Aliasing problem with 3204 and picoscope 6.2.20?

Post any questions you may have about our current range of oscilloscopes
Post Reply
DG5MK
Newbie
Posts: 0
Joined: Tue Nov 17, 2009 9:09 am

Aliasing problem with 3204 and picoscope 6.2.20?

Post by DG5MK »

Dear folkes,

I am using a 3204 with 50 MHz bandwith and 50 MHz sampling rate for amateur radio purposes. Taking a look at a MHz square wave from a DDS generator gives very strange results. If the signal is not a multiple of the sampling frequency, I got large peaks below and above the base frequency in spectrum analyzer view. At 5 MHz and 10 MHz the peaks disappear and the spectrum looks ok. The same pictures from another DDS.

This seems to be an effect of aliasing. The square wave contains harmonics at 3rd, 5th base frequency and so on. But I was not expecting such a big impact on waveforms and spectrums shown in Picoscope. The view is nearly unusable, for a 3 MHz square wave etc.

Now my questions: Is this normal and do others have the same experience? Will it make sense to put in (generally) an anti aliasing filter (lowpass at 25 MHz) in front of the scope? Is there any built in anti aliasing in the scope itself (like it is in PC soundcards etc.)?

Thanks for your help
Michael

Guest

Re: Aliasing problem with 3204 and picoscope 6.2.20?

Post by Guest »

It is aliasing and it is normal because the PicoScope does not have an anti-aliasing filter other than the bandwidth limit. Generally the analog bandwidths are equal to the sampling frequency so they let in a considerable amount of signal information that cannot be properly represented. In cases where you find this confusing or unacceptable you should indeed put an anti-aliasing filter on the scope. I don't know how difficult that might be.

I use mine for amateur radio too and I find the lack of an anti-aliasing filter to be beneficial most of the time. If my 3206 had a filter it would cut off at around 100 MHz. That would prevent me from looking at the close in spectrum of 2 meter signals, for example. Without an anti-aliasing filter I can look at the IMD on a 2 meter PSK-31 signal very nicely by setting the spectrum analyzer view to a few kHz BW. The 2 meter signal aliases down to the displayed BW and it has enough resolution to allow me to see the sidebands and measure the IMD quite nicely.

I can't recall the last time I had to look at the spectrum of a square wave for amateur radio purposes. Most of the signals I examine for amateur radio and other pursuits have low enough harmonic content to make aliasing a minor problem at most.

The PicoScope has been a great little toy for me. A few days ago I used it to measure the power curve on a bicycle trainer. I struggled for a long time to come up with a way to measure this with equipment I had on hand. Then it occurred to me that I could calculate the rotational inertia of the flywheel and by measuring the rate it which it decelerated after winding it up as fast as I could pedal it I could derive the power curve. The flywheel has six holes in it for cooling purposes so all I had to do was shine a flashlight on one side and use an LED clamped in a PicoScope probe on the other side to make a serviceable and accurate tachometer. The PicoScope recorded the data too and when I had a recording of a run I could export it to Excel and do all the post processing there.

Mr. Science would be so proud of me!

Ken

Erik
PICO STAFF
PICO STAFF
Posts: 53
Joined: Mon Oct 23, 2006 8:16 am
Location: Cambridgeshire, UK

Re: Aliasing problem with 3204 and picoscope 6.2.20?

Post by Erik »

You might find that it is the DDS that is giving you the results.

If the sampling rate of the DDS is a multiple of 5 and 10MHz it can do a perfect squarewave for those frequencies. Any other frequencies (not an integer divider of half the dds sampling frequency) close to the samplingrate will have a up to a sample of cycle-cycle jitter on the edges.

DG5MK
Newbie
Posts: 0
Joined: Tue Nov 17, 2009 9:09 am

Re: Aliasing problem with 3204 and picoscope 6.2.20?

Post by DG5MK »

Ken, Erik,
thanks for your reply. I have checked it again with a square wave from a VCO type oscillator (LTC1799). Exactly the same picture.
Yes, it is aliasing. The aliased signal part after A/D conversion maps back to the frequencies below the sample frequency. If the mapped back harmonics of the square wave fit exactly to the original base frequency, the spectrum view shows only the 5 MHz (or 10 MHz etc.). Of course the level of the signals shown is wrong as it contains a part of the aliased harmonics...
Where to use it in Ham Radio?
I am doing some work with Software Defined Radio (SDR) and to produce some orthogonal signals (I/Q) the input signal will be mixed with a square wave, because you can easily produce the required 90 degrees phase shift with square waves and some TTL logic.
What to learn?
1. Take care about what you measure, what you expect and what you really get. It could be very different
2. Digital scopes need to be handled more carefully than analog ones, because the analog to digital conversion can create some surprises.
Would I change back to an analog one or buy another one than a PicoScope?
No never and probably no. Digital scopes have so many big advantages compared to analog ones. And I like my PicoScope very much. Good value for money and fantastic SW!
Thanks again for your support and good luck with cycling...
Michael

Post Reply