Test and Measurement Forum

Resolution enhancement and amplitude problem

Forum for discussing PicoScope version 6 (non-automotive version)

Resolution enhancement and amplitude problem

Postby Ax2 » Sat Aug 30, 2014 12:44 pm

Hello,

Just bought PS2107 and performed a simple frequency sweep test (0 to 1MHz) using fixed voltage level. Using 8-bit mode, amplitude (Vpp) seems to be within tolerance. After increasing bits especially up to 10-12 weird things started to happen. Amplitude varies from 10-100% depending on frequency.
It is definitely a drawback if Resolution enhancement is not reliable!For me correct amplitude is essential.

Ax
Ax2
Newbie
 
Posts: 0
Joined: Tue Aug 26, 2014 5:32 pm

Re: Resolution enhancement and amplitude problem

Postby Martyn » Mon Sep 01, 2014 6:16 pm

Resolution enhancement is achieved using a moving average filter, the more bits you request the more samples are used in the average. If you check the PicoScope help file, available from the Help menu, it will explain this in more detail. Higher frequency signals will be affected by this process so what you see would be expected.
Martyn
Technical Specialist
Martyn
Site Admin
Site Admin
 
Posts: 2350
Joined: Fri Jun 10, 2011 8:15 am
Location: St. Neots

Re: Resolution enhancement and amplitude problem

Postby Ax2_ » Mon Sep 01, 2014 7:22 pm

Hello. Thanks for the reply. When you say "higher", what is the top limit of frequency Resolution when enhancement is still reliable?

Ax
Ax2_
Newbie
 
Posts: 0
Joined: Mon Sep 01, 2014 6:23 am

Re: Resolution enhancement and amplitude problem

Postby Martyn » Thu Sep 11, 2014 9:09 am

It will depend upon the sample rate you have selected, and the number of bits of resolution enhancement.

Each additional 0.5 bits will double the number of samples used in the moving average.

0.5 uses 2 samples
1.0 uses 4 samples
...
3.5 uses 128 samples
4.0 uses 256 samples
Martyn
Technical Specialist
Martyn
Site Admin
Site Admin
 
Posts: 2350
Joined: Fri Jun 10, 2011 8:15 am
Location: St. Neots

Re: Resolution enhancement and amplitude problem

Postby Ax2_ » Thu Sep 11, 2014 8:16 pm

Well, I don't think you answered to my question. The situation was very simple at the beginning: I tried to use Resolution enhancement and made a simple sweep test. After the frequency started to increase, amplitude started to variate between 10% and 100%. It can't be proper behavior! Or if it is, I really can't use resolution enhancement for anything useful. And... I don't think anyone can.
Ax2_
Newbie
 
Posts: 0
Joined: Mon Sep 01, 2014 6:23 am

Re: Resolution enhancement and amplitude problem

Postby Martyn » Fri Sep 12, 2014 6:15 am

Resolution Enhancement is a software technique to remove the Gaussian noise which is often present in electrical signals, the downside is that it can also remove the higher frequency components of the actual signal. The more high frequency components in the signal, and the more bits of resolution enhancement used, the result will be a drop off in the amplitude of the displayed signal. Remember this is just the displayed data, the raw data will be still be true. For this reason we will generally recommend that you always capture the raw data, and use Resolution Enhancement post capture just to see the signal cleaned up a bit.

It does have it's use in cleaning up noisy signals, particularly lower frequency ones, but you have to understand the possible downsides. The effect you get is very similar to inputting a high frequency square wave into a device whose bandwidth is less than 5 times the frequency of the signal, the edges become rounded and the amplitude reduces.

If you require true 8, 10,12 or even 16 bit data then you would need to select a scope model that supports this, however the higher the vertical resolution, the faster the sampling rate, the greater the cost.
Martyn
Technical Specialist
Martyn
Site Admin
Site Admin
 
Posts: 2350
Joined: Fri Jun 10, 2011 8:15 am
Location: St. Neots

Re: Resolution enhancement and amplitude problem

Postby Ax2_ » Sun Sep 14, 2014 10:49 am

Well, you are talking about cleaning up noisy signal which affects to amplitude. You are true naturally, but still it is NOT the issue I'm, talking about.
Maybe you get the idea by examples.
First picture is 8-bit signal from an external signal generator 5 Vpp. PS measurement says 5.1 Vpp which is quite true with a small noise it has.
Second sample is the same signal with 10-bit Enhancement. Vpp is 5.0 which is exactly what is fed to PS and sounds like right after 2-bit Enhancement. So far so good.
Third sample is again the same signal with 11,5 number of bits. Amplitude is 2.2 Vpp!
Fourth sample has 12 bits and amplitude is 960mV Vpp.

Got it? We are not talking about rounded signal shapes nor cleaning up signal. We are talking about massive amplitude error when resolution bits are increased. There are some frequency ranges where this affect started even with 10-bits, but in this example 10 bits is actually very useful. Just that when the function is not reliable in general, I can't trust it at all.

Ax
Attachments
12-bit_01.jpg
11_5bit_01.jpg
10-bitb_01.jpg
8-bit_01.jpg
Ax2_
Newbie
 
Posts: 0
Joined: Mon Sep 01, 2014 6:23 am

Re: Resolution enhancement and amplitude problem

Postby Martyn » Sun Sep 14, 2014 11:54 am

For the signal, exactly as shown on the screen, can you tell me exactly how many samples make up the full screen, and therefore how many make up one cycle of the sine wave.
Martyn
Technical Specialist
Martyn
Site Admin
Site Admin
 
Posts: 2350
Joined: Fri Jun 10, 2011 8:15 am
Location: St. Neots

Re: Resolution enhancement and amplitude problem

Postby Ax2_ » Sun Sep 14, 2014 3:04 pm

I think it is better if you specify what debug data you need to investigate this issue. I don't know if I understand your question right but according to the properties window no. of samples is 2000. (Sample rate 1GS/s). "Number of samples" setting is 100kS, but it does not influence on this issue.

This was one of the first tests I ran with the device and you can easily repeat it by yourself: just feed 5MHz (or 6 or whatever) pure sine signal from an external signal generator and change no. of bits. Above 10 weird things start to happen. 11.5 and 12 are the most affected parameters.

Ax
Ax2_
Newbie
 
Posts: 0
Joined: Mon Sep 01, 2014 6:23 am

Re: Resolution enhancement and amplitude problem

Postby Martyn » Sun Sep 14, 2014 3:50 pm

With 2000 samples across the screen it means that each cycle of the signal is covered by just 200 samples, there are 10 complete cycles shown. When you are applying increasing levels of resolution enhancement, you are getting closer to averaging all the samples of one cycle of the wave, that is why the amplitude drops. If you slow the frequency so that you only have 1 or 2 cycles covered by the 2000 samples you see less of a drop in amplitude.
Martyn
Technical Specialist
Martyn
Site Admin
Site Admin
 
Posts: 2350
Joined: Fri Jun 10, 2011 8:15 am
Location: St. Neots

Re: Resolution enhancement and amplitude problem

Postby Ax2_ » Sun Sep 14, 2014 4:04 pm

Well... you are giving me an explanation why this happens. But for me this is a failure situation. If I feed a pure 5 Vpp sine signal to PS and on screen I see less than 1 Vpp signal, it is a failure.
I changed collection time to show only 5 waves but No. Samples is also reduced to 1000. Amplitude therefore remains the same. And.... wrong signal level remains on screen.
So, please tell me what parameters should be used to get correct result on screen.

It is true that if I reduce frequency this behavior is not visible. Just that I thought I bought a 100MHz scope, not 100kHz bw.

Please consider this as a bug: I'm not buying an explanation why 5 Vpp shows 1 Vpp on scope screen. This is a measurement device after all.

Ax
Ax2_
Newbie
 
Posts: 0
Joined: Mon Sep 01, 2014 6:23 am

Re: Resolution enhancement and amplitude problem

Postby Martyn » Mon Sep 15, 2014 8:25 am

Ax2_ wrote:Please consider this as a bug: I'm not buying an explanation why 5 Vpp shows 1 Vpp on scope screen. This is a measurement device after all.

This is not a bug, you are at the limits of the hardware, and are then applying software techniques to the collected signal that are not suitable.

The 2207 is an 8 bit scope and will acquire signals up to 100MHz quite happily, and if you turn off Resolution Enhancement you will see this is true.

Resolution Enhancement is a feature of the software, it is used to remove Gaussian Noise, however to work it needs an appropriate number of samples across the trace. As you are running the scope at it's fastest sampling interval, of 1ns, and therefore the number of samples across the screen is limited to 200 per cycle of the input sine wave, using Resolution Enhancement is not suitable so you should turn it off.

From the PicoScope software help file
Resolution enhancement is a technique for increasing the effective vertical resolution of the scope at the expense of high-frequency detail. In some scope operating modes PicoScope may reduce the number of samples available to maintain display performance.

For this technique to work, the signal must contain a very small amount of Gaussian noise, but for many practical applications this is generally supplied by the scope itself and the noise inherent in normal signals.

The resolution enhancement feature uses a flat moving-average filter. This acts as a low-pass filter with good step response characteristics and a very slow roll-off from the pass-band to the stop-band.

Some side-effects will be observed when using resolution enhancement. These are normal and can be counteracted by reducing the amount of enhancement used, increasing the number of samples captured or changing the timebase. Trial and error is usually the best way to find the optimum resolution enhancement for your application. The side-effects include:

Widened and flattened impulses (spikes)

Vertical edges (such as those of square waves) turned into straight-line slopes

Inversion of the signal (sometimes making it look as if the trigger point is on the wrong edge)

A flat line (when there are not enough samples in the waveform)


If you require a greater vertical accuracy then you would need to look at a scope that can handle 10 or 12 bits in the hardware such as the 5000 series.

As you are unhappy with the scope for your intended purpose, please contact me via support@picotech.com and we can discuss your options for a refund or upgrade.
Martyn
Technical Specialist
Martyn
Site Admin
Site Admin
 
Posts: 2350
Joined: Fri Jun 10, 2011 8:15 am
Location: St. Neots


Return to PicoScope 6 for Windows

Who is online

Users browsing this forum: No registered users and 0 guests