Siglent SDS6000A DSO's 500MHz-2GHz

[nctnico]

When will Dave get his hands on one of these babies and do a review ?

Yes, if he is serious to test using a Bodnar fast rising pulse generator (30..40ps rise/fall time) so we will see the beef.

For modern DSOs a pulse is useless to check anything. You'll need a levelled RF generator to test the bandwidth and find other signal processing artefacts.

BTW: I found this Xilinx case study about the SDS6000 pro which shows the internal architecture: https://japan.origin.xilinx.com/content/dam/xilinx/publications/powered-by-xilinx/siglent-case-study.pdf

However a teardown would be interesting between a 8bit SDS6000A VS a 12bit SDS6000 pro.

It's simply different board.. Not much to compare.

... or also not. We cannot know until someone had a look under the hood of both instruments. It might well be that there is no hardware difference at all...

It could be. It depends on the cost of the ADCs versus number of units sold.

[luudee]

It's simply different board.. Not much to compare.

Do you know that for a fact, or are you just trolling ?!

[jemangedeslolos]

It would be really surprising if the 8 bit version was supplied with 4 x adc 12 bit 5GSa/s.
The price difference must be huge between 8 and 12 bit 5GSa/s ADC....but they have some headroom given the low-end price.

If it is the same architecture, I will think a lot less to buy it and I think someone will be very interested in hacking it if it's not already done 

[mawyatt]

It would be really surprising if the 8 bit version was supplied with 4 x adc 12 bit 5GSa/s.
The price difference must be huge between 8 and 12 bit 5GSa/s ADC....but they have some headroom given the low-end price.

If it is the same architecture, I will think a lot less to buy it and I think someone will be very interested in hacking it if it's not already done 

Agree! A 12bit 5GSPS ADC is very difficult to develop and thus I'd expect quite expensive, and with 4X even more expensive. This is not the same as using a TI VGA chip in the front end that can be "programmed" for added bandwidth and followed by a high speed fixed 8 bit ADC like in the SDS2000X+ series. The 8 bit ADC Siglent is using in the SDS2000X+ is quite good tho for an 8 bitter, but likely not expensive.

Very disappointed that the 6000 will not be available in NA with 12 bit ADCs, and not sure how good the ERES will behave regarding ENOB. The SDS2000X+ is pretty good tho, so hopefully Siglent will be able to squeeze out a few extra reliable bits like the 2000.

Soon I guess we'll know when a few quality reviews are posted.

Best,

[2N3055]

It's simply different board.. Not much to compare.

Do you know that for a fact, or are you just trolling ?!

Why would I be trolling ?? What kind of a question is that?

Which part of 4 x 8bit ADC versus 2x12bBit  ADC didn't you understand. Which part that it has different memory sizes and general specification didn't you get?

12Bit version has VGA output. 8 bit one has HDMI... etc etc..

Are you one of those people that live in fantasy world that think that PCB layout can be fixed by a software patch? That you can install large BGA ADC converters by software unlock?

Yes, I know it for the fact..  It is different hardware. Only thing they share is case.

[2N3055]

It would be really surprising if the 8 bit version was supplied with 4 x adc 12 bit 5GSa/s.
The price difference must be huge between 8 and 12 bit 5GSa/s ADC....but they have some headroom given the low-end price.

If it is the same architecture, I will think a lot less to buy it and I think someone will be very interested in hacking it if it's not already done 

Agree! A 12bit 5GSPS ADC is very difficult to develop and thus I'd expect quite expensive, and with 4X even more expensive. This is not the same as using a TI VGA chip in the front end that can be "programmed" for added bandwidth and followed by a high speed fixed 8 bit ADC like in the SDS2000X+ series. The 8 bit ADC Siglent is using in the SDS2000X+ is quite good tho for an 8 bitter, but likely not expensive.

Very disappointed that the 6000 will not be available in NA with 12 bit ADCs, and not sure how good the ERES will behave regarding ENOB. The SDS2000X+ is pretty good tho, so hopefully Siglent will be able to squeeze out a few extra reliable bits like the 2000.

Soon I guess we'll know when a few quality reviews are posted.

Best,

Mike,

as always, faster the ADC is, you can interpolate a bit more... So 5GS/s will be able to get a bit better results than 2GS/s for low frequencies.. more samples for averaging is available.

I don't have it so I cannot measure how much exactly.

As a sidenote, I really like that it has 4 converters. I kinda like the simplicity that sampling is always the same at all times for all channels...

[mawyatt]

as always, faster the ADC is, you can interpolate a bit more... So 5GS/s will be able to get a bit better results than 2GS/s for low frequencies.. more samples for averaging is available.

This assumes the 8 bit converter doesn't have any bad characteristics like hysteresis or high DNL. If it's as good as the ADC in the SDS2000X+ regarding these parameters that will be a plus

Not sure this will be enough of a compelling reason for us to consider a 6000 tho, really would like 12 bit ADCs. If those 12 bit ADCs come directly from LeCroy, then understand the market concessions Siglent must make.

Best

[jemangedeslolos]

It's simply different board.. Not much to compare.

Do you know that for a fact, or are you just trolling ?!

Why would I be trolling ?? What kind of a question is that?

Which part of 4 x 8bit ADC versus 2x12bBit  ADC didn't you understand. Which part that it has different memory sizes and general specification didn't you get?

12Bit version has VGA output. 8 bit one has HDMI... etc etc..

Are you one of those people that live in fantasy world that think that PCB layout can be fixed by a software patch? That you can install large BGA ADC converters by software unlock?

Yes, I know it for the fact..  It is different hardware. Only thing they share is case.

Hello Sinisa,

Where did you see that the 12 bit version didn't have the same memory amount and had a VGA output ?
On this specsheet, we can see HDMI and 500Mpts memory. Maybe an out dated specsheet  ?

https://www.siglent.com/upload_file/document/SDS6000_Pro_Datasheet_DS0106P_C02A.pdf

[nctnico]

It's simply different board.. Not much to compare.

Do you know that for a fact, or are you just trolling ?!

Why would I be trolling ?? What kind of a question is that?

Which part of 4 x 8bit ADC versus 2x12bBit  ADC didn't you understand. Which part that it has different memory sizes and general specification didn't you get?

12Bit version has VGA output. 8 bit one has HDMI... etc etc..

Are you one of those people that live in fantasy world that think that PCB layout can be fixed by a software patch? That you can install large BGA ADC converters by software unlock?

Yes, I know it for the fact..  It is different hardware. Only thing they share is case.

Hello Sinisa,

Where did you see that the 12 bit version didn't have the same memory amount and had a VGA output ?
On this specsheet, we can see HDMI and 500Mpts memory. Maybe an out dated specsheet  ?

https://www.siglent.com/upload_file/document/SDS6000_Pro_Datasheet_DS0106P_C02A.pdf

Indeed. If you put this spec sheet next to the one from the SDS6000A then you'll see there are a lot of similaties AND a major difference. The SDS6000A (8 bit) supports 5Gs/s on all channels where the SDS6000 Pro supports 'only' 2.5Gs/s on all channels. Maybe there is some interleaving going on to get more bits. IMHO it is impossible to say whether these oscilloscopes have the same hardware under the hood or not without doing a teardown of both. Since a DSO is a relatively low volume product the NRE costs for 2 ADCs might be higher compared to developing 1 higher spec ADC and use it in an 8bit mode on lower spec models.

For example: Keysight has done something similar on some of their higher end models where they limit the DSO to use half the samplerate the ADC is capable of.

[jemangedeslolos]

It's simply different board.. Not much to compare.

Do you know that for a fact, or are you just trolling ?!

Why would I be trolling ?? What kind of a question is that?

Which part of 4 x 8bit ADC versus 2x12bBit  ADC didn't you understand. Which part that it has different memory sizes and general specification didn't you get?

12Bit version has VGA output. 8 bit one has HDMI... etc etc..

Are you one of those people that live in fantasy world that think that PCB layout can be fixed by a software patch? That you can install large BGA ADC converters by software unlock?

Yes, I know it for the fact..  It is different hardware. Only thing they share is case.

Hello Sinisa,

Where did you see that the 12 bit version didn't have the same memory amount and had a VGA output ?
On this specsheet, we can see HDMI and 500Mpts memory. Maybe an out dated specsheet  ?

https://www.siglent.com/upload_file/document/SDS6000_Pro_Datasheet_DS0106P_C02A.pdf

Indeed. If you put this spec sheet next to the one from the SDS6000A then you'll see there are a lot of similaties AND a major difference. The SDS6000A (8 bit) supports 5Gs/s on all channels where the SDS6000 Pro supports 'only' 2.5Gs/s on all channels. Maybe there is some interleaving going on to get more bits. IMHO it is impossible to say whether these oscilloscopes have the same hardware under the hood or not without doing a teardown of both. Since a DSO is a relatively low volume product the NRE costs for 2 ADCs might be higher compared to developing 1 higher spec ADC and use it in an 8bit mode on lower spec models.

For example: Keysight has done something similar on some of their higher end models where they limit the DSO to use half the samplerate the ADC is capable of.

On the Xilinx PDF you linked, there are 4 individual ADC. Maybe they can't process 5GSa/s 12bit on 4 channels by some kind of hardware limitation or by marketing decision with Lecroy partnership?

[tautech]

Truth is there are significant differences between Chinese SDS6000 Pro and western SDS6000A.
The specs are out there if you know where to find them however for us in the west only the SDS6000A is of interest.

Years back when SDS3000 was released in China the one I played with at Siglent HQ didn't even have a English UI menu option and it's very likely SDS6000 Pro will be the same ....... if some relationship with LeCroy still exists to market SDS6000 Pro in the west.
Only time will tell if we can chose between SDS6000 Pro/LeCroy model *** or SDS6000A. 

[2N3055]

That is interesting, I was looking at older datasheet, and was apparently comparing to 1GHz Chinese version from that datasheet.

But it still stays the same. 12 Bit converters are much more expensive and also not easy to get. So if the scope is 8 bit, converters won't be 12 bit inside and artificially made to behave like 8 bit ones.
Nico's Xilinx whitepaper shows all there is. Today scopes are several ADC connected to large FPGAs and that is it. There is nothing interesting really, they are all the same. All the magic is FPGA bitstreams and software.
Teardowns don't show much apart from what is obvious..

[nctnico]

That is interesting, I was looking at older datasheet, and was apparently comparing to 1GHz Chinese version from that datasheet.

But it still stays the same. 12 Bit converters are much more expensive and also not easy to get.

Again: you have to factor in the NRE costs. A high spec ADC may cost ball park 1 million USD to develop. Certainly not less than $250k. The production costs are pretty much the same after that no matter what resolution the ADC is. Given the relatively low number of units sold it just doesn't make sense to develop 2 seperate ADCs. OTOH it could be the 8 bit version uses off-the-shelve ADCs. It all depends on the economics and we simply don't know enough to be sure.

[rf-loop]

If you put this spec sheet next to the one from the SDS6000A then you'll see there are a lot of similaties AND a major difference. The SDS6000A (8 bit) supports 5Gs/s on all channels where the SDS6000 Pro supports 'only' 2.5Gs/s on all channels.

No.

SDS6000 Pro H12 and H10 2GHz models  have 5GSa/s for all channels simultaneously and then one 500M memory shared for all analog channels and as can see looks like same with all 8bit SDS6000A models.

Difficult to read public data sheets published by manufacturer instead of wild speculations?

---------------

Sidenote/Clarification in general:   Every SDS6000********  model have HDMI out. (SDS6000 Pro H10,  H12: 350MHz to 2GHz and SDS6000A: 500MHz to 2GHz models. ) 

[Performa01]

SDS6000 Pro H12 and H10 2GHz models  have 5GSa/s for all channels simultaneously and then one 500M memory shared for all analog channels and as can see looks like same with all 8bit SDS6000A models.

Indeed.

One could make an educated guess that all the worldwide selling SDS6000A models originate from the chinese 2 GHz SDS6204 H12 Pro flagship.

Some folks here already noticed that the price difference between the various bandwidth models is rather low - so this is a first pointer that
a) the hardware is identical for all models - other than for the SDS5000X
b) this is clearly aimed for the professional market and there is no particularly cheap entry model for all the hackers out there.

We also get a frontend without compromises: full sensitivity at full bandwidth for all models up to 2 GHz.

Whether Siglent has gone to the effort to develop a new acquisition board to fit a vastly different 8 bit ADC - that's another question

[rf-loop]

SDS6000 Pro H12 and H10 2GHz models  have 5GSa/s for all channels simultaneously and then one 500M memory shared for all analog channels and as can see looks like same with all 8bit SDS6000A models.

Indeed.

..... this is clearly aimed for the professional market and there is no particularly cheap entry model for all the hackers out there.

We also get a frontend without compromises: full sensitivity at full bandwidth for all models up to 2 GHz.

Yes, it is for professional markets what may mean also more "things" than just instrument itself.



Small note for BW:

2GHz model BW is specified as 1 GHz when input vertical scale is under 2.3 mV/div. (1:1)

[TurboTom]

Interesting information from datasheets published by @rf-loop:

The specified power consumption of the lower frequency models of the SDS 6000 H10/12 Pro is lower (150W max / 90W typ / 4W s/b) vs. the highest range model (193W / 123W / 4W).

The SDS 6000 A datasheet lists the 193W / 123W / 4W for the whole range of instruments.

Considering Siglent's datasheet are usually quite accurate, this may be a hint towards an interesting conclusion... 

[nctnico]

If you put this spec sheet next to the one from the SDS6000A then you'll see there are a lot of similaties AND a major difference. The SDS6000A (8 bit) supports 5Gs/s on all channels where the SDS6000 Pro supports 'only' 2.5Gs/s on all channels.

No.

SDS6000 Pro H12 and H10 2GHz models  have 5GSa/s for all channels simultaneously and then one 500M memory shared for all analog channels and as can see looks like same with all 8bit SDS6000A models.

Difficult to read public data sheets published by manufacturer instead of wild speculations?

You better check your own links: on pages 3 and 10 of the SDS6000 Pro H12 spec sheet it states that it has 2.5 Gs/s per channel with all channels enabled. Google translate tells me that 5Gs/s is only supported in interleave mode with half the channels (1+2 or 3+4) enabled.

[Performa01]

You better check your own links: on page 3 and page 12 of the SDS6000 Pro H12 spec sheet it states that it has 2.5 Gs/s per channel with all channels enabled. I don't need to understand Chinese to spot that.

Sometimes some understanding doesn't hurt. These are the specifications in question for the SDS6204 H10/12 Pro, translated from the Datasheet:

Real-time sampling rate: 10 GSa/s (ESR) @ per channel
Memory depth              : 500 Mpts/ch（single channel), 250 Mpts/ch (dual channel), 125 Mpts/ch (four-channels)

[nctnico]

You better check your own links: on page 3 and page 12 of the SDS6000 Pro H12 spec sheet it states that it has 2.5 Gs/s per channel with all channels enabled. I don't need to understand Chinese to spot that.

Sometimes some understanding doesn't hurt. These are the specifications in question for the SDS6204 H10/12 Pro, translated from the Datasheet:

Real-time sampling rate: 10 GSa/s (ESR) @ per channel
Memory depth              : 500 Mpts/ch（single channel), 250 Mpts/ch (dual channel), 125 Mpts/ch (four-channels)

Now explain what ESR means... Google translate tells me: :ESR: Enhance the sampling rate, obtain better measurement accuracy through 2x interpolation which still doesn't explain what is does exactly. It looks more like the 2GHz model has different acquisition hardware compared to the 1GHz and lower models.

[2N3055]

For domestic market (China) there are two versions of  hardware: one up to 1 GHz and 2 GHz. 2 GHz version has 4 ADC and has 5GS/s at all times (10 GS/s with ESR). Also, if you look closely,  it has slightly different memory architecture (Unified memory : 500, 250, 125 MS as opposed to to 2x 250/125MS) and therefore different FPGA configuration and such. Also, a 2GHz capable front end.

The architecture for 2GHz domestic one was probably used as basis to develop 8 bit version of that, and with 4 (different type) 8 bit ADC was made in worldwide platform, that is single for all bandwidths (500MHz, 1 GHz and 2 GHZ). They all share same motherboard, front end etc. There are software upgrade options that use that fact, otherwise it wouldn't be possible.. As mentioned by Nico many times, BOM optimization, streamlining development and such.

Therefore domestic Chinese SDS6000 series is same family, but worldwide motherboard is different board and should be viewed as such.

@Tom,  difference in power ( <= 1GHz and 2 GHz domestic China version ) comes from, as explained, different hardware..
Variation in consumption comes from scope CPU load, and channels enabled..
With all math channels, decodes, digital scope power goes up. In stop mode it goes down..

[tv84]

Some folks here already noticed that the price difference between the various bandwidth models is rather low - so this is a first pointer that
a) the hardware is identical for all models - other than for the SDS5000X
b) this is clearly aimed for the professional market and there is no particularly cheap entry model for all the hackers out there.

c) This is the first Siglent model with "licensing hack" by design.

[2N3055]

You better check your own links: on page 3 and page 12 of the SDS6000 Pro H12 spec sheet it states that it has 2.5 Gs/s per channel with all channels enabled. I don't need to understand Chinese to spot that.

Sometimes some understanding doesn't hurt. These are the specifications in question for the SDS6204 H10/12 Pro, translated from the Datasheet:

Real-time sampling rate: 10 GSa/s (ESR) @ per channel
Memory depth              : 500 Mpts/ch（single channel), 250 Mpts/ch (dual channel), 125 Mpts/ch (four-channels)

Now explain what ESR means... is it some equalent time sampling system? Google translate tells me: :ESR: Enhance the sampling rate, obtain better measurement accuracy through 2x interpolation which still doesn't explain what is does exactly.

Yes, ESR is interpolation method. There is LeCroy ESR whitepaper that explains it. It is still 5GS/s sampling but it does help with built in measurements, making them more accurate.

[rf-loop]

You better check your own links: on page 3 and page 12 of the SDS6000 Pro H12 spec sheet it states that it has 2.5 Gs/s per channel with all channels enabled. I don't need to understand Chinese to spot that.

Sometimes some understanding doesn't hurt. These are the specifications in question for the SDS6204 H10/12 Pro, translated from the Datasheet:

Real-time sampling rate: 10 GSa/s (ESR) @ per channel
Memory depth              : 500 Mpts/ch（single channel), 250 Mpts/ch (dual channel), 125 Mpts/ch (four-channels)

Now explain what ESR means... Google translate tells me: :ESR: Enhance the sampling rate, obtain better measurement accuracy through 2x interpolation which still doesn't explain what is does exactly. It looks more like the 2GHz model has different acquisition hardware compared to the 1GHz and lower models.

Samplerate is, as we normally talk about, 5GSa/s  simultaneously for every channel in SDS6000********  2GHz models in China and in all SDS6000A models outside of China. From which it follows that fNyquist based to fastest native samplerate is 2.5GHz. 10GSa/s ESR do not change this. 

ESR in LeCroy oscilloscopes is well explained in attached LeCroy white paper.  But this is LeCroy. How it is implemented in Siglent model... there is not available public Siglent white paper about it.

ETA: lack of my browser refresh after come bac from out and  I did not see @2N3055  previously wroted good answer.

[mawyatt]

The LeCroy white paper reference by 2N3055 is a must read for anyone using any of these mentioned instruments.

http://cdn.teledynelecroy.com/files/whitepapers/enhanced-sample-rate-whitepaper.pdf

Edit: A quick search shows TI has candidate 12 bit ADCs, as does AD, and they aren't cheap at ~$1K per chip.

Best,