Performance of Picoscope 4262 does not reach expectation

Post any questions you may have about our current range of oscilloscopes
Post Reply
richiechen
Newbie
Posts: 0
Joined: Mon Nov 05, 2012 12:59 pm

Performance of Picoscope 4262 does not reach expectation

Post by richiechen »

We just bought a Pico 4262 because of this pic:
pico4262.jpg
It demonstrates a noise floor of -127dB, with its own AWG generating sine waves.

When we got the scope and tested in the exact same way, here comes the result:
Software version: Pico 6 beta 6.11.4.929
raven_real.png
The comparisons are as following:
1. Website: A noise floor of -127dBu
Real: A noise floor of -110dBu
2. Website: 10k over highest harmonic distortion: 103dB
Real: 10k over highest harmonic distortion: 98dB

Could anyone explain the reasons to me?
1. Is it because I need a properer signal wire? (When I disconnect the channel from anything, the noise floor is still around -110dBu, so probably it is not the problem of signal wire.)
2. Is it the problem of the scope that I get?
3. Is it because the software version I am using?

Thank you very much!

Regards,
Richie

alan
Site Admin
Site Admin
Posts: 195
Joined: Thu May 25, 2006 8:02 am

Re: Performance of Picoscope 4262 does not reach expectation

Post by alan »

Hi,

Try increasing the number of points in the FFT (in spectrum options with the drop down arrow). Alternatively change the display mode (also in options) to averaged view. Both of these will reduce the noise floor.

As for the SFDR it may be the cable - I have seen 5 to 10dB difference when looking at signals in the 100dB down region. I notice your cable seems to have SMA connectors then adapts to BNC. Try a good quality short BNC cable.

I hope one / both these helps.

Regards
Alan

Audioguy
Active User
Active User
Posts: 22
Joined: Thu Aug 09, 2007 9:34 pm

Re: Performance of Picoscope 4262 does not reach expectation

Post by Audioguy »

Regarding performance expectations of 4262, are there any known aliasing problems with this scope?

I can not get clean results above approx 1MHz no matter if linear or sin(x)/x interpolation is used. This does not go well with the advertised bandwith of 5MHz.

Feeding a clean sine (harmonics/spurious signals 50dB below fundamental) cause a modulated rendition on the screen, indicating aliasing. What you see on the screen is modulation between full level and zero. Think what a waveofrm look like if you mix two sines with same level which are close to eachother in frequency.

I can understand this is happening if feeding a sine above at/above 5MHz (in single channel mode) but not below.

Since the advertised bandwith is 5MHz I assume it is possible to capture a sine wave up to 5MHz.

Can the Pico team please see if this is possible and if so, which settings that are necessary to get a correct reading of a pure sine input up to say at least 4MHz?

Thanks!

Gerry
PICO STAFF
PICO STAFF
Posts: 1145
Joined: Mon Aug 11, 2014 11:14 am

Re: Performance of Picoscope 4262 does not reach expectation

Post by Gerry »

Hi Audioguy,

We are not aware of any known aliasing issues with the PicoScope 4262, and we can get clean results (without any significant aliasing) right up to 5MHz signals.

The transfer of data (or lack of) between the different views depending upon whether they are split, switched, and which one is the original view can sometimes have misleading effects. It would be worthwhile having a look at how you have things setup.

Could you tell us what version of Software you are using, and what you signal source is. Could you also post your psdata file so that we can have a look at what is going on.

Regards,

Gerry
Gerry
Technical Specialist

Audioguy
Active User
Active User
Posts: 22
Joined: Thu Aug 09, 2007 9:34 pm

Re: Performance of Picoscope 4262 does not reach expectation

Post by Audioguy »

Hi!

Please see attached files. One FFT to show quality of signal generator (Agilent DSOX-3032) stimuli at 1MHz (=low distortion) and three scope views of three different sines below 5MHz.

It seems like what we can see is just what you would see on the output of a poorly filtered DAC giving mirror signals above the nyquist frequency. Sin(x)/x interpolation makes a difference but only minor.

Sampling 4.9MHz at 5MS/s will give a mirror signal at 5.1MHz on a DAC output if poorly filtered. These two sines will modulate eachother with a frequency/repitition of 200kHz.

My set up is fine at 1MHz but already at 2MHz the waveform starts to become jumpy in the time domain as seen on the screen.

Pico version 6.10.16.2.

The 4262 is a unit from 2014.
Attachments
20160418-0004.psdata
(21.45 KiB) Downloaded 545 times
20160418-0003.psdata
(73.6 KiB) Downloaded 541 times
20160418-0002.psdata
(19.4 KiB) Downloaded 514 times
20160418-0001.psdata
(21.16 KiB) Downloaded 556 times

Gerry
PICO STAFF
PICO STAFF
Posts: 1145
Joined: Mon Aug 11, 2014 11:14 am

Re: Performance of Picoscope 4262 does not reach expectation

Post by Gerry »

Hi Audio Guy,

Thanks for posting the data files. I can see what you are referring to now.

As I'm sure you are aware, the signal bandwidth specification applies to the frequency domain (not the time domain) the bandwidth maximum being the 3dB point on a frequency plot. That point for the PicoScope 4262 is at 5Mhz and, with reference to our product, this is an indication of the performance of the input stage.

Once again, utilizing a concept you already know, we need to ensure that there are no frequency components that are equal to or greater than the Nyquist frequency to avoid aliasing. The maximum sample rate of the PicoScope 4262 is 10MS/s, so no pure sine-wave with a frequency of less than 5MHz is going to produce alias components. Aliasing manifests itself as a new beat frequency, where too few sample points actually look like a different frequency waveform.

What you are seeing in your first posted data file is a modulated envelope, which is not an artifact of aliasing (the frequency of the signal is still the same, and not being replaced by an alias). The modulation is actually common to most digital waveform recorders and oscilloscopes and is due to their inherent reconstruction of the data for display. The effect is minimized at signal frequencies that are at most 1/5 of the sample rate, which is a recommendation we make on our website for scope choice and signal bandwidth.

However, as the amplitude of signal harmonics diminish with increasing number, the modulation effect diminishes with increasing harmonic. For some usage cases of the applications for which the PicoScope 4262 was intended, the ability to display higher frequency pulses (e.g. 250kHz freq, 500nS width), and square waves (e.g. 1MHz) that have harmonics beyond 1/5 the sample rate (albeit modulating slightly) is still a useful feature.

I hope this clarifies things for you.

Regards,

Gerry
Gerry
Technical Specialist

Audioguy
Active User
Active User
Posts: 22
Joined: Thu Aug 09, 2007 9:34 pm

Re: Performance of Picoscope 4262 does not reach expectation

Post by Audioguy »

Thanks Gerry!

Sorry for bad wording regarding aliasing and "imaging". I do realize the difference between sampling signals above Nyquist that folds down below Nyquist (aka aliasing) and sampling signals below Nyquist but trying to reproduce them without a lowpass reconstruction filter (aka "brickwall-filter") which will casue imaging signals above Nyquist that was not there in the input.

I also realize that a sharp lowpass would give worse display of transient signals.

I guess I was expecting the sin(x)/x interpolation mode to use a brickwall lowpass filter similar to what you would find in an audio converter. Also an analog scope will be able to display a steady sine at the -3dB point in contrast to what we see in this case.

I understand now that for measuring sine waves properly between 2-5MHz you must use the FFT in Picoscope.

Using the math function in scope mode would work as well but I'm not sure how freely you can set filter characteristics.

Using SDK or exporting data to Excel or Matlab would be other options.

I don't necessarily need to have more than 1MHz "proper" bandwith in this scope but was pussled by what was going on. A friend of mine that recently started to use Pico products noticed the same thing and was surprised.

I also see that you have started to use more "oversampling" in the latest products. Having a bandwith spec significantly below Fs/2.

Having audio gear and analog scopes as a reference I guess it's easy to be confused by this.. althoug in hindsight it seems obvious. :-)

Thanks!

Post Reply