Siglent SDS6000A DSO's 500MHz-2GHz

[Performa01]

5000 MHz / 128 ?

Oh yes, of course!

[gf]

5000 MHz / 128 ?

Oh yes, of course!

What still irritates me is the low sample rate of only 39M that the ADC cores would have and the large number of interleaved cores. For comparison, the HMCAD1511 which is used in several low-cost 1GSa/s scope models has 8 interleaved ADC cores, with 125MSa/s each. And for faster ADCs, I'd rather expect an even higher sample rate per core. But who knows...

[hpw]

5000 MHz / 128 ?

Oh yes, of course!

What still irritates me is the low sample rate of only 39M that the ADC cores would have and the large number of interleaved cores. For comparison, the HMCAD1511 which is used in several low-cost 1GSa/s scope models has 8 interleaved ADC cores, with 125MSa/s each. And for faster ADCs, I'd rather expect an even higher sample rate per core. But who knows...

This looks as a clear 5000 MHz / 128  = 39.0625Mhz SQUARE Wave or other combination(s)... guessing as exact 128 diff by the reference frequency (synthesizer) or oscillation on the VRef power.
The other one as about -100dB are for me the spurs...

Hopefully the upcoming SDS3000 12 bit, SDS7000 12 bit are any better or just send them back to the development group.

Hp

 

[mawyatt]

Some interesting high speed ADCs utilizing multiple individual channels, recall from ~2008 when we saw the early Fuji 56GPSP ADC later expanded to 64GSPS, this was developed in 65nm CMOS with 2 ADCs per chip.

Later some other interesting ADCs utilizing multiple individual channels emerged, like the 2014 IBM 90GSPS in 32nm SOI, with 5.2 ENOB at 20GHz which utilized 64 ADC channels @ 1.4GSPS per ADC, and the 2017 Broadcom 64GSPS in 20nm with 128 channels per ADC (has 4 ADCs per chip).

https://indico.cern.ch/event/121657/attachments/68435/98170/ADC_DAC_CERN.pdf

https://www.electronicspecifier.com/products/mixed-signal-analog/fujitsu-launches-ultrafast-56gsa-s-8-bit-adc-technology-for-100g-coherent-receivers-and-high-performance-test-equipment

https://www.fujitsu.com/uk/Images/c45.pdf

Best,

[Performa01]

Trigger Jitter

The datasheet specifies the trigger jitter as follows:

•   <9ps RMS (typical) for ≥300MHz sine and ≥6 divisions peak to peak amplitude for vertical gain settings from 2.5mV/div to 10V/div.
•   <5ps RMS (typical) for ≥500MHz sine and ≥6 divisions peak to peak amplitude for vertical gain settings from 2.5mV/div to 10V/div.

The EXT trigger jitter is not exactly stellar at <200 ps RMS – but then again this is a classical analog (comparator) trigger and not the fully digital trigger system that we get when using one of the input channels as trigger source.

Now let’s verify this with a 1 GHz sine signal from an OCXO-driven AWG (SDG7102A), fed into channels 3 and 4 of an SDS6204 H12 Pro (which has an individual ADC for each channel) via a 12.4 GHz resistive power splitter. This way we can observe the jitter in the trigger channel as well as a not triggered channel, where both are using different ADCs, hence are completely unrelated.

The high quality 1 GHz sine signal has been chosen for its fast edges and low inherent jitter – after all we want to characterize the DSO and not the signal source. See the attached screenshot which has been taken after more than 15 minutes running with infinite persistence:


SDS6204 Pro H12_Trigger_Jitter_1GHz_15m

At a timebase of 100 ps/div, we can see the peak to peak jitter in the triggered as well as the non-triggered channel after more than 15 minutes at infinite persistence.

The jitter measurements are as follows:
Triggered channel: 6 ps pk-pk, 698 fs rms;
Un-triggered channel: 11.3 ps pk-pk, 1.564 ps rms;
Skew Ch.3-Ch.4: 12.06 ps pk-pk, 1.594 ps rms;

[Martin72]

Hi,

For a new project we want to look at the signals of a 1000 Base T connection.
I have roughly thought about it and would consider a 1Ghz scope for this - is that enough or would we have to go even higher?
Either way, only the 6000A would come into question, it is one of the cheapest on the market despite its performance.
The only question is, 1Ghz or 2Ghz....

[2N3055]

Hi,

For a new project we want to look at the signals of a 1000 Base T connection.
I have roughly thought about it and would consider a 1Ghz scope for this - is that enough or would we have to go even higher?
Either way, only the 6000A would come into question, it is one of the cheapest on the market despite its performance.
The only question is, 1Ghz or 2Ghz....

Short version, 2GHz version would be better, but 1GHz should be usable..
Some lit... Holler if you need more..

DS7000A has compliance package  .... Just saying...

[Martin72]

Hi my friend,

DS7000A is too big for this project in financial terms..

Incidentally, this is an issue that affects both models.
Both want to be Pro models, and the data sheet says they are - but it's astonishing how few options come with these scopes.
The protocol and analysis options for our waverunner fill a whole page, unfortunately the 500Mhz are not enough.

[nctnico]

Look at the Tektronix appnote. Is says 8ns bit times. Just like 100Mbit ethernet, 1Gbit ethernet has a maximum frequency content of around 125MHz. CAT5E wiring just ain't going to pass higher frequencies. I'm not sure what the goal is here, but a 500MHz oscilloscope is going to do just fine to look at the signals. Compliance testing is of little use as you likely can't change anything on the transmitter / receiver chip (unless you are designing these kind of chips). What is left is the transmission system (wiring, filters, transformers) and for that you can use a network analyser to check the impedances of the transmission lines / system. All in all: what is the goal of the measurements?

[2N3055]

Look at the Tektronix appnote. Is says 8ns bit times. Just like 100Mbit ethernet, 1Gbit ethernet has a maximum frequency content of around 125MHz. CAT5E wiring just ain't going to pass higher frequencies. I'm not sure what the goal is here, but a 500MHz oscilloscope is going to do just fine to look at the signals. Compliance testing is of little use as you likely can't change anything on the transmitter / receiver chip (unless you are designing these kind of chips). What is left is the transmission system (wiring, filters, transformers) and for that you can use a network analyser to check the impedances of the transmission lines / system. All in all: what is the goal of the measurements?

Says minimum 400ps risetime, BW 10x of 125Mhz, sample rate of 5GS/S. That fits to roughly 1 GHz scope. That is minimum
125Mhz is not the problem here but risetime, as always. 125Mhz is guaranteed for long cable runs. For short cables it will be more..
You will see that they performed actual testing with 4 GHz scope with 10GS/s...

Naturally, if you want to measure what characteristics are is different to just checking if it passes basic requirements.

I don't know what requirements are. From what little was said, it is not cable testing... And for that, a LAN analyser with compliance testing would be better match anyways.

[Martin72]

I only got the flippant announcement "to snoop around in it"...
But I'll ask again.

[nctnico]

Look at the Tektronix appnote. Is says 8ns bit times. Just like 100Mbit ethernet, 1Gbit ethernet has a maximum frequency content of around 125MHz. CAT5E wiring just ain't going to pass higher frequencies. I'm not sure what the goal is here, but a 500MHz oscilloscope is going to do just fine to look at the signals. Compliance testing is of little use as you likely can't change anything on the transmitter / receiver chip (unless you are designing these kind of chips). What is left is the transmission system (wiring, filters, transformers) and for that you can use a network analyser to check the impedances of the transmission lines / system. All in all: what is the goal of the measurements?

Says minimum 400ps risetime, BW 10x of 125Mhz, sample rate of 5GS/S. That fits to roughly 1 GHz scope. That is minimum

No, that is Tektronix wanting to sell you an oscilloscope which is extreme overkill. Probably their lower bandwidth oscilloscope lack the analysis capabilities. If you look through the appnote you'll see that some measurements end at 100MHz and most of the accesoires (like filters and splitter) are specced up to a couple of hundred MHz.

[2N3055]

Look at the Tektronix appnote. Is says 8ns bit times. Just like 100Mbit ethernet, 1Gbit ethernet has a maximum frequency content of around 125MHz. CAT5E wiring just ain't going to pass higher frequencies. I'm not sure what the goal is here, but a 500MHz oscilloscope is going to do just fine to look at the signals. Compliance testing is of little use as you likely can't change anything on the transmitter / receiver chip (unless you are designing these kind of chips). What is left is the transmission system (wiring, filters, transformers) and for that you can use a network analyser to check the impedances of the transmission lines / system. All in all: what is the goal of the measurements?

Says minimum 400ps risetime, BW 10x of 125Mhz, sample rate of 5GS/S. That fits to roughly 1 GHz scope. That is minimum

No, that is Tektronix wanting to sell you an oscilloscope which is extreme overkill. Probably their lower bandwidth oscilloscope lack the analysis capabilities. If you look through the appnote you'll see that some measurements end at 100MHz.

Not Tektronix, the actual working group...

[Martin72]

In the meantime, the wishes have become more concrete, so the Siglent models are out.
Because protocol analyzes should also be possible and, as I already mentioned, the available options for the large models are rather limited.
For our Waverunner model, however, there are dozens of options available, but the hope that its bandwidth (500Mhz, smallest model) would be sufficient has died.
Lecroy recommends 1Ghz for T100 and 2.5Ghz for T1000.
But this doesn't belong here, maybe I'll open a new thread.

[nctnico]

In the meantime, the wishes have become more concrete, so the Siglent models are out.
Because protocol analyzes should also be possible and, as I already mentioned, the available options for the large models are rather limited.
For our Waverunner model, however, there are dozens of options available, but the hope that its bandwidth (500Mhz, smallest model) would be sufficient has died.
Lecroy recommends 1Ghz for T100 and 2.5Ghz for T1000.
But this doesn't belong here, maybe I'll open a new thread.

Opening a new thread sounds like a good plan. Having quite a bit of hands-on experience with various types of electrical and optical ethernet and other types of high speed serial interfaces, I strongly doubt an oscilloscope or protocol analyser is going to help solve a problem you may have. Getting the problem definition clear is likely way more useful as it will point into the direction of the solution.