Oscilloscope Alias Followup
ฝัง
- เผยแพร่เมื่อ 30 มิ.ย. 2024
- A follow-up to this video comparing scopes anti-aliasing.
• Oscilloscope Anti Alias?
Support the EEVblog on:
Patreon: / eevblog
Odysee: odysee.com/@eevblog:7
EEVblog Web Site: www.eevblog.com
Main Channel: / eevblog
EEVdiscover: / eevdiscover
AliExpress Affiliate: s.click.aliexpress.com/e/c2LRpe8g
Buy anything through that link and Dave gets a commission at no cost to you.
Donate With Bitcoin & Other Crypto Currencies!
www.eevblog.com/crypto-currency/
T-Shirts: teespring.com/stores/eevblog
#ElectronicsCreators #oscilloscope - วิทยาศาสตร์และเทคโนโลยี
![[Live สด] การออกรางวัลสลากกินแบ่งรัฐบาล งวดวันที่ 1 กรกฎาคม 2567](http://i.ytimg.com/vi/o6JPYXalcXU/mqdefault.jpg)
The "oddball" frequencies are also aliasing - look at the frequency display which is still wrong. It's just that the signal/sample alias frequency and the display subsampling are beating at a frequency above the Nyquist limit.
The “measured” frequency will let you know how accurate your scope’s timebase is. If you show a 2 Hz signal with 10MHz input, you’ve only got a 0.2ppm error between your sig gen and scope.
I remember the first time I started to use my brand new oscilloscope and function generator.
I had the generator set up to output a sine wave of a 10 MHz, but the oscilloscope showed a very low frequency (in the order of Hz) with a timebase of 100ms per division.
The hardware frequency counter was on and it showed 10MHz.
I was going crazy. Everything was telling me the frequency was 10MHz except the oscilloscope screen, when I changed time base the signal started going crazy and produce a mess on the screen. Spent like a couple of hours trying to identify the problem. I was sure my signal generator was broken.
That day I learned about aliasing.
Just like the early Tek high bandwidth scopes did the display, using repetitive sampling and aliasing to get a signal that would actually pass through the CRT deflection amplifiers. That way you could get a CRT that possibly went to 100MHz in real time to sample and display signals in the GHz range.
This is sort of true, but keep in mind that if you were looking at signals above the actual bandwidth of the scope, any vertical measurements would be inaccurate because higher frequency components are attenuated. Even the cheapest USB scopes at least have a basic RC low pass filter on the front end.
But yeah you can measure the frequency of signals quite a bit beyond the bandwidth of the scope. Daniel Bogdanoff did a video on this a few years back - th-cam.com/video/S8eSDjyRceg/w-d-xo.html
I think there's an interesting interaction between memory depth and acquisition mode on these sorts of scope. If you run 'normal' acquisition and a small memory size, you can get aliasing in situations where you don't with large memory depths. With large depths the sampling is running at a higher rate, and the display is giving you a view based on all of the captured data, so if you, say, have 10M data points and only a 1000 pixel wide screen, 10k samples are being displayed per line, which leads to the nicer DPO style view.
If you turn on peak acquisition (that's what it's called on my Siglent) then again, I think the sampling happens at a faster rate, but rather than storing all of the data in memory, it just stores the aggregate data from the multiple samples, so at smaller buffer sizes you still get a DPO style view of a fast signal without aliasing. If you turn the buffer up larger however, it doesn't have as much data to aggregate per peak value, and some odd stuff re-appears.
So, either large buffers normal acquisition or small buffers, peak acquisition seem to give me the best results in Siglent land.
Definitely need to explore how those two interact to fully understand aliasing in digital systems and what you can do about it
The reason that peak detect breaks up the aliasing effect is because it is picking the samples in in each interval based on their value, not always picking them at regular intervals, and this small amount of irregularity in sampling is enough to avoid synchronous aliasing.
The Fluke 190 series II and III has a glitch mode (is default enabled) where the incoming signal is sampled (and saved) not only on the sample frequency, but also on a high clock frequency, just to avoid these kinds of problems. The minimum and maximum samples seen during the sample time are also stored (and displayed).
Hello, thank you for your good video
Yes, this problem exists in the oscilloscope.
To solve this problem, I choose the memory depth of 50 and the problem is solved.
Oscilloscope 1000 also has this problem (when HiRes mod is ON)
When don't have enough memory depth for the frequency vs timespan you can change the adquire mode to peak to let the osc sample at full rate.
At least you'll get a proper envelope. Noisy but proper..
Is it possible that with the odd frequencies the function generator is actually not producing a pure sine and instead has some small phase error (possibly dithering) which is enough randomness to prevent the aliasing? To get what looks like a pure sine with aliasing requires high spectral purity (less than about 1 in a million when dealing with a pure frequency in the MHz).
Nyquist...
Al sampling devices have this feature.
Read up on your sampling theory.
Is there aliasing (from the screen resolution) on top of aliasing (from the sampling rate) happening?
Hi Dave, do we have a livestream today?
After using the Siglent and Rigol scopes for a few years now, do you prefer one brand over the other?
I would have thought the frequency counter and trigger frequency getting close or a harmonic would give you a warning at the least or prevent aliasing by some other means.....
Sticking to my ANALOG scope that doesn't have these problems and nearly instant display of signals.
To be fair, this is happens because a given sample can only store 1 value. It simply captures a single sample in time.
The scope could instead continue to sample at the full 2 GS/s instead, and statistically crunch it down into a "5 MS/s" sample rate for it to achieve the desired memory depth. But still contain the average, max and min values for that "sample". (The sample would be a lossy compression of 400 samples.) This however means that we need to store 3 values per sample, instead of just one, so it would lead to 1/3rd the memory depth. Technically one could store even more statistics if desired, like RMS values, duty cycle, etc, and if it is selectable then the scope could get some rather powerful long term statistics gathering capabilities. Simply pick and choose who it stores this "sample", and how many types of data it shall contain. (I should quickly add here that no scope to my knowledge actually does this.)
Have you tried your old Rohde Schwarz HMO scope, mine behaves slightly different ...
I'm getting an MXO4 next week.
The greyscale values. looks weirdly low..
what levels does these HDO4k and 1k series have?
hopefully not only 56 levels?
Interesting
On the bright side, with this you can get a very detailed view of a waveform even though it's way above the scope's sampling rate.
No, actually you most likely can't. The analog bandwidth will limit what your scope can display.
I can't reproduce this on my analog scope! 🤪😂
Bloody Nyquist! Will bite you every time! 😉😁
Not that fund of the scope design of HDO4K and 1K and those hard plasticknobs got a Fnirsi/Hantek vibe over them..
The Siglent SDS2000x series do look better ..those neoncolors / housingsticker aroudn the sockets, also dont ring my bell.
and that dotted part, is that indent a sticker or actual casing, nahhI reckon its actual casingmold as it would be to tacky if it was simply a sticker that was used to coverup a brief plane difference to the indent around the USB sockets...
Though the screen UI do look decent and somewhat modern.
You are comparing it to a product that costs 3 times as much. The low end Siglents also use hard plastic knobs.
@@jaro6985 you clearly dont know what HDO4K refers too.?
th-cam.com/video/6qjqhnQiQXQ/w-d-xo.html
and no, its not a good locking scope as per se..
a SDS2104X (4ch, 100) 1399 US (Siglentna) does look quite a bit better then a Rigol HDO1104 (4ch 100) 1399 (rigolna) on the pure outer design of the scope.
- as mentioned above. the screen UI do look somewhat modern & decent.
but nothing new.. Rigol leaves a lot to be desired when it comes to their scope design.
You almost need to go up around their flagship DS70000-series before its a sexy beast with knobs that looks like its with duallayered inlays for fingergrip, and highlighted backlid-knobs and all that jazz.
Seems Bogus...machine automatically drops the sample frequency but doesn´t LPF the input to 1/2 Nyquist? Gahhh!
Analogue scope superiority ?
A trap for young players if I ever saw one!
WTF 0_0
And you pay how many dollars for these scopes..... Not good enough
wtf #69 Like Again thats 10 this month no no NO i,m not touching it :)