New Rigol DS1054Z oscilloscope

[Bzzz]

They also have a nice bundle of the DS1054Z with a hakko FX888D for 374€ which is a good bargain.
http://www.batterfly.com/shop/oscilloscopi/70MHz/bundle-ds1054z-fx888d

Excluding VAT, unfortunately

[michalK]

I’m considering buying DS1054Z but because I’m “young player” in electronics I would like to ask you guys about something.

That scope has 50 MHz BW and 250 MSa/s on four channels. That gives exactly 5 times more sample rate than BW which is enough with sin(x)/x interpolation. But if we increase BW using that hack you’ve mentioned up to 100 MHz the situation changes and we get only 2.5 times the max BW. Can you tell me how that influences the readings and if that can cause aliasing? What is the theoretical limitation to avoid aliasing? How that issue is solved in 1104Z model if it has 100MHz and 250 MSa/s (4ch) as default?
Does it make sense at all to have scope with 100MHz and only 250 MSa/s?

[Fungus]

I’m considering buying DS1054Z but because I’m “young player” in electronics I would like to ask you guys about something.

That scope has 50 MHz BW and 250 MSa/s on four channels. That gives exactly 5 times more sample rate than BW which is enough with sin(x)/x interpolation. But if we increase BW using that hack you’ve mentioned up to 100 MHz the situation changes and we get only 2.5 times the max BW. Can you tell me how that influences the readings

Simple: It means you don't actually get 100MHz bandwidth.

(nb. This is also true on the DS1104Z with 4 channels enabled).

Does it make sense at all to have scope with 100MHz and only 250 MSa/s?

No.

But...it's a 4 channel scope and you can turn some channels off, eg. With only one channel enabled you have 1 GSa/s.

[marmad]

That scope has 50 MHz BW and 250 MSa/s on four channels. That gives exactly 5 times more sample rate than BW which is enough with sin(x)/x interpolation. But if we increase BW using that hack you’ve mentioned up to 100 MHz the situation changes and we get only 2.5 times the max BW. Can you tell me how that influences the readings and if that can cause aliasing? What is the theoretical limitation to avoid aliasing? How that issue is solved in 1104Z model if it has 100MHz and 250 MSa/s (4ch) as default?
Does it make sense at all to have scope with 100MHz and only 250 MSa/s?

It's not just the number of times the sample rate that you have to be concerned with - it's also the BW frequency response of the DSO. At 250MSa/s, the Nyquist frequency is 125MHz. That means, in order to use sin(x)/x interpolation reliably at that sample rate, all frequencies above that should be attenuated by at least -12db. On the DS1104Z, they are not. So when running with 3/4 channels, you should have interpolation set to LINEAR (unless using the built-in 20MHz BW limiter) - which, given the usual BW/10 linear interpolation formula, gives you a truly usable BW of 25MHz.

I'm not sure how the BW limiting is done for the 50/70MHz models. As Mark_O pointed out earlier in this thread, these models may / may not actually provide a higher reliable BW in 3/4 channel mode, depending on how the filtering is done.

[edavid]

Simple: It means you don't actually get 100MHz bandwidth.

I don't think that's an accurate way of describing the problem, since you do get 100MHz bandwidth.  The problem is that you may also get aliasing.  If you know the input signal is bandlimited (for example, it came out of a lowpass filter), you can use the full bandwidth.

[michalK]

Thank you both for your answers.

So to sum up (if I understood correctly):
1) 1104Z IS a real 100MHz scope even while using 4 channels
2) upgrading 1054Z to 100MHz in fact gives you capability to measure signals up to that frequency only on one (or two?) channels

Can anyone make some tests or show some examples what is the difference while measuring some 100MHz signals on one and four channels using upgraded 1045 model?

[marmad]

If you know the input signal is bandlimited (for example, it came out of a lowpass filter), you can use the full bandwidth.

Aside from decoding in the digital realm, when exactly would you be looking at 3 or 4 independent signals which you knew were all band-limited to 100MHz?

So to sum up (if I understood correctly):
1) 1104Z IS a real 100MHz scope even while using 4 channels

No, that's not the way I would describe it, but I guess it depends on what you mean by "real". IMO, unless decoding/examining digital signals of a known speed, the analog 'practical' BW of the DS1000Z when running 3/4 channels should be considered ~25MHz (unless the DS1054Z/DS1074Z have filtering in the gain-stage).

[Fungus]

So to sum up (if I understood correctly):
1) 1104Z IS a real 100MHz scope even while using 4 channels

Nope.

It's exactly the same 'scope as the "adjusted" (ahem) DS1054Z.

Can anyone make some tests or show some examples what is the difference while measuring some 100MHz signals on one and four channels using upgraded 1045 model?

There's some in the middle of this thread somewhere. The tests have been done. The results have been published (see the first part of this reply).

[alank2]

Isn't there a nice PDF which talks about bandwidth vs sample rate and Nyquist, etc.??

[pascal_sweden]

The remark about the 25 MHz BW is incorrect.

The scope has an analog BW of 100 MHz on ALL channels.

Note that analog BW and digital sample rate are two different things.

Analog BW = 100 MHz on ALL channels, even if you use all 4 channels at once.

Digital sample rate is 1 GS/s when you use 1 channel, 512 MS/s when you use 2 channels, 250 MS/s when you use 4 channels.

For a 100 MHz BW, 250 MS/s is enough to represent the original signal as it complies with the Nyquist theorem.

The Nyquist theorem states that a signal must be sampled at a rate greater than twice the highest frequency component of the signal to accurately reconstruct the waveform; otherwise, the high-frequency content will alias at a frequency inside the spectrum of interest.

More detailed information about the relation between Bandwith, Sample Rate and the Nyquist Theorem is available here:
http://www.ni.com/white-paper/2709/en/

[Fungus]

For a 100 MHz BW, 250 MS/s is enough to represent the original signal as it complies with the Nyquist theorem.

Technically true, but only useful to musicians, not electrical engineers.

[edavid]

If you know the input signal is bandlimited (for example, it came out of a lowpass filter), you can use the full bandwidth.

Aside from decoding in the digital realm, when exactly would you be looking at 3 or 4 independent signals which you knew were all band-limited to 100MHz?

Maybe if you were looking at the input and output of a radio transmitter.  (But even without an example, I think it's important to understand what the real limitation is.)

IMO, unless decoding/examining digital signals of a known speed, the analog BW of the DS1000Z when running 3/4 channels should be considered ~25MHz (unless the DS1054Z/DS1074Z have filtering in the gain-stage).

It's misleading to describe it that way, because people will think that you can't see signals > ~25MHz, which is not true.  If you said alias-free bandwidth, that would be clearer.

[edavid]

Can anyone make some tests or show some examples what is the difference while measuring some 100MHz signals on one and four channels using upgraded 1045 model?

There's some in the middle of this thread somewhere. The tests have been done. The results have been published (see the first part of this reply).

Has anyone applied a 150MHz signal in 4 channel mode and looked for aliasing?

[marmad]

Maybe if you were looking at the input and output of a radio transmitter.

That's 2 channels - not a limitation, in terms of the sampling rate - and so not applicable.

It's misleading to describe it that way, because people will think that you can't see signals > ~25MHz, which is not true.

Only people that don't understand what we're talking about. OTOH, using your logic, I could describe my Rigol DS2302 as being a 450MHz DSO, since I can see those signals.

EDIT: BTW, every time I've written about this, I've used terms like "a truly usable BW" or "a practical BW", etc. to indicate that it's not the actual BW of the DSO - but the 'workable' BW in many circumstances.

[BobCollins]

For a 100 MHz BW, 250 MS/s is enough to represent the original signal as it complies with the Nyquist theorem.

Technically true, but it's only useful to musicians, not electrical engineers.

No, electrical engineers, not musicians (assuming musicians can accept some distortion and call it music).

The point is Nyquist theorem only applies theoretically if there is no content above the sample rate /2. The real low pass filters used before the sampler are not perfect. If they were you could use bandwidth right up to the Nyquist point.

[edavid]

Maybe if you were looking at the input and output of a radio transmitter.

That's 2 channels - not a limitation, in terms of the sampling rate - and so not applicable.

The transmitter in my example has IQ inputs, and I also need to look at the LO 

It's misleading to describe it that way, because people will think that you can't see signals > ~25MHz, which is not true.

Only people that don't understand what we're talking about. OTOH, using your logic, I could describe my Rigol DS2302 as being a 450MHz DSO, since I can clearly see those signals.

What's the name of the logic fallacy that marmad is employing here?

EDIT: BTW, every time I've written about this, I've used terms like "a truly usable BW" or "a practical BW", etc. to indicate that it's not the actual BW of the DSO - but the 'workable' BW in many circumstances.

Above you called it "analog BW" 

[marmad]

It's misleading to describe it that way, because people will think that you can't see signals > ~25MHz, which is not true.

Only people that don't understand what we're talking about. OTOH, using your logic, I could describe my Rigol DS2302 as being a 450MHz DSO, since I can clearly see those signals.

What's the name of the logic fallacy that marmad is employing here?

There's no logical fallacy. My point was this: when a DSO is advertised as being an XX MHz, N channel DSO - that implies that it's BW frequency response is at least XX at -3db - and that it can sample at it's highest rate (without fear of aliasing) that XX MHz with all N channels ON. The DS1104Z can not - period. Since it does not adhere to these normal expectations, this would be no different than breaking one of the other implied specifications: e.g. that a 300MHz DSO has a 450MHz BW (even though that BW is at -9db).

It seems you want to tell newbies and other prospective buyers, "Sure, it does 100MHz with 3/4 channels on, as long as you make sure the input signal is band-limited", whereas I (and Mark_O and some other posters) are pointing out that this is not the realistic, working BW of the DSO with 3/4 channels on. Yes, under special circumstances you can achieve that BW  - but it's certainly not the normal, everyday usage of the DSO.

Above you called it "analog BW" 

OK, true - that is perhaps misleading and I will change it...

[alank2]

and that it can sample at it's highest rate (without fear of aliasing) that XX MHz with all N channels ON. The DS1104Z can not - period.

But with all 4 channels on, 250MSa/s, Nyquist is still 125 MHz, right?  Am I missing something?

[pascal_sweden]

Has anyone here actually proof that the DS1104Z can not do 100 MHz on all 4 channels at same time?

Or are the speculations just that it can not, because it is supposed to have no fancy anti-aliasing filter or whatever it might be called?

For me the analog frontend should just limit the max. frequency to make sure that there is no conflict with the corresponding ADC specs (sample rate) in the path beyond, and my understanding is that a 100 MHz scope has proper filtering by default to take care of this requirement in combination with > 200 MS/s ADC.

100 MHz scope == proper filter in the analog frontend to limit max. frequency to 100 MHz, period.

Isn't the anti-aliasing filter just a band pass filter? How complex can that be?

[Retep]

Has anyone here actually proof that the DS1104Z can not do 100 MHz on all 4 channels at same time?

The DS1104Z can do 100 MHz on all 4 channels at same time, however if the signal contains frequencies beyond 125 MHz you get aliasing when all 4 channels are enabled. 125 MHz is dangerously close to 100 MHz which means that if you really need 100MHz with all 4 channels enabled you might want to look further.

100 MHz scope == proper filter in the analog frontend to limit max. frequency to 100 MHz, period.

100 MHz scope means that you can expect that the -3dB point of the analog front-end is at least 100 Mhz, it does not say anything about the attenuation of higher frequencies. It is not possible to create an analog filter that passes all signals below 100 MHz with little or no attenuation and at the same time completely blocks any signal with frequency higher than 100 MHz. Even with a very steep filter (which tend to have undesirable side effects) it is still possible that frequency components above the Nyquist frequency make it to the ADC resulting in aliasing artifacts.

[pascal_sweden]

Is all the filtering done in the analog front-end only?
Or what kind of (additional) filtering can be done in the digital stage? Or is it then already too late?

[Lightages]

To filter a signal to prevent aliasing, it needs to be done before the A/D. In other words it needs to be an analog filter on the front end before any digital sampling. The closer to the Nyquist limit, the steeper the filter.

[pascal_sweden]

Wanted to share a link from Maxim with a nice explanation about Filter basics and Anti-aliasing
http://www.maximintegrated.com/en/app-notes/index.mvp/id/928

They also cover interesting section about Undersampling and Oversampling techniques.

[marmad]

Here's a graph illustrating the problem with the 100MHz BW - 250MSa/s combination - scaled from the image in this Agilent paper. If the Rigol had a flat-response, it would be better served in terms of aliasing (although not rise-time measurement accuracy) - but it's evident from BW measurements posted by users that it has something much closer to a Gaussian-response. BTW, this image assumes an exact 100MHz @ -3db, which the Rigol doesn't have - meaning it's response could have even more aliased content in it:

[pascal_sweden]

Would it be possible to make an automated measurement with a setup of LXI devices, including the Rigol scope itself, to verify this?

The described setup on the page below should do the trick:
http://pueski.de/?p=99

As the LXI command set is generic for all Rigol scopes, it would be easy to verify this for all the different families of Rigol scopes.

Anybody up for the job of implementing this automated benchmark performance using LXI? =)

BTW: How to inline an attachment picture in the actual thread posting? (As in the above posting)