The Siglent SDG1000 and SDG800 thread

[rf-loop]

Chip yield is something different because the process is hard to control. You can't compare that to a complete electronic circuit.

True but also in analog circuits tens of components random values (inside tolerances) produce differencies. There can use more tight specified components and so on but still variations.
In cheap equipment can not use expensive narrow tolerances componets.  It only need accept that products are more or less different and some kind of compromise. Example I have seen differencies in individual SDG's. (same model) and some is better with jitter, some is better with level accuracy and some differencies in rise times and some other differencies. Some clock have more phase noise, some have more temp sensitive and so on. All are different. But inside specifications. (mostly)

Then there have been some "fun" (unexpected) cases. One example here:
https://www.eevblog.com/forum/testgear/siglent-sdg1020-square-signal-%27dirty%27/msg224941/#msg224941

This  amplifier (b) is Texas Instruments OPA695 (it is used in some units).
It need change to Intersil EL5166ISZ
(also Siglent have used CLC1606 in some units and this also looks like ok
EDIT/ADD: in problem case also this need change to Intersil EL5166ISZ  )

[Tepe]

Firmware
Reportedly, Siglent's firmware versions always end in an odd number.

At the time of this writing, the latest SDG1000 firmware update available on http://siglenteu.com/gjjrj.aspx?id=15
is called 1.01.01.36R1. I haven't seen an Rn suffix before.

[ivan747]

Firmware
Reportedly, Siglent's firmware versions always end in an odd number.

At the time of this writing, the latest SDG1000 firmware update available on http://siglenteu.com/gjjrj.aspx?id=15
is called 1.01.01.36R1. I haven't seen an Rn suffix before.

Yes, this confuses me as well. Any other reports? RF-loop, I think you know about this.

[aviphysics]

I wouldn't be the slightest bit surprised if they used lower spec parts in the lower frequency models. There are plenty of parts that are pin compatible with much higher spec parts. Would also not be surprised if they used the same spec parts in all models, and just software limited the unit. I seriously doubt that they are binning. If the design is at all reasonable, it doesn't seem like you would expect that some parts could reach 50 Mhz and others would only reach 5 Mhz. That would be a ridiculous amount of variation.

[pickle9000]

Exchanging chips to reduce bandwidth is a valid approach if you have no intention of selling upgrade (soft) options. Very effective compared to depopulating a board. If a chip(s) is missing you only need to find out what it is and reinstall. If you need to remove a chip (and change some filter items for example) it becomes more of an issue.

If you look at the Rigol 1054z the bits missing on the main board are for the logic analyzer stuff and a panel connector to tie it in. They also have a set of resistors that id the board, that is probably for the logic analyzer and sig gen. I expect it won't take long before the logic analyzer stuff is hacked. A few chips, power reg, connector assy and figure out the board id.

Anyway Siglent's approach to the issue is a good one, at least as good as you can get. Interesting subject.

 

[rf-loop]

Firmware
Reportedly, Siglent's firmware versions always end in an odd number.

At the time of this writing, the latest SDG1000 firmware update available on http://siglenteu.com/gjjrj.aspx?id=15
is called 1.01.01.36R1. I haven't seen an Rn suffix before.

Yes, this confuses me as well. Any other reports? RF-loop, I think you know about this.

I do not know where from this kind of beliefs are coming but:
There have been and is currently FW version ending with odd and even  numbers and there is and have beeen also Rn
There have been also odd andd even ending FW versions also for SDG1000 series.
current public list for many Siglent equipments:

Example for SDG800:
1.08.01.06 
1.08.01.07
1.08.01.08

Other equipments currently
1.01.01.03R1
1.01.01.06R2
1.01.01.37.2
1.08.01.07
1.01.01.09
1.01.01.36R1
1.08.01.06
5.06.02.19
5.08.02.30
3.01.02.01
1.01.01.35.01
1.01.01.31
5.01.01.10
5.06.02.16
5.01.02.13
5.01.02.22
5.03.02.12

[Jamieson]

Does anyone know if the 800 series has a fan?

Also, what's the deal with the giant rubber bumpers on function generators?  Are manufacturers assuming people are throwing them in the back of pickup trucks? 

[ivan747]

Also, what's the deal with the giant rubber bumpers on function generators?  Are manufacturers assuming people are throwing them in the back of pickup trucks? 

They grip to the bench more than a Formula 1 tire and it looks cool  O0 I like them.

[rf-loop]

Does anyone know if the 800 series has a fan?

I believe not. but I know sure that lowest model, SDG805,  do not have fan. (as also SDG1010 and SDG1005 do not have)

(I have one SDG805 for analyze,  disassemble, assemble and after then use for demo/promotion purposes if it is still ok after all handling and proof tests. Time when I buy it, all other models was sold out and I did not want wait. Later I will do some tests and peek inside (because I need study also how to repair/service these models if need and also I do not give my warranty extensions over factory warranty if I have not analyzed HW building quality and construction and if it pass, then these units what have been in my own burn in test phase and passed I will give (local inside Fi)) warranty extension. So, after some time I know more... )

[Jamieson]

Function Generator -- Big Bumpers Extreme Edition!

[Siglent America]

Function Generator -- Big Bumpers Extreme Edition!

Yes, that is our off-road function generator 
Steve

[Jamieson]

Does anyone have a teardown of the SDG800 series? 

Is the chassis rust still an issue on recent Siglent function generators?  (Dave's teardown video #497 is about 1.5 years old.)

thanks,
jamieson

[rf-loop]

Does anyone have a teardown of the SDG800 series? 

Is the chassis rust still an issue on recent Siglent function generators?  (Dave's teardown video #497 is about 1.5 years old.)

thanks,
jamieson

 I can pick-up negative things from what ever equipment from what ever manufacturer just if I want. I have seen so terrible crap things made by HP, Tek, Philips, ChingChangChung, Fluke, R&S and so on...  . Just if want I can find these and today even more that 60's or 70's or 80's..

Also if same negative thing is in other brand, rewiever can be blind ("I did not see it") and override whole thing. This looks more like normal practice.

Do you mean something like this?
(this image is not from Siglent but from other and brand new "high-end" equipment)

This kind of rust color is not any kind of issue in this kind of position. There is not "Rust issue" in Siglent equipments any more than in other this kind of equipments. Also I can tell that outer box is aluminium and/or plastic.  Of course if rewiever bias (intentional or unintentional) is for finding negative things so that peoples stay more on the channel. Media business is...

[rf-loop]

Does anyone have a teardown of the SDG800 series? 

Yes, here:
https://www.eevblog.com/forum/testgear/teardown-of-siglent%27s-sdg805-arbitrary-waveform-generator/

[ivan747]

Does anyone have a teardown of the SDG800 series? 

Yes, here:
https://www.eevblog.com/forum/testgear/teardown-of-siglent%27s-sdg805-arbitrary-waveform-generator/

-Added to the list 

[rf-loop]

Something with SDG805. 

This is 5MHz model in SDG800 series.

In all pictures signal coming from SDG805 and output level is same  exept last image where is modulation (no any other reason but this image is from other test set)

How it go over 5MHz? Is it hacked?   
(no explanations   )

sidenote: 20kHz is only as reference for level and also because my Keithley 2015THD measurement is Limited to 20kHz.
0.005% THD is nice for cheap function generator.












Then break the 5 MHz great wall.








image with 20MHz accidentally loosed, but signal drop is less than 1dB related to 15MHz image))

So, for 20MHz I will show image from other test set with SDG805
20MHz carrier amplitude modulated with 3kHz with 50% mod depth.


Later play with around 7-8ns rising and falling times with SDG805.

[rf-loop]

About SDG805  rising and falling times.
Here two pictures where is ramp start and end edge.




Ramp down, starting edge rise. (first ?8ns rising edge up and after then ramp fall slowly down)




Ramp up, ending edge fall. (first ramp rise slowly up and then this ?8ns falling edge down)

[rf-loop]

It need note that SDG800 series data sheet (DQS02008 --E02A) have mistake.

1. Output level: There is only 50 ohm load levels.  Hi-Z load levels are double. This information is missing.

2. Pulse function (SDG805):  Minimum is 48ns.  (16ns information is mistake)
(I do not know with my own eyes SDG810 or SDG830)

---------------------------
Square wave function:

SDG800 series Square wave is very different if compare with SDG1000 series.
With low frequencies SDG800 have much less cycle to cycle time jitter.

First just 5MHz Square  (delay one cycle)
Then 1kHz Square rising edge and here delay 1ms (delay agen one cycle)


5MHz One cycle delay (200ns).
30s persistence


5MHz 1000000ns delay. (1ms)
30s persistence




1kHz One cycle delay  (1000000ns delay)
In image 1 horizontal div is 20ppm from 1ms time delay.
30s persistence

It need note that 1ms delay time jitter is perhaps coming mainly from system clock. (this same "wow and flutter" can see with all waveforms and it is around same.  Other jitter is is 8ns systematic jitter with ramp, arb and pulse what is coming it working principle fundamentals (125MSa/s)  In measurement there is of course summed also oscilloscope possible time jitter.
With 1ms time 1ns is 1ppm. (0.0001%)

Later more frequency and under 48ns width pulses etc. ("Breaking" SDG805   5MHz limit)

[ivan747]

About SDG805  rising and falling times.
Here two pictures where is ramp start and end edge.



Ramp down, starting edge rise. (first ?8ns rising edge up and after then ramp fall slowly down)



Ramp up, ending edge fall. (first ramp rise slowly up and then this ?8ns falling edge down)

As far as I can tell, the edge is not actually measured in those pictures, and the timebase is too slow to make it easy and accurate to measure it visually. So, yes, we can see it doesn't ring (mostly because of good termination, not the generator itself, which I am sure it's good), but we cannot see how fast it actually rises.

[ivan747]

How it go over 5MHz? Is it hacked?   
(no explanations   )

  Well, it's a nice experiment for you, but not much for everyone else.

[ivan747]

It need note that SDG800 series data sheet (DQS02008 --E02A) have mistake.

1. Output level: There is only 50 ohm load levels.  Hi-Z load levels are double. This information is missing.

2. Pulse function (SDG805):  Minimum is 48ns.  (16ns information is mistake)
(I do not know with my own eyes SDG810 or SDG830)

---------------------------
Square wave function:

SDG800 series Square wave is very different if compare with SDG1000 series.
With low frequencies SDG800 have much less cycle to cycle time jitter.

First just 5MHz Square  (delay one cycle)
Then 1kHz Square rising edge and here delay 1ms (delay agen one cycle)

5MHz One cycle delay (200ns).
30s persistence

5MHz 1000000ns delay. (1ms)
30s persistence


1kHz One cycle delay  (1000000ns delay)
In image 1 horizontal div is 20ppm from 1ms time delay.
30s persistence

It need note that 1ms delay time jitter is perhaps coming mainly from system clock. (this same "wow and flutter" can see with all waveforms and it is around same.  Other jitter is is 8ns systematic jitter with ramp, arb and pulse what is coming it working principle fundamentals (125MSa/s)  In measurement there is of course summed also oscilloscope possible time jitter.
With 1ms time 1ns is 1ppm. (0.0001%)

Later more frequency and under 48ns width pulses etc. ("Breaking" SDG805   5MHz limit)

Thanks for that info.

[rf-loop]

About SDG805  rising and falling times.
Here two pictures where is ramp start and end edge.



Ramp down, starting edge rise. (first ?8ns rising edge up and after then ramp fall slowly down)



Ramp up, ending edge fall. (first ramp rise slowly up and then this ?8ns falling edge down)

As far as I can tell, the edge is not actually measured in those pictures, and the timebase is too slow to make it easy and accurate to measure it visually. So, yes, we can see it doesn't ring (mostly because of good termination, not the generator itself, which I am sure it's good), but we cannot see how fast it actually rises.

Edge is not there measured but with enough accuracy my eyes can see it is around 8ns what is enough accurate to tell in this case.

Top and bottom "ringing" (most part of it) come from SDG800 (internal system filter what is compromise between many things) . (this kind of risetime do not need any special transfer line matching more than good quality coax and oscilloscope input 50ohm "impedance" what really is not at all 50ohm impedance exept with DC. But still better than nothing and for this pulse shape well enough.  Situation is very different if we have fast rising edge what really have high amount of harmonics, and more than scope can handle. But here we have highly lp filtered signal (with reasons)

Here two images for better estimate more accurate risetime if it is needed.
(note delays - 0ne cycle for look cycle - cycle time jitter. This 8ns "jumping" and how it is distributed statistically depends frequency and how it need adjust it. (because this is "simple machine"))

[ivan747]

Thanks for the valuable info. A lot of things are missing for the SDG800 series, so thank you 

[rf-loop]

How it go over 5MHz? Is it hacked?   
(no explanations   )

  Well, it's a nice experiment for you, but not much for everyone else.

Now some explanation..

It is for everyone. It need only "hack" factory shipping carton and take SDG out and next step is RTFM and use.

Arb memory lenght is 16384 points as told in manual and specifications.
Arb max frequency setting is 5MHz. Nowhere is told Arb produced maximum frequenxcy is 5MHz.

IF Arb frequency setting is 5MHz and Arb memory contain 1 cycle of sinewave output is 5MHz sinewave.

No one have told that Arb memory can contain only one cycle. It is for free waveforms - arbitrary waveforms.  If this Arb memory waveform  is 2 cycle of sinewave then 5MHz arb freq setting give out 10MHz.

For this, attached 2Xsine.csv.txt
For "first time" users: Just download it, do not make any changes but rename it: 2Xsine.csv. Save it it to USB stick and reaad it to SDG Arb memory. Set Arb freq 5MHz and you get 10MHz sinewave out.


But when think and do own waveforms  it need understand what it means it have 125MSa/s speed!
It means 8ns steps.
5MHz one cycle is 200ns!
Arb memory is 16384 (14bit vertical value) steps.
Using 5MHz setting for Arb frequency means that it need handle whole Arb memory lenght using 200ns. In 200ns time it can read only 25 points.

It can think inverse. If we have oscilloscope what have 125MSa/s sampling speed and 5MHz one cycle. There is now 25 samples per cycle.  If we have there 5 cycles it means there is 5 samples per cycle. If there is 10 cycle, it can take only 2.5 sample per cycle. With well working Sin(x)/x it still can draw this sine.  Same fundamental principles to both direction, in and out.  We can (simplified) imagine that in signal generator Arb memory we have now "sigal" what generator is sampling and now, it do not show this result to screen... because it is not dso, it send result to output because it is signal generator..   In this process there is filtering and also need solve frequency adjustment. (125MSa/s is fixed in this simple machine principle)

In reality, as long as the rules of Nyquist are not violated, an oscilloscope can reconstruct a user’s signal identically. This reconstruction process is often referred to as sin(x)/x interpolation. Whether the sample rate is 25x the Nyquist frequency, or 2.5x the Nyquist frequency, interpolation can be used to reproduce the waveform exactly as it appeared at the oscilloscopes input connector, removing all doubt about a signal’s behaviour between samples.

And just this quoted Agilent paper (or other basic fundamentals) we can read inverse. If it can not work inverse - (from group of sample points to signal generator output in this case), it is bullshit theory. No, it is not bullshit, it works just forward and backward.

Using these informations and own skills and experience - with fundamental limits - you can do what you want with also SDG800. Just out from box. Only what need Hack is first this factory carton and power it up.

Attached 2Xsine.csv.txt
(it is just original .csv after you rename it (remove   .txt)

[tautech]

How it go over 5MHz? Is it hacked?   

It is for everyone. It need hack factory shipping carton take SDG out and next step is RTFM and use.

Arb memory lenght is 16384 points as told in manual and specifications.
Arb max frequency setting is 5MHz. Nowhere is told Arb produced maximum frequenxcy is 5MHz.

IF Arb frequency setting is 5MHz and Arb memory contain 1 cycle of sinewave output is 5MHz sinewave.

No one have told that Arb memory can contain only one cycle. It is for free waveforms - arbitrary waveforms.  If this Arb memory waveform  is 2 cycle of sinewave then 5MHz arb freq setting give out 10MHz.

Using these informations and own skills and experience - with fundamental limits - you can do what you want with also SDG800. Just out from box. Only what need Hack is first this factory carton and power it up.

Very nice, I'd better get some in stock.