Siglent SDS1104X-E (Lliberated to 200MHz) VS Rigol DS1202Z-E Bandwidth.

[TheBay]

About to order a new scope, currently using a new old stock Hameg HM-1004-3 with now a handful or hours on it. Great scope but purely analogue apart from the text overlays.

I sold my DS2074 (200MHz Hack) And am really missing having a DSO.
I'm about to order the Siglent SDS1104-XE as I REALLY want a 4 channel scope this time and plan to "Liberate it for 200MHz" as I need at least 200MHz bandwidth.

I did consider the Rigol DS1054Z but it seems like a big downgrade compared to the Rigol DS2074 I sold, I was really tempted to get it to gain the extra 2 channels but the bandwidth is just not enough even with all the "hacks"

However a big spanner has been thrown in the works, a friend just purchased a Rigol DS1202Z-E 200MHz 2 channel scope and it is punching well above it's weight, it must have a totally different front end to the DS1054 as it will do true 200MHz with ease and no sign of attenuation until after that point I can see. And looking at some quick waveforms of a wide frequency range it doesn't start to fall apart until over 400MHz, at 450MHz it will not display a trace. But easily capable of 200MHz+. I Have not seen any tear downs of this unit to see what configurated the front end is, actual true bandwidth and what ASIC/FPGA/CPU they are using. I don't even think I can find a video or a writeup on a real world test.

If they only made a 4 channel version of this, I'd buy it straight away but there isn't.

I've trawled this forum and Youtube for the SDS1104X-E and the SDS1204X-E but unable to find the information I need, I really like the look of this scope and the functionality. I've just cancelled the order for my DG812 as I may go for the Siglent AWG add on or a SDG-1032X.

What is the actual usable bandwidth on the SDS1104X-E "Liberated" / SDS1204X-E and what is the absolute max it will display?
How does it fare against the DS-1102Z-E in terms of bandwidth. Not taking in to consideration things like the uV/mV div range, GSa/s and offsets as I know the SDS1104/1204X-E are better in that regard.

[rf-loop]

At this time I do not have available any better image (perhaps here in forum is, others or also mine but can not find and now busy days, so can not try find.

This is SDS1104X-E what have some identity mistake...

Swept from 5MHz to 500MHz.  If you carefully look you can find -3dB is over 250MHz and -6dB is somewhere around 350MHz and -10db at 400MHz  (400MHz is 0.8*fNyquist. in this case) (note sweep start -4dB level)
Of course this test was made using real high quality RF generator and signal connected to scope with 50ohm feed thru. (I do not remember if also attenuator due to oscilloscope inputs reactance)
With this kind of (too)wide analog front end user need really understand how to use it for avoid falling to trap named "aliasing".

[bdunham7]

as I need at least 200MHz bandwidth.

Telll us more about what you do that requires 200MHz bandwidth.  Having a visible (triggered) but attenuated and probably distorted response to over-bandwidth signals is only helpful in certain situations.  The fact that a '200MHz' scope triggers on and displays a 400MHz signal is more of a problem than a benefit in many circumstances.  I would like to see a 100MHz BW limit feature on the SDS1204X-E, but I can't really complain to Siglent about that...

[Fungus]

The fact that a '200MHz' scope triggers on and displays a 400MHz signal is more of a problem than a benefit in many circumstances.

Yep. It has to have the sample rate to match, which would mean over 1GHz per channel.

Neither the Rigol nor the Siglent really has enough sample rate for their full analog bandwidth when you turn on multiple channels.

[tautech]

I sold my DS2074 (200MHz Hack) And am really missing having a DSO.
I'm about to order the Siglent SDS1104-XE as I REALLY want a 4 channel scope this time and plan to "Liberate it for 200MHz" as I need at least 200MHz bandwidth.

And you don't consider you're down grading going to a SDS1104/1204X-E ? 

You had 2 GSa/s and 56 Mpts and 8" display whereas SDS1000X-E series cannot offer any of that.

I did consider the Rigol DS1054Z but it seems like a big downgrade compared to the Rigol DS2074 I sold, I was really tempted to get it to gain the extra 2 channels but the bandwidth is just not enough even with all the "hacks"

Yes again that will be an even bigger downgrade.

Your DS2000 offered 5x sampling rate @ 200 MHz with both channels active whereas SDS1000X-E can only offer 2.5 the sampling required @200 MHz with all channels active.

Yes BW is a bitch as you must have sufficient sampling to match it and particularly as you're wanting to push past rated BW which will risk aliasing so IMO you will be better served with a DSO with 2 GSa/s sampling.
SDS2104X Plus can easily do this as despite its rated BW being just 100 MHz we have tested its -3dB point as ~185 MHz and as you're into hacking 350 MHz on all 4 channels can be achieved and 500 MHz on 2 channels if you push it right to the limit. There are reports of ~600 MHz BW for the 500 MHz option which still provides 3x+ sampling for the two 2GSa/s ADC's.

Take a moment to think more deeply about DSO selection.

[TheBay]

Most of the time I will be using 2x Channels and the SDS1104X-E has 2 ADC's so will be 2x 1GSa which is the same as the DS2000 (I hope I've got this right) I found the DS2000 to be slow to use and a bit noisy looking at small signals and from what I have seen the Siglent models are better in this department, also there is no 4 channel DS2000 and the DS4000 range has been discontinued.

When I need to use 4 channels the bandwidth is not high, so the SDS1104X-E should be fine I hope, examples where I need 4 channels are: Multiple power rails and DC-DC, Primary and secondary sides to SMPSU's or other topologies, looking at "Dead time", this is often where I run out of channels quickly and need to look at multiple parts of a circuit/psu while watching dead time. So 4 channels would really help here.
Also audio/AF in and outs.

Use case for me is very broadbanded, electronic and electrical repair, circuit design, amateur/commercial radio such as repair and also design, serial uarts, playing with uC's, mainly the lower end STM32's. I'm really used to having up to 200MHz and would like that again.

Also this is a birthday present and I've been nagged what do I want as I'm difficult to buy for and she suggested a new "sillyscope" as she puts it.

I sold my DS2074 (200MHz Hack) And am really missing having a DSO.
I'm about to order the Siglent SDS1104-XE as I REALLY want a 4 channel scope this time and plan to "Liberate it for 200MHz" as I need at least 200MHz bandwidth.

And you don't consider you're down grading going to a SDS1104/1204X-E ? 

You had 2 GSa/s and 56 Mpts and 8" display whereas SDS1000X-E series cannot offer any of that.

I did consider the Rigol DS1054Z but it seems like a big downgrade compared to the Rigol DS2074 I sold, I was really tempted to get it to gain the extra 2 channels but the bandwidth is just not enough even with all the "hacks"

Yes again that will be an even bigger downgrade.

Your DS2000 offered 5x sampling rate @ 200 MHz with both channels active whereas SDS1000X-E can only offer 2.5 the sampling required @200 MHz with all channels active.

Yes BW is a bitch as you must have sufficient sampling to match it and particularly as you're wanting to push past rated BW which will risk aliasing so IMO you will be better served with a DSO with 2 GSa/s sampling.
SDS2104X Plus can easily do this as despite its rated BW being just 100 MHz we have tested its -3dB point as ~185 MHz and as you're into hacking 350 MHz on all 4 channels can be achieved and 500 MHz on 2 channels if you push it right to the limit. There are reports of ~600 MHz BW for the 500 MHz option which still provides 3x+ sampling for the two 2GSa/s ADC's.

Take a moment to think more deeply about DSO selection.

[bdunham7]

OK, to answer your specific question on bandwidth:

Using the PP510 probe that came with it in 10X mode, my SDS1104X-E has a -3db point of right around 200MHz on the higher ranges (10mV/div actual, or 100mV/div @ 10X) and triggers cleanly on a 400MHz signal, which is attenuated about 10dB.  Using a coax and 50R terminator, the 3dB point is 270MHz.  You can trigger on signals above 400MHz, including aliased signals above 500MHz, but they are noisy, attenuated and the frequency counter doesn't read them properly.

The lower ranges are slightly different and seem to have a steeper roll off and more noise or instability. It appears that an additional gain stage is switched in at this point.  It can't trigger cleanly on lower-level signals above 300 to 350MHz, but the 50R -3dB point appears to be about 300MHz. 

That's all with one channel active.  With the ADC running multiple channels at 500MSa/s, you need to avoid any signal with anything near those frequencies because  you can get some truly odd results.

[tautech]

Most of the time I will be using 2x Channels and the SDS1104X-E has 2 ADC's so will be 2x 1GSa which is the same as the DS2000 (I hope I've got this right)

Experienced DSO users know this basic stuff but we are never quite sure hence my concern that you may have overlooked it.
Also this is where a 4ch DSO with a single ADC comes unstuck with its sampling rate shared by all channels when all are active and it can be a trap for young players pushing past BW limits.

[radiolistener]

However a big spanner has been thrown in the works, a friend just purchased a Rigol DS1202Z-E 200MHz 2 channel scope and it is punching well above it's weight, it must have a totally different front end to the DS1054 as it will do true 200MHz with ease and no sign of attenuation until after that point I can see. And looking at some quick waveforms of a wide frequency range it doesn't start to fall apart until over 400MHz, at 450MHz it will not display a trace.

This is a strong indication of a broken oscilloscope. Such indication means that antialias filter doesn't works properly, because this oscilloscope has sample rate 1 GS/s and it's sampling bandwidth is 500 MHz. So, it should attenuate 400-500 MHz for at least 60 dB. 

I advise you to avoid buying such oscilloscopes, because it will suffer from aliases.

If you're needs 200 MHz bandwidth, then look for oscilloscope with sample rate not less than 2 GS/s.

[Fungus]

However a big spanner has been thrown in the works, a friend just purchased a Rigol DS1202Z-E 200MHz 2 channel scope and it is punching well above it's weight, it must have a totally different front end to the DS1054 as it will do true 200MHz with ease and no sign of attenuation until after that point I can see. And looking at some quick waveforms of a wide frequency range it doesn't start to fall apart until over 400MHz, at 450MHz it will not display a trace.

This is a strong indication of a broken oscilloscope. Such indication means that antialias filter doesn't works properly, because this oscilloscope has sample rate 1 GS/s and it's sampling bandwidth is 500 MHz. So, it should attenuate 400-500 MHz for at least 60 dB. 

Nowhere did he say that 400MHz is displayed at full amplitude.

[radiolistener]

Yep. It has to have the sample rate to match, which would mean over 1GHz per channel.

Neither the Rigol nor the Siglent really has enough sample rate for their full analog bandwidth when you turn on multiple channels.

Yes, my SDS1102X @ 1 GS has just -45 dB attenuation for 500 MHz.
With such 100 MHz analog frontend it would be better to have 2 GS ADC, but SDS1102X use just 1 GS.

And it going to be even worse when you use two channels, because sample rate is reduced to 0.5 GS.

It has:
-1.338 dB @ 100 MHz
-4.24 dB @ 200 MHz
-11 dB @ 300 MHz
-25 dB @ 400 MHz
-45 dB @ 500 MHz
-48 dB @ 600 MHz

[tautech]

Yep. It has to have the sample rate to match, which would mean over 1GHz per channel.

Neither the Rigol nor the Siglent really has enough sample rate for their full analog bandwidth when you turn on multiple channels.

Yes, my SDS1102X @ 1 GS has just -45 dB attenuation for 500 MHz.
With such 100 MHz analog frontend it would be better to have 2 GS ADC, but SDS1102X use just 1 GS.

And it going to be even worse when you use two channels, because sample rate is reduced to 0.5 GS.

It has:
-1.338 dB @ 100 MHz
-4.24 dB @ 200 MHz
-11.175 dB @ 300 MHz
-25.86 dB @ 400 MHz
-45.129 dB @ 500 MHz
-48.577 dB @ 600 MHz

SDS1000X is a 2ch 200 MHz design. 

[radiolistener]

SDS1000X is a 2ch 200 MHz design. 

I'm using SDS1102X which bandwidth is limited in hardware with capacitor to 100 MHz.

Does SDS1000X-E series use antialias filtering with more sharp slope?

[tautech]

SDS1000X is a 2ch 200 MHz design. 

I'm using SDS1102X which bandwidth is limited in hardware with capacitor to 100 MHz.

Does SDS1000X-E series use antialias filtering with more sharp slope?

I believe so as BW is controlled by SW and gives a little steeper roll off.
But all scopes are very different and as an example a SDS2104X Plus rated as 100 MHz has a -3db point of some 185 MHz however that is a 500 MHz front end design.

[bdunham7]

This is a strong indication of a broken oscilloscope. Such indication means that antialias filter doesn't works properly, because this oscilloscope has sample rate 1 GS/s and it's sampling bandwidth is 500 MHz. So, it should attenuate 400-500 MHz for at least 60 dB. 

I advise you to avoid buying such oscilloscopes, because it will suffer from aliases.

If you're needs 200 MHz bandwidth, then look for oscilloscope with sample rate not less than 2 GS/s.

I agree that more samples is better, but I don't agree with characterizing products like these as substandard or defective because they don't meet some arbitrary requirement.  I'm more interested in understanding their characteristics so as to best use them and avoid any traps.  Makers of good test equipment used to explicitly list equipment limitations and shortcomings in manuals just for this reason.  Now the marketing wankers only want superlatives and optimistic puffery.

Attenuating a 400MHz signal 60dB, even in a scope with the -3dB point right at 200MHz, would require a brick wall filter that would give you a cartoon for a step response.  I don't think the SDS1204X-E aliasing response to over-bandwith signals is intolerable at 1GSa/s and for a given waveform, a steeper filter would result in just as much distortion of a different character.  The real issues arise at 500MSa/s.

Nowhere did he say that 400MHz is displayed at full amplitude.

It is only attenuated 10dB and is displayed clear and steady.

Does SDS1000X-E series use antialias filtering with more sharp slope?

I'm not sure what the slope is in the 100MHz version, but in the 200MHz version it is not a sharp slope and appears to just be a single pole lowpass.  I think I posted a while back about the 100MHz version--this issue isn't the slope, it's that the (marketing) BW filter is digital after the ADC, so it is not effective in the 4-channel mode.  Here is a shot of two signals, one is 100MHz and the other is 400MHz.  IIRC the issue is identical even with the 100MHz BW (1104X-E) in place.

[Fungus]

Nowhere did he say that 400MHz is displayed at full amplitude.

It is only attenuated 10dB and is displayed clear and steady.

 

Anybody know the real bandwidth of the 100MHz model? It's cheaper and might be better (if -3dB bandwidth is near 200MHz...)

[MarkF]

Rigol DS1074Z bandwidth (hacked to 100MHz).
You decide.

[bdunham7]

Anybody know the real bandwidth of the 100MHz model? It's cheaper and might be better (if -3dB bandwidth is near 200MHz...)

That shouldn't be a surprise.  A single-pole filter is 6db/octave, right?

IIRC, the -3dB point of the 100MHz version (which is exactly the same price as the 200MHz version  ) was 120MHz or so.  However, as I mentioned, there's not much point to not hacking it because the 100MHz BW filter is not an anti-aliasing filter--the exact same issue of 400Mhz aliasing down to 100Mhz still happens at 500MSa/s.  The SDS1104X-E is the best entry level two channel scope on the market.  The only cost-effective solution to this--if it is a problem--would be lower performance probes.  The cheapo PP510s that come with the 1104X-E aren't the greatest--I only got one and it sort of falls apart--but their performance is astonishingly good--which can be a bad thing in some cases.

[nctnico]

Yep. It has to have the sample rate to match, which would mean over 1GHz per channel.

Neither the Rigol nor the Siglent really has enough sample rate for their full analog bandwidth when you turn on multiple channels.

Yes, my SDS1102X @ 1 GS has just -45 dB attenuation for 500 MHz.
With such 100 MHz analog frontend it would be better to have 2 GS ADC, but SDS1102X use just 1 GS.

And it going to be even worse when you use two channels, because sample rate is reduced to 0.5 GS.

It has:
-1.338 dB @ 100 MHz
-4.24 dB @ 200 MHz
-11 dB @ 300 MHz
-25 dB @ 400 MHz
-45 dB @ 500 MHz
-48 dB @ 600 MHz

Actually that is pretty bad. You really want an attenuation of 40dB or more at the Nyquist frequency (in this case 250MHz) otherwise the sin x/x reconstruction doesn't work and you get really wobbly signals. Try measuring a square wave in a logic circuit for example; it will show way worse than it has to.

[bdunham7]

Actually that is pretty bad. You really want an attenuation of 40dB or more at the Nyquist frequency (in this case 250MHz) otherwise the sin x/x reconstruction doesn't work and you get really wobbly signals. Try measuring a square wave in a logic circuit for example; it will show way worse than it has to.

I did some experiments with RF signals of 80, 100 and 120MHz and some 3rd, 4th and 5th harmonics 6dB below the fundamental.  The conclusion I came to was that Siglent really had dialed in the compromises for a nice correlated response--I could shift the phase so that it rolled through from squarish to lumpy to triangularish to lumpy and so on--just as would be expected--up to almost 400MHz, then above that the effects diminished greatly but the signal remained stable.  With harmonics above 480MHz, you see a bit of shimmery noise, but not terrible.  Essentially the roll off appears to be a perfect compromise for realistic harmonic levels.  That's at 1GSa/s, but at 500MSa/s, it's a train wreck.

It might be different if I was trying to get an eye diagram or something, but I can't think of a readily available logic signal that would be a good test--they're either too fast or too slow.  Any ideas on something common that you think would break the scope?

[nctnico]

Actually that is pretty bad. You really want an attenuation of 40dB or more at the Nyquist frequency (in this case 250MHz) otherwise the sin x/x reconstruction doesn't work and you get really wobbly signals. Try measuring a square wave in a logic circuit for example; it will show way worse than it has to.

I did some experiments with RF signals of 80, 100 and 120MHz and some 3rd, 4th and 5th harmonics 6dB below the fundamental.  The conclusion I came to was that Siglent really had dialed in the compromises for a nice correlated response--I could shift the phase so that it rolled through from squarish to lumpy to triangularish to lumpy and so on--just as would be expected--up to almost 400MHz, then above that the effects diminished greatly but the signal remained stable.  With harmonics above 480MHz, you see a bit of shimmery noise, but not terrible.  Essentially the roll off appears to be a perfect compromise for realistic harmonic levels.  That's at 1GSa/s, but at 500MSa/s, it's a train wreck.

It might be different if I was trying to get an eye diagram or something, but I can't think of a readily available logic signal that would be a good test--they're either too fast or too slow.  Any ideas on something common that you think would break the scope?

Any signal with reasonably sharp edges (like you find in a typical microcontroller circuit) will wreck it. I have hacked my GW Instek GDS2204E to 300MHz but it made it crap to look at digital signals with 3 or 4 channels on (500Ms/s); I still need to revert the hack.

[bdunham7]

Any signal with reasonably sharp edges (like you find in a typical microcontroller circuit) will wreck it. I have hacked my GW Instek GDS2204E to 300MHz but it made it crap to look at digital signals with 3 or 4 channels on (500Ms/s); I still need to revert the hack.

I meant at 1GSa/s, where the step response is exemplary.  No need to struggle breaking it in 4-channel mode.  Here's a very fast 10MHz square wave in either mode.





And a brick wall pre-ADC filter isn't everyone's answer either, although it would give a stable display which might be good for digital.  Here's the same 10MHz fast edge square wave put through a 60MHz brick wall (7th order Chebyshev or some such monstrosity) compared to a reconstructed square wave using 3rd and 5th harmonics plus a diminished 7th to try and sort of match the bandwidth response of the filter.  No bueno if you care about phase...

[MarkF]

I posted the step response on my Rigol DS1000Z hacked to 100MHz.
If you truly need 200MHz, the DS1000Z is probably not for you.
But, I will give up bandwidth for four channels any day.



How much are you willing to spend?  Have you looked at the Rigol DS5000 Series.
I believe Dave did a video and there is a thread about hacking it.
Don't know if it's been hacked yet or not.

[TheBay]

But that is a 4 Channel DS1000Z Series, the 2 Channel DS1202Z-E 200MHz has a different front end. It is not comparable and also not comparable to the one that IMSAI guy has.

I posted the step response on my Rigol DS1000Z hacked to 100MHz.
If you truly need 200MHz, the DS1000Z is probably not for you.
But, I will give up bandwidth for four channels any day.



How much are you willing to spend?  Have you looked at the Rigol DS5000 Series.
I believe Dave did a video and there is a thread about hacking it.
Don't know if it's been hacked yet or not.

[StillTrying]

What's wrong with Dots + Persistence.