Products > Test Equipment
Is Rigol DS1054Z still a good option in 2019?
Mr. Scram:
--- Quote from: Fungus on August 01, 2019, 09:33:00 am ---a) "Display is sluggish" is a lie, the display is just fine.
b) It's true that the vertical position control (and only the vertical position control) is a bit sluggish but in daily use it turns out to be a superficial, "First world" problem that doesn't affect work at all.
--- End quote ---
The fun part is that almost all oscilloscopes are sluggish. The Keysights seem to be the only real exception due to doing most things in hardware.
bdunham7:
--- Quote from: Fungus on August 01, 2019, 09:21:12 am ---
--- Quote from: bdunham7 on August 01, 2019, 01:21:00 am ---As for the DS1054Z vs the SDS1104X-E, I have them both. They both work fine, especially at their price points, but neither is perfect.
--- End quote ---
This.
Dismiss any idea that the Siglent is "perfect", it has issues too.
--- Quote from: bdunham7 on August 01, 2019, 01:21:00 am ---The only quirk I worry about is that the Rigol has some issues about what it is actually displaying and how it is interpolating that I don't understand--this has been discussed elsewhere.
--- End quote ---
It's not difficult to understand it's just that the Rigol-hater-club that inhabits this forum tries to exaggerate it beyond all reason and the details get lost.
Some oscilloscopes have the option to turn off sin(x)/x signal reconstruction. Not all of them do (most expensive ones don't!) because it makes no sense to do so. It's a fundamental part of signal reconstruction and drawing the wiggly lines on screen.
The Rigol does have it but it only appears when you turn on more than two channels enabled and it's sampling at 250Mhz. It's greyed out at all other times because the sample rate isn't anywhere near the Nyquist limit so sin(x)/x is the correct thing to do.
What the Rigol-hater club discovered is that when you turn sin(x)/x "Off", it doesn't really turn off. Instead it changes to a "Rigol interpolation".
We don't know the exact math of this interpolation, hence the FUD. The only 'issue' is that it's not what the OCD types think "sin(x)/x turned off" ought to be.
(note: Nowhere is it written what "sin(x)/x turned off" should be, it's undefined, it's an error condition...)
OTOH we can see that Rigols' mystery interpolation reduces the Gibbs Phenomenon when you're on the limit, ie. it's better than sin(x)/x when you're on the Nyquist limit.
--- End quote ---
Ease up on the "Rigol-hater" rhetoric. The issues are more complex than that. Interpolation is involved with triggering, as well as drawing your squiggly line. IMO, the Rigol does a really good job of generating a stable trigger overall--better in fact than any of my analog scopes except the 2465B, and excepting low signal levels where some of the analog scopes trigger cleanly on much lower signals.
However, the issue I have, and which has been discussed endlessly elsewhere, is that the 'DOTS' mode displays dots which are not necessarily data samples. The whole point of wanting to eliminate interpolation is that sometimes you want to see the raw data that the scope is receiving so you have a better idea of what it actually knows and what it is 'reconstructing'. This is very important because the reconstruction theory depends on certain conditions being met--bandwidth limitations and so on--and sometimes there isn't an actual guarantee that they ARE met. I've seen some pretty bizarre results from this on and not just on contrived corner-case examples. I think it is a limitation, and more importantly, it isn't an obviously apparent one. Sometimes the dots are sample points--and sometimes they aren't. I'm not sure what it is doing and I haven't seen a clear explanation yet.
None of this would stop me from buying the scope again if I needed to stay at that price point.
Fungus:
--- Quote from: bdunham7 on August 01, 2019, 03:17:37 pm ---The whole point of wanting to eliminate interpolation is that sometimes you want to see the raw data that the scope is receiving so you have a better idea of what it actually knows and what it is 'reconstructing'.
--- End quote ---
I'm not sure how looking at dots tells you anything though. It's like looking at the pixels in a zoomed image and telling me you know the detail in between them.
The correct thing to do would be turn off other channels and get 500MHz or 1GHz sample rate on the channel you're interested in. That way you move away from the Nyquist limit and see the true signal.
--- Quote from: bdunham7 on August 01, 2019, 03:17:37 pm ---This is very important because the reconstruction theory depends on certain conditions being met--bandwidth limitations and so on--and sometimes there isn't an actual guarantee that they ARE met.
--- End quote ---
It's perfectly possible that they aren't being met on a Rigol DS100Z. It has about 130MHz bandwidth and the sample rate drops to 250MHz when you turn all channels on.
The Siglent SDS1204 has the exact same problem, it has about 280Mhz measured bandwidth and sample rate can drop to 500Mhz with all channels on - not enough to garantee the conditions required by signal theory.
To beat this you have to go up in price, eg. The Rigol MSO5000 has 350MHz bandwidth and at least 2GHz sample rate per channel. It's never going to get anywhere close to Nyquist with that ratio.
bdunham7:
--- Quote from: Fungus on August 01, 2019, 05:31:52 pm ---
--- Quote from: bdunham7 on August 01, 2019, 03:17:37 pm ---The whole point of wanting to eliminate interpolation is that sometimes you want to see the raw data that the scope is receiving so you have a better idea of what it actually knows and what it is 'reconstructing'.
--- End quote ---
I'm not sure how looking at dots tells you anything though. It's like looking at the pixels in a zoomed image and telling me you know the detail in between them.
The correct thing to do would be turn off other channels and get 500MHz or 1GHz sample rate on the channel you're interested in. That way you move away from the Nyquist limit and see the true signal.
--- Quote from: bdunham7 on August 01, 2019, 03:17:37 pm ---This is very important because the reconstruction theory depends on certain conditions being met--bandwidth limitations and so on--and sometimes there isn't an actual guarantee that they ARE met.
--- End quote ---
It's perfectly possible that they aren't being met on a Rigol DS100Z. It has about 130MHz bandwidth and the sample rate drops to 250MHz when you turn all channels on.
The Siglent SDS1204 has the exact same problem, it has about 280Mhz measured bandwidth and sample rate can drop to 500Mhz with all channels on - not enough to garantee the conditions required by signal theory.
To beat this you have to go up in price, eg. The Rigol MSO5000 has 350MHz bandwidth and at least 2GHz sample rate per channel. It's never going to get anywhere close to Nyquist with that ratio.
--- End quote ---
No, I think we're on two different pages here. First, just to clarify, the "measured bandwidth" refers to approximately the 3db point. The 1054Z will cleanly trigger on and display a signal out to 400MHz, which is near what you call the "Nyquist limit", but that is when it is fed a clean 400MHz sine signal. Nyquist, however, requires that all of the input signal above the limit be attenuated below the noise level or it will show up folded back in (aliased). You can demonstrate this doesn't happen with the 1054Z by using 500MHz, 600MHz and 700MHz signals--try it if you can.
I'm not referring to the sampling/bandwidth issue directly, I'm referring to interpolation sometimes being implausible. How do I know what is between the dots? I don't--and this is my point. I want to see the samples displayed directly with no interpolation or interpretation so that I understand what information the scope is working with. With complex or transient signals, the interpolation is just defective guesswork that I don't trust--and that is because I've seen it be wrong.
rsjsouza:
In my opinion, the Siglent and the Instek are the best bang for buck and have a longer roadmap of updates as they are newer platforms, with an edge for the Siglent as the manufacturer actually supports and advertises the protocol decoding. The Instek can have these enabled via hacking, but the manufacturer does not have them as an option, which increases risk as they may simply vanish in a future update. You don't seem too worried about this, thus you should be fine with either.
The independent controls for each channel are nice, but the shared controls can be learned and one develops some mechanical memory over time. I personally prefer the independent controls, but YMMV.
The discussions above on very specific details about each one's preferred brand are common here. The same regular actors tend to turn each question into a battlefield of ideas, preferences, assumptions and guesses... Don't feel bad feeling your question started a war - it always happens around here! :P
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version