Author Topic: FLUKE PM3394B sampling strange behavior  (Read 1277 times)

0 Members and 1 Guest are viewing this topic.

Offline cafi95Topic starter

  • Newbie
  • Posts: 7
  • Country: fr
FLUKE PM3394B sampling strange behavior
« on: May 14, 2021, 10:18:03 am »
Hi everyone,

I'm a French newone on that forum and I would have been glad to introduced myself  on the forum, but I haven't found where. So I do it on that thread.

I'm a former electronic  & TCM ingenier long years ago but I still work on electronic devices (vintage audio, ham radio, laser biological applications ... and lot of other stuff I generally design by myself).

So, recently, I've experienced a tricky issue on my Fluke PM3394B 4 channels combiscope (full options equiped)

I'm using that outstanding device from more than ten years without any problem.

Description of the issue :

Until MTB is not less than 250ns/div, everything is ok.

But below (so from 200ns to 2ns), the sampling rate appears to be much too slow.

On the trace, sampling dots can be clearly seen ...

In addition,  when a signal is active and the input mode is toggled to ground, it takes a lot of time for the trace to return to the ground state. It comes to that state dot after dot...

At the 2ns MTB setting, it needs more than 5s for the display to stabilize at a ground level.

Of course,  below 250ns/div, no single event can be displayed,  for the same reason.

Question : do you experience such a behavior or is it a scope issue  ?

To be noticed :

- I'm sure every parameter has the right setting.

- everything is nominal in analog mode.

Many thanks for any advice.

All the best.

Philippe  ;)
 

Offline wn1fju

  • Frequent Contributor
  • **
  • Posts: 553
  • Country: us
Re: FLUKE PM3394B sampling strange behavior
« Reply #1 on: May 14, 2021, 11:14:31 am »
Similar behavior on my PM3390B Combiscope.  It is because below 250 ns, it uses equivalent time sampling.  In other words, the scope constructs a picture of the input signal by accumulating the samples over many wave cycles.  Thus it takes time for the screen to change and the display to settle down.

You will note that it is not a time-base problem with your unit since the scope will work fine in analog mode below 250 ns/div. 

You can turn off the equivalent time sampling in the TB menu and you will see that the scope won't allow you to go less than 250 ns/div. 

A lot of the early digital scopes used equivalent time sampling.  Works great for repetitive signals only.
 

Offline cafi95Topic starter

  • Newbie
  • Posts: 7
  • Country: fr
Re: FLUKE PM3394B sampling strange behavior
« Reply #2 on: May 14, 2021, 05:29:08 pm »
Many thanks for your answer   :-+

That was exactly my thoughts but I was'nt quite sure about it (and as a matter of fact, setting the real time sampling to ON, limits the TB to 250ns/div) ...

Fine !

The spec (and the user manual) are not very clear on that point.

So, the best to catch a single event of that speed is to set the TB at 250 ns/div and then, to magnify the trace ...

My incoming night will be peaceful  :=\

All the best. ;)

Philippe
 

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5452
  • Country: de
Re: FLUKE PM3394B sampling strange behavior
« Reply #3 on: May 15, 2021, 07:20:33 am »
Under 250 ns, the PM3394B scope goes into a special acquisition mode.
Maybe you can post a picture of how it looks.

 
There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline cafi95Topic starter

  • Newbie
  • Posts: 7
  • Country: fr
Re: FLUKE PM3394B sampling strange behavior
« Reply #4 on: May 15, 2021, 01:13:35 pm »


You're right.

On that example, channel 1 has just been grounded. Just notice that the other trace (still busy) is affected by the same behaviour. That's quite logical as the sampling is on both channels  ::)

 

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5452
  • Country: de
Re: FLUKE PM3394B sampling strange behavior
« Reply #5 on: May 16, 2021, 08:09:46 pm »
Something is wrong with your scope.

How does the signal look if nothing is hooked up and both channels set to ground?

There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline wn1fju

  • Frequent Contributor
  • **
  • Posts: 553
  • Country: us
Re: FLUKE PM3394B sampling strange behavior
« Reply #6 on: May 16, 2021, 08:43:08 pm »
You have to be careful.  It is both the equivalent time sampling (ETS) and the trigger that is producing the behavior in your picture.

1.  If you take a 100 MHz signal, for instance, and hook it up only to ch#1 with a 10 nsec timebase on the (digital) PM3394B, and you are triggering on ch#1, then you will see the effects of the ETS as you enable and disable the input signal (takes some time for the waveform to appear and disappear).
It is annoying, and probably is further complicated by residual capacitance in the acquisition, but my other vintage digital scopes do the same thing.
The PM339x seems a bit slower than most.

2.  If you take the same 100 MHz signal and supply it to ch#1 and ch#2, then what you will see depends on the trigger.  For instance:

   a.  If you are triggering on ch#1 and enable/disable the signal input on ch#2, then ch#2 will change rapidly without seeing much of the ETS.  Obviously, the ch#1 waveform will remain unchanged.

   b.  If you are triggering on ch#1 and enable/disable the signal input on ch#1 (your picture), then ch#1 will slowly bleed off to zero via ETS and ch#2 will look like it is - but it really isn't.  It's now an untriggered waveform so it will look like random garbage.  That's not from the ETS.

   c.  Same thing will happen if you are triggering on ch#2, but with the conditions in a. and b. reversed.

Bottom line:  I think your scope is doing what it is supposed to.  If not, then yours and mine are both broken!  It is also why I am not a big fan of the early digital scopes from the early 1990's.  I'd rather use a 1990's analog scope.  With your Combiscope, you have both.
 

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5452
  • Country: de
Re: FLUKE PM3394B sampling strange behavior
« Reply #7 on: May 17, 2021, 10:28:11 am »
All my PM3394B have the same behavior.

When I swich the MTB from 10ns  to 5ns, it will cause some spikes in the fast edge of a signal.
But after 1/10 of a second these spikes are gone and the fast edge is shown clearly.

See attachments.

 

There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline cafi95Topic starter

  • Newbie
  • Posts: 7
  • Country: fr
Re: FLUKE PM3394B sampling strange behavior
« Reply #8 on: May 17, 2021, 07:13:17 pm »
That sounds good !

In fact you're right, there is an issue.

First screenshot: every input grounded. Everything looks ok.

Second screenshot: Trace 1 is just moved downwards using the appropriate knob. 45 s is needed for the new trace to stabilize !

According to specs, for 2 ns/div,  the acquisition time is 2s, that's to say within 2s, 99% of the samples must have been updated.

Your observation : 1/10 th of a second (at 5ns) !

Clearly,  my scope is far from that.

I have now to investigate the digitalizer diagram to figure out what could be wrong. Fortunately, I have some spares ... But before to do anything, I must understand.

Definitely not a piece of cake  |O
 

Offline wn1fju

  • Frequent Contributor
  • **
  • Posts: 553
  • Country: us
Re: FLUKE PM3394B sampling strange behavior
« Reply #9 on: May 17, 2021, 10:06:21 pm »
To both HighVoltage and cafi95:

The pictures shown from both of you are from two different experiments:

Experiment 1:  Put in a signal and change the timebase
Experiment 2   Set the input coupling to ground and simply move the vertical position knob

With my 3390B, I get similar results to both of you.  Namely, in Experiment 1, when I change the timebase (say from 10 ns to 5 ns) with say a 100 MHz sinewave, the display changes almost instantaneously (like HighVoltage's).  However, in Experiment 2, when I ground the input and twist the vertical position knob, it can take upwards of 30 seconds for the display to give me a flat line (like cafi95's).

Moreover, it I mix the two experiments, namely I take the 100 MHz signal and move the vertical position knob, the display changes almost instantaneously.

So there appears to be something going on in the absence of a signal.

I would like to know if HighVoltage sees the same thing for Experiment 2. 

Finally, I have a Tektronix TDS520 digital scope and it does exactly the same thing.  Instantaneous changes with timebase or vertical position when a signal is present, and a large delay (about 5 seconds) when there is no signal (input grounded) and I move the vertical position knob.

I'm open to suggestions as to why....
 

Offline cafi95Topic starter

  • Newbie
  • Posts: 7
  • Country: fr
Re: FLUKE PM3394B sampling strange behavior
« Reply #10 on: May 18, 2021, 10:27:20 am »
That's a very good observation.

I performed the same experiments and I got the same results.

On 2 ns, with an active 20 Mhz signal on ch1, moving trace needs only 3 s for all samples to be updated.

On that same configuration, with ch2 grounded, moving trace 2 is quite instantaneous.

So you're fully right, there  seems to be a strange behavior when no active signal is present on all channels.

Perhaps is it a matter of triggering that does'nt start a proper sampling (below 250ns/div).

I'm on holidays for a week and I'll start a trial campaign as soon as I  come back.

That's an interresting tricky thing to understand.  :-BROKE
 

Offline cafi95Topic starter

  • Newbie
  • Posts: 7
  • Country: fr
Re: FLUKE PM3394B sampling strange behavior
« Reply #11 on: May 19, 2021, 09:19:22 am »
I haven't noticed yet the answer #6 of WN1FJU.

I do agree with everything you've said as I experimented the same things. 

Well, now I understand the limit zone of that scope, I will be able to use it without misinterpretation.

That was the first time I noticed that phenomenon but as my ham radio call sign has been reactivated , the frequencies I observe are much higher than before !

So really many thanks for your help during that thread .

Really good job !

I will stay active on that forum as I'm using a lot a vintage apparatus .

May be one day someone will need help about one of them.

 

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5452
  • Country: de
Re: FLUKE PM3394B sampling strange behavior
« Reply #12 on: May 19, 2021, 09:57:57 am »
Hello wn1fju

Interesting, I never had a need to use my scopes this way in the ns-range and never observed this behavior.

I just did your proposed test and can confirm that the it is as you said.
Having a second channel with the signal on and grounded will cause the spikes to stay on for a long time.

I need to dig out my fast 1 GHz generator to test this again with a fast sine wave of 100 or 200 MHz
But from what I see now, this is a limitation of this scope.

As long as we know what the limitation is, we can work with this.


For fast signals I use modern Keysight scopes anyways and not these old Fluke scopes.
But having said that, I love these Fluke scopes for many other purposes where modern scopes will fail.

There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline wn1fju

  • Frequent Contributor
  • **
  • Posts: 553
  • Country: us
Re: FLUKE PM3394B sampling strange behavior
« Reply #13 on: May 19, 2021, 05:27:51 pm »
Yes, I'm convinced it has something to do with the triggering.  The only time I see this long delay on the screen is when the red ARM'D LED on the front panel is lit indicating no trigger.  This can be in the two previously described situations where the input is grounded and the vertical position knob is rotated, or when you disconnect ch#1 (trigger channel) and there is a waveform on ch#2 (both traces then take a long time to disappear).

If Fluke/Philips is using a digital trigger and there is no signal to trigger on, perhaps the software doesn't know what to do, or has to wait for enough noise spikes to trigger something?  Or maybe in this case it just forces triggers at a much slower rate so at least some trace shows up on the screen.

But no big problem in any event.  The original post asked the scope was working correctly.  The answer seems to be yes.
 

Offline cafi95Topic starter

  • Newbie
  • Posts: 7
  • Country: fr
Re: FLUKE PM3394B sampling strange behavior
« Reply #14 on: May 19, 2021, 09:47:51 pm »
I'm very happy about that.

It's the best scope I own and I'm using it for more than 15 years !

What's wonderfull with it in particular are the math functions which are gorgeous.

One day I had to observe a laser flash lamp discharge in order to design the right PFN (pulse forming network).

I was able to view in real time the voltage, current, instant power and energy on a single flashlamp shot (1kV, 200 J). Everything on a single screen with the right units !

My previous scope was an Hameg 412 from the 80's ! ( but still working as new).

What such a terrible gap for me !

But well, next week I'll keep going with some new experiments to sort that triggering question out.

Just for fun ...
 

Offline shakalnokturn

  • Super Contributor
  • ***
  • Posts: 2100
  • Country: fr
Re: FLUKE PM3394B sampling strange behavior
« Reply #15 on: May 20, 2021, 03:18:44 pm »
I have one of these scopes too, nice model for their time still handy today as a general purpose scope. I appreciate the bandwidth and ability to switch analog/digital.

I haven't tried to confirm it behaves the same way, but that's the way equivalent time sampling works, on fast repetitive signals they'll sample one point of the input on every trigger event.
You don't actually ever see a single waveform on the display, it's a collection of as many samples as memory depth used, each at a different time from trigger and different cycle of the input waveform.

If there is no fast trigger event and the scope slowly updates by itself (why they do slow trigger I don't know) it takes a long time to refresh the whole memory.
It isn't much of a limitation, ETS has no use on DC.

If you like the math functions on these Combiscopes, try an old LeCroy 9344 or above, there's so much more!

 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf