it is misleading at the first glance.. But not a lie.

In documentation for the instrument they make it clear that they decided to call "Peak Detect mode" same thing that other manufacturers call "Envelope detection".

And I agree with you, I hate when people don't stick with the standards, reinventing terminology, deliberately or not...  They might decide to call Volt or Ohm something else too. 

I might forgive them for that if it was the only place Rigol did it but it was not.  They systematically redefined several industry standard terms including peak detect, delay, delayed trigger (1) to make their product look good and then doubled down when I asked them about it insisting that their oscilloscopes supported functions which they do not.

Further, they *changed* the definition when they released later oscilloscopes which do support peak detection.  The Rigol DS1000E defines it one way and the DS1000Z another with the later being the industry standard for decades.  So the old definition needed to be changed when they did not support it but once they supported it, the old definition was fine.

Things get interesting when both peak detect and envelope mode are supported.

Offhand I do not know of any DSOs which support peak detection without also supporting envelope detection in one way or another but when tracking this down years ago, I ran across something funny.

The documentation for the old Tektronix 2440 series calls it "glitch capture" but there is no such glitch capture or peak detect mode; there is only envelope mode.  On these DSOs, peak detection in the traditional sense is always active when envelope mode is used and when you set the envelope mode to accumulate 1 acquisition, that is the same as peak detect mode on other DSOs and it does exactly the same thing.  I just always found it odd that they did not include it as a separate mode or refer to it as peak detection.

This threw me off when I was evaluating the DS1000E because I kept looking for a way to adjust the number of acquisitions in the displayed envelope to set it to 1 but there is no such adjustment and the hardware cannot support peak detection during decimation.

(1) And I would add record length to that list as well.  If measurements are made on the display record, then what is the actual record length?

[2N3055]

David,

Thanks for explanation... Yeah, that's sleazy... I agree..
I hate marketing tricks...

Regards,

Sinisa

[tggzzz]

This is a simple 10kHz waveform output.

The old analogue scope display is clear and simple - and correct.

But the (single shot) digitising scope display is woefully incorrect. When not in single shot mode, the trace twinkles like a demented Christmas decoration!

The DSO needs peak detection (1) to display this waveform correctly which the Rigol DS1000E series does not despite what its lying manual says.  When I first evaluated Rigol's DSOs (before the DS1000Z series became available), I asked their customer service about it and just got more lies.  "Of course they support peak detection, it is right there in the manual!"  Having Rigol double down was very frustrating.

My 27 year old 2232 DSO would have no trouble displaying this and it has done so.  My slightly older 2230 might have problems because its peak detection only goes down to 100ns.

I'm not overly interested in Rigol bashing per se; my points are more general and related to low-end digitising scopes in general.

In this case I'll settle for envelope detection as being sufficient. That's also sufficient for measurement of the period (albeit with unnecessarily limited resolution on this particular scope).

I have no idea why the burst pulse is shown incorrectly on the DS1000E; that is just broken.  The bandwidth is not high enough to reproduce the edges but the sample rate and bandwidth should be plenty for the peaks and where is that trailing edge trash coming from?  That is not caused by low bandwidth or insufficient sample rate.  I call shenanigans; that DSO is broken by design unless there is a problem with the probe.

Yes, and I don't fully understand that display either.

It may be there is something I have missed deep in a menu; but since I'm scarcely a beginner if that's the case then it illustrates the problems that beginners can and do have.

(EDIT added for clarification. That is a very common waveform in hard realtime systems, when you are blipping a bit to see how fast/slow a loop is occurring, or  the latency between two events, typically a stimulus and a response)

I have often made this measurement in real time systems.  With a good oscilloscope you can measure the distribution of response time or latency and capture long latency glitches which completely destroy real time performance.  If the distribution is gaussian, then you can even measure it quantitatively with a delta delayed sweep on an analog oscilloscope.

Yes indeed.

Clue: Toggle the pin instead of blipping it then you can use the DSO's pulse width measurement to get the elapsed time instead of trying to count fractions of squares on a a green screen.

That is a good work around for an oscilloscope which has trouble making this measurement but instead of toggling, I set and reset so the pulse polarity is always consistent.

Just so.

(You do have pulse width measurement on the Analog 'scope, right?)

Mine does.

So does mine, but I didn't want to further distract Fungus down unimportant alleys.

[rstofer]

I've also noted that professional sometimes hammer screws into wood, except for the last turn.

That doesn't mean that amateurs can.

Hammering in a screw takes years of experience. 

Or the lack of a screwdriver...

[Paul Moir]

Out of curiosity, what was the bit rate? ...

Sorry, I had it cranked way down to I think 50kHz but not for photographic purposes.  I only had one 10x probe with me so the data channel is being served by one of those Ponoma minigrabber to BNC lead sets. 
EDIT:  This was done on a Tek 465, a single beam scope.  On data high rates I guess you couldn't view this.  Chop won't work with fast signals and ALT will capture (likely) the next data packet rather than the one associated with the displayed SCLK signal.  So DSOs are better after all! 

[David Hess]

(You do have pulse width measurement on the Analog 'scope, right?)

Mine does.

So does mine, but I didn't want to further distract Fungus down unimportant alleys.

I would not mind having a DSO or universal counter which can provide a histogram of the pulse timing but this is a high end feature and measurement cursors work well enough.

[tggzzz]

(You do have pulse width measurement on the Analog 'scope, right?)

Mine does.

So does mine, but I didn't want to further distract Fungus down unimportant alleys.

I would not mind having a DSO or universal counter which can provide a histogram of the pulse timing but this is a high end feature and measurement cursors work well enough.

Curiously I'm using the XMOS and DevKit processor to count edges in two independent 16ns period bit streams (call them Fin and Fref, both 1Hz-15MHz), do some arithmetic to create two reciprocal frequency counters, and report the Fin/Fref results up a USB link to a host PC - all in software with no added hardware other than input buffers and protection diodes.

When the "work of art" is "completed", I may move onto a statistical frequency counter. Your histogram doesn't seem so far removed from that

[nctnico]

(You do have pulse width measurement on the Analog 'scope, right?)

Mine does.

So does mine, but I didn't want to further distract Fungus down unimportant alleys.

I would not mind having a DSO or universal counter which can provide a histogram of the pulse timing but this is a high end feature and measurement cursors work well enough.

My GW Instek has some statistical functions which can analyse recorded segments and show a histogram using results from an automatic measurement.

[Mechatrommer]

corporate specification: 200Mbps digital (square) signal, that is somewhere 2GHz BW or above...
the corporate wise decision was... use 1GSps 50MHz BW obsolete model DSO...
talking about choosing the right tool for the right job?

ok ok the point was. higher BW analog can be had at cheap...
ok point taken but... too many other points hidden...
a very well known strategy try to take over the leader... distraction...
show my 1 strong point, hide my other 100 weak points...
ee is just not about seeing led blinking at 200Mbps...

conclusion: never send a programmer to do engineers job...

[tggzzz]

corporate specification: 200Mbps digital (square) signal, that is somewhere 2GHz BW or above...
the corporate wise decision was... use 1GSps 50MHz BW obsolete model DSO...
talking about choosing the right tool for the right job?

ok ok the point was. higher BW analog can be had at cheap...
ok point taken but... too many other points hidden...
a very well known strategy try to take over the leader... distraction...
show my 1 strong point, hide my other 100 weak points...
ee is just not about seeing led blinking at 200Mbps...

conclusion: never send a programmer to do engineers job...

I'm not entirely sure what point you are making, but I refer you to paragraphs 2 and 6 in the first post ("The standard statement..." and "Before replying...").

[Mechatrommer]

That statement should provoke the immune response
The standard statement seen all too often on this forum is that an entry-level Rigol scope is better than an old analogue scope. But is that really true for a common basic use-case:

The first point to note is that the 80:20 transition times of the output is 1ns (90:10 is 1.8ns). Using the BW=0.35/tr rule-of-thumb, that implies a 350MHz scope is required to measure the risetime. Hence a 50MHz scope is completely inadequate when checking signal integrity of a modern microprocessor's outputs. There is absolutely no doubt about that, and no surprises either.

Before replying, please note my repeated statements elsewhere in the forum: choose the right tool for the job. Occasionally a digitising scope is necessary, occasionally an old analogue scope is necessary, frequently either are sufficient.

1) you are not using the right tool for the job...
2) as a way of saying. hammer is bad than a screwdriver because hammer cant turn a screw.. all too often people saying hammer is better than screwdriver... whats missing is at putting nails into the wood
3) hence yeah... (refer the colored statements)... the OP is full of contradictions and missing parts... this thread is a troll... someone should lock this down..

[tggzzz]

That statement should provoke the immune response
The standard statement seen all too often on this forum is that an entry-level Rigol scope is better than an old analogue scope. But is that really true for a common basic use-case:

The first point to note is that the 80:20 transition times of the output is 1ns (90:10 is 1.8ns). Using the BW=0.35/tr rule-of-thumb, that implies a 350MHz scope is required to measure the risetime. Hence a 50MHz scope is completely inadequate when checking signal integrity of a modern microprocessor's outputs. There is absolutely no doubt about that, and no surprises either.

Before replying, please note my repeated statements elsewhere in the forum: choose the right tool for the job. Occasionally a digitising scope is necessary, occasionally an old analogue scope is necessary, frequently either are sufficient.

1) you are not using the right tool for the job...
2) as a way of saying. hammer is bad than a screwdriver because hammer cant turn a screw.. all too often people saying hammer is better than screwdriver... whats missing is at putting nails into the wood
3) hence yeah... (refer the colored statements)... the OP is full of contradictions and missing parts... this thread is a troll... someone should lock this down..

Sigh; there's definitely a lost-in-translation issue here, somewhere.

[691175002]

For what its worth, when I was looking to buy an entry level scope analog just wasn't an option.

For a complete beginner buying used equipment of unknown condition is a massive risk.  Not only are you incapable of repairing problems (even simple ones) with the scope, but you don't even feel qualified to judge whether the scope is working properly in the first place.

Used scopes can become pretty pricy outside of the USA, and spending that kind of money just doesn't seem smart when the 1054z exists.

Now when you start looking to buy a second scope many of the used analogs start to become very interesting.

[tggzzz]

Sorry, read the title then saw it was about Rigol scopes vs some analog relic.  I gave away my old Hitachi analog scope and my newer Lecroy but kept the first one I ever owned, the 7200.   VME chassis, 68K based relic from the USAF's past.   Not exactly what I would consider the high end of the curve.   I would even go so far as to call it a very low end DSO by todays standards, or even my other DSOs.   

I'll have to check the little micro use used.   Rather than go that route, I just programmed my old Sony Arb with your pattern and roughly the same levels.   From the 2nd picture, you will see this is a single shot.   Third picture, turned on the auto measure and RIS with the 4GHz sampler.   

Like I said, it's a relic but I still use it any time I think I may damage something.  I can't think of a time I would have needed an analog scope or an analog handheld meter for that matter.   I do still have my vacuum tube grid dip meter.

I hadn't spotted that you had changed the contents of your posting, adding significant content.

Bear in mind that my whole posting is not related to expensive professional scopes, but is related to the class of cheap scopes frequently and forcefully recommended in this forum - and to what extent they are suitable for bit-bashed digital signals.

What's the 7200's basic spec, particularly the front-end analogue bandwidth? Given that LeCroy is a long-standing decent professional brand, I would hope and expect  that it would perform "without surprises". I don't see any surprises in the second screenshot.

I don't understand what the Arb Generator screenshot is showing, and can't see the waveform it is generating. Certainly 400kHz is much slower than the waveform I generated (bit period 4ns). The voltage levels ought to be irrelevant.

I don't understand what the last screenshot is showing, nor how it relates to the issue I've mentioned.

If you are interested in hard realtime multicore parallel systems, the XMOS processors and xC are definitely worth understanding. Their pedigree (people, theoretical basis, practical experience) dates back to the 70s in the form of CSP, Occam, Links/Channels and the Transputer. Some of the concepts have found a place in very different systems, e.g. some TI DSPs.

[tggzzz]

For what its worth, when I was looking to buy an entry level scope analog just wasn't an option.

For a complete beginner buying used equipment of unknown condition is a massive risk.  Not only are you incapable of repairing problems (even simple ones) with the scope, but you don't even feel qualified to judge whether the scope is working properly in the first place.

Used scopes can become pretty pricy outside of the USA, and spending that kind of money just doesn't seem smart when the 1054z exists.

Now when you start looking to buy a second scope many of the used analogs start to become very interesting.

We pretty much agree.

A beginner should only have a working scope, of whatever type. Taking on a "restoration project" is only for those that enjoy doing that and are prepared to take some risks!

If money is a serious consideration, which is likely on this forum, then it may be worth considering a very cheap very basic analogue scope. I don't know where you are based, but in the UK if you are patient you can find working examples at about £1/MHz, and frequently they are simply given away. (My local hackspace will shortly be throwing half a dozen out, unless members want them!)

£20 for a 20MHz scope is a viable way to learn about scopes in general, and to find out what you really do and don't need, and why. Think of it as a cheap and valuable "learning experience" before you spend significantly more money.

And I agree, while there's little point in having two very similar scopes (or tools), it can be worth having significantly different scopes (or tools) - so that you can use the unique advantages of each as appropriate. That's why I also have a new 10MHz (not 100MHz) digitising storage scope.

[Specmaster]

I have been reading the arguments and counter arguments with much interest and I have to be completely honest and say that much of it is way above my level of understanding but speaking purely as a person who is effectively getting into electronics again after about a 40 year break and also maybe from a different perspective to some of the members contributing to this thread. My interest lays in the ability to do relatively simple things like fault find on broken items of consumer electronics such as radios etc then it becomes a simple choice, spend a little (especially when your like me, retired on a fixed income) or if you can afford it go for something with all the bells and whistles on it such as a DSO. A cheap analogue scope is just as capable of telling you if the item under test be it a transistor, valve or an IC etc is A/ receiving an input and B/ giving an output and that I would suggest is basically what an awful lot of people really require. Anything else is in some cases is just guilding the lily, i.e. nice features to have but how often will the extra functionality be used or indeed actually required.

The scope needs to be matched to likely and expected work it is going to be used on and I can't help but think that sometimes features on scopes etc are for a lot of people, like features on cars, bragging points that can used in a game of top trumps. My car has a TV built in and surround sound cinema system, great for bragging rights but how often have I used it, about once in 4 years of owning the car.

[tggzzz]

The scope needs to be matched to likely and expected work it is going to be used on and I can't help but think that sometimes features on scopes etc are for a lot of people, like features on cars, bragging points that can used in a game of top trumps.

That is an absolutely key point that many people on this forum repeatedly mention - and many people dogmatically don't mention. The aphorism in my .sig has direct analogies with electronics, as your posting indicates you already know

Apart from that, "welcome back". It is off-topic, but you'll find many things have changed depressing little; the fundamentals are still more or less the same. The major differences are a smaller, much cheaper, and significantly faster. Embedded computers haven't changed: still mostly C on 8/16/32 bit single-core machines. The major technology revolutions are in the speed/resolution of affordable ADCs and DACs, and nano-power technology.