afg1062 setup

Hello,

I'm trying to realise two pulses, one on CH1 and another on CH2 where i can change the time between them with CH1 as reference (a start/stop setup).
the setup of the AFG1062 is as follows: CH1 in pulse (burst) and CH2 in pulse/burst.  When triggering the burst, I see that the polarity of the output channel 2 is not consistent with the gear's display and is outputting a negative polarity in function of the StartPhase.  I did a manual sweep between -360° and +360° and noted the output polarity with below the results:
-360 (-)
-354 (+)
-352 (-)
-348 (+)
-324 (-)
+2 (+)
+36 (-)
+360 (-)
All values in between have the same polarity until the next change and should all be positive?
As my circuit is flank driven this is a very annoying behaviour.
Is there a way to solve this?

Thanks for helping on this:

 

Hi Hans,

There are known issues with changing phase on the AFG1000.
I've found that upgrading to the latest version (v1.1.0) fixed these issues on my instrument.

Hi Alfonso,

That upgtrse (?) is the first I did in the hope this would be solved but clearly without success.

I also thought that a minus phase shift on Ch2 with respect to CH1 would result in a leading CH2 pulse which is not the case.

To be honest I'm quite disappointed about  this instrument...

Hi Hans,

I certainly understand your disappointment with the instrument, I am working with our product line team to keep fixing these problems.

I have found that pressing "Align Phase" in the "Inter CH" menu fixes the phase alignment issue, at least until you change the phase angle again.
Additionally, from my testing, different waveforms and run modes had less problems with phase alignment, so I'd suggest you try out other settings, if possible.

Hello,

I've been playing with the different modes to get a better understanding of the instrument but finally I need to stay within the TTL limits so I switched back to pulse.
I tried the following setups:
CH1 PULSE/BURST, Slope positive, start phase 0°, trigger external, width 500 ns, frequency 300 kHZ
CH2 PULSE, start phase variable,  width 500 ns, frequency 300 kHZ
Not usable setup as it looks there is is no relation between CH1 and CH2 regarding  inter pulse timing.
The polarity is ok for both channels.

CH1 PULSE/BURST, Slope positive, start phase 0°, trigger external, width 500 ns, frequency 300 kHZ
CH2 PULSE/BURST, Slope positive, start phase variable,  width 500 ns, frequency 300 kHZ
For the negative start phase the timing between the start of the pulses looks ok (using the formula (1/frequency) x startphase / 360°) except the polarity of the signal on CH2 is negative.
For the positive start phase the positive pulses are stretched by +/_ a factor 5.  The timing between the pulses follows here the formula
period - {(1/frequency) x startphase / 360°)} with the negative polarity on CH2.  When the polarity is positive the timing between the pulses is not between both leading edges but between CH1 leading edge and CH2  trailing edge!!

For the measurements I did the scope always triggers on CH1 of the AFG.
Is there a possibility that these problems can be solved?



 

Hi Afonso,

Could you please elaborate a little more your fix to the problem.  Plus I don' grasp "The AFG uses the last point of the pulse as the value to hold between pulses, this is what makes it look like the pulse "inverts"."

I lowered the frequency to 10 KHz and sill get an negatve pulse on CH2 with following setup:

CH1 puls / burst, frequency 10 KHz, pulse width 20 us, 1 cycle, trigger external, 0° StartPhase

CH2 puls / burst, frequency 10 KHz, pulse width 20 us, 1 cycle, trigger external, variable StartPhase.

The result is as in the added pictures: 

1) For 60° on CH2 the polarity is positive, the pulse width is stretched  4 times (this varies also) away from the 20us

2) For 170° on CH2 the polarity is negative (!!) and the width correct.

I zoomed in on the scope, you can see the output of CH2 stays high or low in function of the StartPhase!

A related thing I found is that without triggering the AFG externally and sweeping the StartPhase from -360° to +360°  the output of CH2 goes (in my case) from 0V to 3V and back to 0V  several times.  I suppose this is NOT the correct behaviour of the AFG!!

AMHO this is a bug :-(

Groeten,

Hans

next pictures

Hi Hans,

To set a certain phase the AFG has to know your frequency, which it cannot possibly know when it's externally triggered, so it uses the pulse frequency. The problem is that this isn't the correct frequency and so when you increase your phase angle to such a point that the pulse overlaps the end of the period, you get a situation where the pulse looks stretched out as it stays high between triggers.
This is easy to test, if you set your external trigger frequency to the same as your pulse frequency, you will see it looks as you'd expect, but if you then change the external trigger frequency you will see the problem reappear.

So yes, this behavior is surprising, but not incorrect.
I will say that on the AFG31000, when generating an externally triggered pulse it doesn't allow you to set phase. Instead, you can set a delay and this delay is limited such that the pulse never overlaps the end of the waveform, avoiding this problem. The designers of the AFG1000 decided not to do it this way, which leads to some surprising behavior, but isn't exactly wrong.

This is a difficult issue to understand so I apologize if I haven't done the best job at explaining it.

Hi Afonso,

I still don't get the explanation of the AFG1062's strange behaviour!  And how do you then explain that the output of CH2 changes when changing the startphase?  Shouldn't it be zero all the time until triggered??

Also the signal on the display doesn't corespond the the sinal on the output, although this is the least of my problem.

Groeten,

Hans

Hi Hans,

And how do you then explain that the output of CH2 changes when changing the startphase?  Shouldn't it be zero all the time until triggered??

This is a good question and it hints at the core reason behind this unexpected but correct behavior. When the AFG is waiting for a trigger, it holds the output at the last point of the waveform. When your angle is 0 degrees, this point is 0 V, but at 1 degree the waveform loops around and the last point actually becomes 3 V, this is what causes the apparent "inversion" of the pulse.

Also the signal on the display doesn't corespond the the sinal on the output, although this is the least of my problem.

This is because what the signal looks like on the oscilloscope depends on your external trigger, which the AFG doesn't know beforehand so it cannot possibly draw the signal correctly.

I think a good away to visualize this is if you vary the trigger frequency you will see the pulse shape change, which is an inevitable consequence of what I mentioned above.

Defining phase for an externally triggered pulse doesn't really make sense, which is why in our new AFG31000 we don't allow you to set the phase, instead you can set a delay, which is much more intuitive in how it works. So this isn't a bug in the AFG1000, it's working as intended, but it would be better if they allowed you set a delay instead of phase.

Hi Afonso,

How can you then setup a test environment when the gear behaves 'unexpected but correct"?  At my humble opinion this is a contradiction.

This implementation of startphase is thus completely useless for doing any time measurements!  The "delay" approach makes indeed more sense than the phase approach which makes the AFG1062  unusable for certain measurements.

I guess this weird implementation can be rectified by your product line team to keep fixing these problems....

A happy customer anyway ;-)

Groeten,

Hans

Hi Hans,

I agree with you that this is something which can be improved on the AFG1062 and I will be working with our product line towards that.
Thank you for bringing this to our attention and working with me to understand it!

Regards,
Afonso Teles

Hi Afonso,

I'm pleased you got the insight that there is something wrong with the behaviour of the instrument.

I hope your team is choosing for the "delay approach" instead of the phase one.

If you allow, I'm willing to test new releases.

Groeten,

Hans

Hi Afonso,

Is there any progress towards a solution concerning the unpredictable behaviour of the phase setting in burst mode?

Groeten,

Hans

Hi Hans,

I haven't received an update no.
There's a very good chance this won't be changed as it's working properly, it implements a phase shift as commanded, but it should instead be done as a time delay, in my opinion. This would require a feature change which would break backwards compatibility for other users.

I'll let you know once I get a more definitive answer though.