Pulse at end of buffer with funcPlayPattern

[Meindert Norg]

Hi,

I'm playing back a 5 second long 8 kHz sine signal at 6 MHz using funcPlayPattern. (30e6 sample) The sine signal is modulated with a ramp which changes from 1 to 0 over the length of the sine signal, resulting in an amplitude of ~1 at the start of the 5 sec playback, and 0 at the end. 

The issue is that if I zoom into the tail end of the 5 seconds data set on my o'scope, I can see an unexpected  ~50 ns long pulse. See scope photos. 

Any thoughts on what might cause that pulse and how to prevent it? 

Input data: 

1 sec / div

2 ms / div

50 usec / div

500 ns / div with the unexpected pulse triggered.

 

 

I uploaded the file which I use to switch back and forth between using a PlaybackOnlyExample2.pypre-recorded signal (not included) and a synthesized signal. The latter is the default setting, generating the signal at output 2. 

 

[attila]

Hi @Meindert Norg

Have you tied with unlimited run time ?
dwf.FDwfAnalogOutRunSet(hdwf, iWavegen, c_double(0.0))

Generating such sweep may be more accurate by using the device's AM/FM... see WF SDK/ samples/ AnalogOut_Sweep.py

[Meindert Norg]

Using

dwf.FDwfAnalogOutRunSet(hdwf, iWavegen, c_double(0.0))

made it 'worse', See photo below. 

Note that I'm ultimately not really to make a sweep of any kind (hence I think the AnalogOut_Sweep.py example will not suit my needs). I want to repeatedly play back a 6 MHz synthesized (home-made) signal and (for now) am using a sine with decreasing amplitude to prove that the output looks like what I generated. 

 

I was able to reproduce the issue (PlaybackOnlyExample3.py) to its simplest form where instead of a sine I'm now using a slope from 1 to 0. At the point where the waveform is repeated, the analog output is

1. set to the new value (you can test that by changing the start value) for ~50 ns, then 
2. set to the old value (you can test that by changing the end value) for ~500 or 600 ns
3. set to the new value to start the actual playback of the signal. 

PlaybackOnlyExample3.py

10 Hz slope from +1 to -0.5 V, at a playback frequency of 6 MHz (166 ns period).

Zoomed in to where the signal jumps from -0.5 V to +1 V and starts at the start of the downwards slope again:

 

Any thoughts? 

 

Edited Wednesday at 01:50 PM by Meindert Norg

[attila]

Hi @Meindert Norg

The play glitch is corrected in the latest version.
The finite run with repeat requires a few nanoseconds of wait to fetch data from DDR RAM.
You can use continuous run and the waveform will be repeated without any delay.

 

[Meindert Norg]

Thank you Attila!

Rev beta v3.23.19 indeed fixed the pulse and now continuous run 

dwf.FDwfAnalogOutRunSet(hdwf, iWavegen, c_double(0))

also works. Thanks again.

Meindert
 

[Meindert Norg]

Attila, 

I indeed measure about 30 ns of 'drift' that you contribute to fetching data from RAM. However, I don't seem to be able to eliminate that. 

Can you please educate me on how to configure what you call 'finite run with repeat' vs 'continuous run'? 

I interpret 'finite run with repeat' to be set with:

WaveFormLib.FDwfAnalogOutEnableSet(hdwf, channel, 1) 
WaveFormLib.FDwfAnalogOutFunctionSet(hdwf, channel, dwfc.funcPlayPattern)
WaveFormLib.FDwfAnalogOutFrequencySet(hdwf, channel, ctypes.c_double(hzRate))
WaveFormLib.FDwfAnalogOutAmplitudeSet(hdwf, channel, ctypes.c_double(ampOutVoltageRange))
WaveFormLib.FDwfAnalogOutRepeatSet(hdwf, channel, ctypes.c_double(0))
# WaveFormLib.FDwfAnalogOutRunSet(hdwf, channel, ctypes.c_double(0))
WaveFormLib.FDwfAnalogOutDataSet(hdwf, channel, rgdPlay, nrSamples)
WaveFormLib.FDwfAnalogOutConfigure(hdwf, channel, 1)

and 'continuous run' to be set with

WaveFormLib.FDwfAnalogOutRunSet(hdwf, channel, ctypes.c_double(0))WaveFormLib.FDwfAnalogOutEnableSet(hdwf, channel, 1) 
WaveFormLib.FDwfAnalogOutFunctionSet(hdwf, channel, dwfc.funcPlayPattern)
WaveFormLib.FDwfAnalogOutFrequencySet(hdwf, channel, ctypes.c_double(hzRate))
WaveFormLib.FDwfAnalogOutAmplitudeSet(hdwf, channel, ctypes.c_double(ampOutVoltageRange))
# WaveFormLib.FDwfAnalogOutRepeatSet(hdwf, channel, ctypes.c_double(0))
WaveFormLib.FDwfAnalogOutRunSet(hdwf, channel, ctypes.c_double(0))
WaveFormLib.FDwfAnalogOutDataSet(hdwf, channel, rgdPlay, nrSamples)
WaveFormLib.FDwfAnalogOutConfigure(hdwf, channel, 1)

but they both give the same 30ns per record drift. 

The way I measure this: 

1. Generate 100 kHz signal with external signal generator, view on ch1 of benchtop o'scope and connect to AD pro input 1
2. Connect AD Pro output 2 to ch2 or benchtop o'scope
3. Run RecordOnlyExample.py (generates .pkl file with recorded data)
4. Run PlaybackOnlyExample (plays back data from PlaybackOnlyExample.pyRecordOnlyExample.py.pkl file)
5. Watch the 2 signals on the o'scope drift w.r.t. each other.
6. Count the time for 1 period drift (~33 sec).
   1. 33 sec for at a record length of 100 ms = 330 'drift events'
   2. 1 period of 100 kHz signal == 10 usec
   3. 10 usec / 330 events = 30 ns. 

Thank you for your input.