Siglent SDS2000X Plus

[tautech]

As for the missing calibration certificate, I still haven't had any reply from Telonic on the matter. I'll let it stew for another day or two before pressing them for an answer. As Tautec said (tnx btw) there's no excuse for withholding this piece of documentation from the end customer.
JBG

If you get stuck let me know.
Flick me your SN# via PM and I can get a PDF copy from the factory.
Still, at some point I'll need your email so to forward it on to you.

[Vestom]

Does anyone have any experience with the €138 bag? Is it really worth it?

Yes I store and travel all my demo models in them and IMO they provide good protection although as you say they are not very cheap. 
I asked the factory if we could get the Pelican hard case that Siglent NA offer but it's specific to the US market:
https://siglentna.com/product/hard-shell-carry-case/

Thanks, that is a great tip! It looks like a PELI 1520 or PELI Stom iM2450 should fit perfectly. That is almost same price for a robust flight case compared to a nylon cloth bag!

[Bad_Driver]

Hi John,

short answer to the question regarding SDG2042 (2122) against SDG1032.

I‘m dealing thisdays with my NOS schottky TTLs and a 65 MHz OCXO ( now VC)
to design a by-13-divider with my available ICs to bring it into suitable frequencies to
follow the GSPDO path 

The multisim simulation shows a problem above 50MHz I can not there identify.

For a breadboard design test it‘s helpfull to have the SG that goes easy above 60 MHz.
So this is for me the first advantage of my decision for the better and little bit more expensive SG.
And with the SDS as well 

For testing on breadboard I built these simple probes:

http://nihtila.com/2019/03/16/tip-3-robust-high-bandwidth-passive-diy-probes/

I used selected 1k (0.1%) resistors and than the selectable divider factor (21) of the SDS inputs with 50 ohms
was first time used and appreceated.

btw: do you have a schematics of your MK2 design by the hand?

[Martin72]

Hi,
We should better stay On-Topic..

[Elasia]

I use these

https://www.amazon.com/Milwaukee-Electric-Tool-48-22-8425-Large/dp/B0776MCYM8

With foam such as

https://www.amazon.com/Feldherr-FS100R-Bundle-Customizable-Separate-Bottom/dp/B06W5BM2FN

less than 100 bucks and a better case, i actually stack these and cart around multiple pieces on wheels.. real handy for busting out in the field with an inverter and other test gear

[Labrat101]

I‘m dealing thisdays with my NOS schottky TTLs and a 65 MHz OCXO ( now VC)
to design a by-13-divider with my available ICs to bring it into suitable frequencies to
follow the GSPDO path 

Hi Unless I have miss understood if you want 10Mhz from 65Mhz you need to divide by 6.5
 if you divide by 13 then times which is 5Mhz then double that to get 10Mhz you will also  be changing
the error from the 65mhz ocxo ie if your ocxo is say 100ppb the divide will give you better by 13 .but then x 2
 will put the error out by a long way as there will be losses in such a large divide and then x2 .
 In theory it should work . You will need really good JK flips flops with a high response time on the x 2
Also remember on the VC to use a zero drift op amp to control the voltage control.
 
Have fun
RNS

 

[Bad_Driver]

Hi Martin,
sorry for that detour  but I think the link with the low cost self made probes can be usefull for
some of us working with digital electronics.

@ labrat: I‘m aware of the problems but it is hobby and learning by doing so let‘s see where it brings me to.
And I never expected to run in problems with this by-13-divider 

[Labrat101]

@ labrat: I‘m aware of the problems but it is hobby and learning by doing so let‘s see where it brings me to.
And I never expected to run in problems with this by-13-divider 

I also did a lot of home experiments on this as well . I found some very good reading on the NIST site how the big guns do it and its basic the same principle as the clock divider 32.768hz divide by 16 to give 1hz good for a second
a year . from a low grade xtal.
I also found that some of those cheesinessy GSPDO are Not as accurate as they say. The numbers on the screen look impressive but are they real . The xtals they use are for sure not certified . But for $100 - $200 they are what they are .
 I used there way without GSPDO thinking the standards came before satellite gps . (chicken or the egg came first)

  Have Fun Good luck and it will work  good
RNS

[2N3055]

I‘m dealing thisdays with my NOS schottky TTLs and a 65 MHz OCXO ( now VC)
to design a by-13-divider with my available ICs to bring it into suitable frequencies to
follow the GSPDO path 

Hi Unless I have miss understood if you want 10Mhz from 65Mhz you need to divide by 6.5
 if you divide by 13 then times which is 5Mhz then double that to get 10Mhz you will also  be changing
the error from the 65mhz ocxo ie if your ocxo is say 100ppb the divide will give you better by 13 .but then x 2
 will put the error out by a long way as there will be losses in such a large divide and then x2 .
 In theory it should work . You will need really good JK flips flops with a high response time on the x 2
Also remember on the VC to use a zero drift op amp to control the voltage control.
 
Have fun
RNS

You can divide all you want, error will stay the same in percentage (ppb) just smaller number of Hz (absolute value)...   

[Labrat101]

I‘m dealing thisdays with my NOS schottky TTLs and a 65 MHz OCXO ( now VC)
to design a by-13-divider with my available ICs to bring it into suitable frequencies to
follow the GSPDO path 

Hi Unless I have miss understood if you want 10Mhz from 65Mhz you need to divide by 6.5
 if you divide by 13 then times which is 5Mhz then double that to get 10Mhz you will also  be changing
the error from the 65mhz ocxo ie if your ocxo is say 100ppb the divide will give you better by 13 .but then x 2
 will put the error out by a long way as there will be losses in such a large divide and then x2 .
 In theory it should work . You will need really good JK flips flops with a high response time on the x 2
Also remember on the VC to use a zero drift op amp to control the voltage control.
 
Have fun
RNS

You can divide all you want, error will stay the same in percentage (ppb) just smaller number of Hz (absolute value)...   

Tell Seiko that ..  & NIST also work on that as well . The Error is also divided try it .

[nctnico]

@ labrat: I‘m aware of the problems but it is hobby and learning by doing so let‘s see where it brings me to.
And I never expected to run in problems with this by-13-divider 

I also did a lot of home experiments on this as well . I found some very good reading on the NIST site how the big guns do it and its basic the same principle as the clock divider 32.768hz divide by 16 to give 1hz good for a second
a year . from a low grade xtal.

This is complete and utter nonsense. The error stays the same no matter what kind of dividing you do. And getting such good results from a crystal is only possible through careful charactisation and compensation but you'll still have an aging effect. Furthermore; organisations like NIST aren't using crystals to track UTC but a whole bunch of different clocks (Cesium or better) which then are compared with eachother to figure out which ones are most stable or not to figure out which weighted average between all those clocks provides the most accurate UTC time. And even that wanders around by nanoseconds over a period of several days. Getting this right is a science in itself.

[Johnny B Good]

As for the missing calibration certificate, I still haven't had any reply from Telonic on the matter. I'll let it stew for another day or two before pressing them for an answer. As Tautec said (tnx btw) there's no excuse for withholding this piece of documentation from the end customer.
JBG

If you get stuck let me know.
Flick me your SN# via PM and I can get a PDF copy from the factory.
Still, at some point I'll need your email so to forward it on to you.

 Thank you, I might well take you up on that offer - still no response from Telonic. However, I did get two emails from BG, the first a 'thank you' for the order, the second confirming delivery will be this Monday, Result!  For us Brits at least, the Chinese warehouse ordering option isn't always the cheapest (in this case a combination of delivery charges and the effect of 'promo' pricing precluding the option to cash any BG points in for a cash discount). If you order anything from BG, check out the alternative warehousing options, if any, to make sure you're not paying extra for the privilege of a four week delay in delivery.

 Taking heed of Martin72's advice over keeping topic drift in check (apologies to everyone here), I can offer some initial observations on the subject of this topic thread, namely that my example (the 2104X+) consumes 4W in standby and 55.5W operating (turning channels off or on makes no difference - I haven't tested consumption using the FFT function so far). Also, I noticed that it takes 30 seconds longer to boot up than the 16 seconds boot time of the humble SDS1202X-E after being shutdown from the trace display mode - it takes a few seconds longer if it had been displaying an FFT on the previous shutdown, an effect I'd expect to see with the 2000X+ models when I eventually get to use the FFT features.

@Noreply, I've attached pictures of my hand drawn circuit diagrams for you to peruse. I haven't shown the dc-dc converter details nor labelled which is which but you should be able figure it out.   

JBG

[2N3055]

You can decrease phase noise, but not frequency error..

[Labrat101]

The Error is also divided try it .

Can I assume that you can eliminate the error by continuous dividing?

No , It does work quite like that .. its like 22/7 Pi.Never ends.
 The error will be smaller to 10¹² . in your case
 Forget the percentage which is 1/100 which will always look wrong as its a percentage of the last digit (12)
 I have run some tests on a cheap 24.000 Mhz xtal  Actual 24.00162843
 divided it by 12 though my test setup with 2x HP 5385 (Now calibrated) comes out at 2.000135683
  May not be to Nasa standards but the error did move by 12 .
I also ran the same test with 2 certified 20mhz ocxo  @ 0.02ppb (New) the division by 2  before 20.0000000002 after 10.0000000001 .
 Come to your own conclusion 

[Martin72]

Maybe a mod could separate the offtopic into a new thread - It´s interesting but in my opinion, it´s worth a own thread. 

[nctnico]

The Error is also divided try it .

Can I assume that you can eliminate the error by continuous dividing?

No , It does work quite like that .. its like 22/7 Pi.Never ends.
 The error will be smaller to 10¹² . in your case
 Forget the percentage which is 1/100 which will always look wrong as its a percentage of the last digit (12)
 I have run some tests on a cheap 24.000 Mhz xtal  Actual 24.00162843
 divided it by 12 though my test setup with 2x HP 5385 (Now calibrated) comes out at 2.000135683
  May not be to Nasa standards but the error did move by 12 .
I also ran the same test with 2 certified 20mhz ocxo  @ 0.02ppb (New) the division by 2  before 20.0000000002 after 10.0000000001 .
 Come to your own conclusion 

You are wrong. The relative error (which is the only thing that counts) stays the same.

[jemangedeslolos]

And.....if I'm not mistaken.....tv84's remark was ironic, reasoning by the absurd.

[Bad_Driver]

Thanks to Martin again.
To end this off-topic issue. I‘m aware of all the jitter problems.
If I really need a GPSDO I will buy the Leo Bodnar one with so many good
reviews and not this china crap with old XOs.
My problem was not the design of a divider-per-13, it is the quest to find the working
solution with the ICs on the shelf I have by the hand. (some NOS Tesla Schottky TTL ICs,
and this Tesla dosn‘t belong to Elon Musk, it‘s a Checz manufacturer in the eighties)
It is my hobby not my job 

[Labrat101]

The Error is also divided try it .

Can I assume that you can eliminate the error by continuous dividing?

No , It does work quite like that .. its like 22/7 Pi.Never ends.
 The error will be smaller to 10¹² . in your case
 Forget the percentage which is 1/100 which will always look wrong as its a percentage of the last digit (12)
 I have run some tests on a cheap 24.000 Mhz xtal  Actual 24.00162843
 divided it by 12 though my test setup with 2x HP 5385 (Now calibrated) comes out at 2.000135683
  May not be to Nasa standards but the error did move by 12 .
I also ran the same test with 2 certified 20mhz ocxo  @ 0.02ppb (New) the division by 2  before 20.0000000002 after 10.0000000001 .
 Come to your own conclusion 

You are wrong. The relative error (which is the only thing that counts) stays the same.

  Can you give A true working example of your statement that proves you right.

[tautech]

As for the missing calibration certificate, I still haven't had any reply from Telonic on the matter. I'll let it stew for another day or two before pressing them for an answer. As Tautec said (tnx btw) there's no excuse for withholding this piece of documentation from the end customer.
JBG

If you get stuck let me know.
Flick me your SN# via PM and I can get a PDF copy from the factory.
Still, at some point I'll need your email so to forward it on to you.

 Thank you, I might well take you up on that offer - still no response from Telonic.

No worry, just when you need it.....

.............
Purposely removing OT content..... 

I can offer some initial observations on the subject of this topic thread, namely that my example (the 2104X+) consumes 4W in standby and 55.5W operating (turning channels off or on makes no difference - I haven't tested consumption using the FFT function so far). Also, I noticed that it takes 30 seconds longer to boot up than the 16 seconds boot time of the humble SDS1202X-E after being shutdown from the trace display mode - it takes a few seconds longer if it had been displaying an FFT on the previous shutdown, an effect I'd expect to see with the 2000X+ models when I eventually get to use the FFT features.

Hardly surprising when comparing their overall feature sets.
OTOH a brand spanking new 2kX Plus ex factory unboxed yesterday for company sticker placement and probe compensation seemed somewhat quieter than the only one other I've had through my hands.
Those impressions had it a bit noisier than 5kX which has a smart fan controller but we have had one 2kX+ FW update that may have added fan control however it wasn't mentioned in the release notes. 

Todays trial will have 2kX+ and 5kX running alongside one another for a better judge of noise levels of each......

@Noreply,

And all........see how PM's might work for ya'all 

[nctnico]

The Error is also divided try it .

Can I assume that you can eliminate the error by continuous dividing?

No , It does work quite like that .. its like 22/7 Pi.Never ends.
 The error will be smaller to 10¹² . in your case
 Forget the percentage which is 1/100 which will always look wrong as its a percentage of the last digit (12)
 I have run some tests on a cheap 24.000 Mhz xtal  Actual 24.00162843
 divided it by 12 though my test setup with 2x HP 5385 (Now calibrated) comes out at 2.000135683
  May not be to Nasa standards but the error did move by 12 .
I also ran the same test with 2 certified 20mhz ocxo  @ 0.02ppb (New) the division by 2  before 20.0000000002 after 10.0000000001 .
 Come to your own conclusion 

You are wrong. The relative error (which is the only thing that counts) stays the same.

  Can you give A true working example of your statement that proves you right.

Just calculate the absolute error when dividing the frequencies down to 1 second and use it for a digital clock (to show the time). It doesn't matter if you start with 1GHz +10ppm or 100Hz +10ppm; the resulting clock will run with the same error. In both cases the clock will be ahead by one second every 1.26 days.

[2N3055]

  Can you give A true working example of your statement that proves you right.

So you want us to prove you that elementary school math is correct... That's precious... 

Read this..

https://www.eetimes.com/perfect-timing-performing-clock-division-with-jitter-and-phase-noise-measurements/

And with apologies to everybody, this will be me my last offtopic post here.
If you want a detailed discussion, open a topic; i will join and explain math to you, no problem. Just not here, please..

[Johnny B Good]

Hi John,

short answer to the question regarding SDG2042 (2122) against SDG1032.

I‘m dealing thisdays with my NOS schottky TTLs and a 65 MHz OCXO ( now VC)
to design a by-13-divider with my available ICs to bring it into suitable frequencies to
follow the GSPDO path 

The multisim simulation shows a problem above 50MHz I can not there identify.

For a breadboard design test it‘s helpfull to have the SG that goes easy above 60 MHz.
So this is for me the first advantage of my decision for the better and little bit more expensive SG.
And with the SDS as well 

For testing on breadboard I built these simple probes:

http://nihtila.com/2019/03/16/tip-3-robust-high-bandwidth-passive-diy-probes/

I used selected 1k (0.1%) resistors and than the selectable divider factor (21) of the SDS inputs with 50 ohms
was first time used and appreceated.

btw: do you have a schematics of your MK2 design by the hand?

 I see your need for a function generator with a wider than 65MHz BW but the SDG2000X series max out at 25MHz for square waves so you'll obviously be using the sine wave output for your tests. If you look at the MK I's circuit diagram, you'll note the use of a 74193 to divide by 13. In this case I had to double the 13MHz to 26MHz just so the 2MHz output met the minimum clock input frequency requirement for the final times 5 clock multiplier used to generate the 10MHz locked to the 13MHz OCXO's output.

 Even so, I had to select my 74193 carefully to get this to work at all (none of my HC193s were good enough for this job - believe me, I tested my whole stock of HC193s before resorting to the old skool 74193 - hence the 16 pin dip socket on the board).

 Whereas I've had to double up to 26MHz to satisfy the clock multiplier's minimum input frequency requirement, you're going to have to divide that 65MHz down to a more manageable 32.5MHz before feeding it into your divide by 13 modulo N counter for a 2.5MHz output you can multiply back up to the required 10MHz (I hope you have a reasonable quantity of '193s to pick from  ). After my own experience with the MK I, I'd recommend against dividing a 65MHz OCXO output down to 10MHz for GPSDO use other than as a "Fun Learning Project" (fvsvo "Fun").

 I suspect the issue with the multisim simulation was it recognising the input clock speed limits of the 74S logic family. I tried to look for a datasheet on a 74S193 to check this but I don't think any '193s were ever manufactured in this logic family. Even so, I suspect even a 74S193, if any ever existed, would struggle with such a high frequency input clock as 65MHz, hence my suggestions above.

 Until around 18 months ago, I didn't know that any other type of 10:1 and 20:1 passive probe than the classic "Low impedance" type described in that article even existed. These new fangled High impedance 'scope probes rather disturbed my engineering sensibilities when I first came across them with my first modern DSO purchase  . Anyway, I've gotten over the shock (of the new) now. They're fine for the job as long as you keep in mind that a 200MHz rated Hi-Z 'scope probe is only good for a 10MHz or so bandwidth when switched to 1:1 (x 1 position). 

 Since I'm now expecting to take delivery of a bunch of FTDI232 adapters this Monday, I won't be cannibalising the MK I for its FTDI module this week after all. It is scheduled to be dismantled to recycle its components back to my spares bin - in a week or two's time, just not this week is all.

JBG

[Johnny B Good]

Hi,
We should better stay On-Topic..

 Agreed!

JBG

[Martin72]

Finally, I would appreciate this...