The Siglent SDG2042X Thread

[uncle_bob]

The fan in my unit started making a slight annoying buzzing noise, so I opened it up to see if I could manipulate the fan to get rid of the noise (which I was able to).

While I had it open I noticed it had the Siglent signature rust on the edges of some of the metalwork. It was only slight and am sure it won't be a problem but does show poor attention to detail.

Hi

Swipe it down with some silicone grease and move on ...

(yes it is a bit tacky in a brand new instrument)

Bob

[Break42]

Synchronization between CH1 and CH2 in sweep mode

Hi folks,

just a bit another point. I have on the CH1 sweep between 10.4Mhz and 11.0Mhz, middle 10.7MHz. Sweep time 300ms. On the CH2 I would like to get a markes (short pulses) at 10.4, 10.7 and 11.0 MHz (that is 0%, 50% and 100% of the 300ms sweep time). My question: How to get these synchronus pulses on the CH2?

Regards
Break

[nugglix]

That's simple: you don't.       

If you need to sync, make the channel emit something at the right moment.
Might be a 150ms pulse every 300ms, that's as close you can get.

Cheers

[Break42]

Synchronization between CH1 and CH2 in sweep mode

Thanks nugglix, I will check how can I manage this today evening.

Anyway I think that such feature ("Synchronize a channel with other sweep channel") would be very helpful for HF development.

[uncle_bob]

Synchronization between CH1 and CH2 in sweep mode

Thanks nugglix, I will check how can I manage this today evening.

Anyway I think that such feature ("Synchronize a channel with other sweep channel") would be very helpful for HF development.

Hi

This is sort of why people set up computer driven auto-test systems. The needs of this or that test regime get complicated quickly. If you are testing billions of cell phones ... somebody will make a all in one test set. For the rest of us, that's not going to happen. each of our unique needs does not represent a big enough market. So ... off to the computer and write up some code. Let the computer tell each instrument what to do and when to do it. That's been part of my life for 40+ years. These days with even the "affordable stuff" showing up with standard control ports, it's a lot easier than it was in the 1870's ... errr ... 1970's when I started.

Bob

[markone]

Synchronization between CH1 and CH2 in sweep mode

Thanks nugglix, I will check how can I manage this today evening.

Anyway I think that such feature ("Synchronize a channel with other sweep channel") would be very helpful for HF development.

I'm too playing with sweep function to build a sort of scalar network analyzer and i'm facing a phase problem between the two output channels, for instance if i program the device in the following mode :

CH1 : SWEEPed Sine Waveform 1Mhz to 120Mhz, SWEEP Time 16ms, Trig Out ON
CH2 : PULSE function, PERIOD 4MS (16/4), PULSE Width 200us, Delay 0

at first seems to works ok with a pulse every sweep's quarter in perfect phase, bu if i look closer i can see that channel two position slowly drift back, let's say 8ns every 20 seconds, while the generator trigger out signal is luckily "locked" to the sweep "ramp" with a fixed offset of about 500nS.

If you turn off then on the sweeped channel the moving phase offset will be reset but restarts soon to drift again.

Some screens in attachment to explain the matter. 

[uncle_bob]

at first seems to works ok with a pulse every sweep's quarter in perfect phase, bu if i look closer i can see that channel two position slowly drift back, let's say 8ns every 20 seconds, while the generator trigger out signal is luckily "locked" to the sweep "ramp" with a fixed offset of about 500nS.

Hi

Ok, let's take a step back and ask the question backwards:

If I wanted to have it drift 8 ns / 20 seconds how could I force it to do that? On a 1 GS/s device, 1 ns is one sample in a second. 20 ns in 20 seconds = I'm dropping one sample each second. If I'm below a sample per second ... that's either dropping samples every couple of seconds or it is something else. I'd bet on "something else".

The generator has a time base in it. It is software corrected for frequency offset. That calculation likely drives some math in the DDS that gives you exactly this or that frequency. The arbitrary wave side of the generator's brain knows that 1 billion samples must each go out on a specific clock edge, no drops, no adds no goofs. The two approaches will eventually collide with each other. If the trigger pulse lives in the "arbitrary waveform" side of the brain and the sine wave sweep lives in the "corrected" side .... you get what you see.

I have *absolutely* no clue if that is even close. The only way I can think to test it would be to somehow zero the calibration coefficients. I'm equally clueless to how that might be done.

Bob

[markone]

Ok, let's take a step back and ask the question backwards:

If I wanted to have it drift 8 ns / 20 seconds how could I force it to do that? On a 1 GS/s device, 1 ns is one sample in a second. 20 ns in 20 seconds = I'm dropping one sample each second. If I'm below a sample per second ... that's either dropping samples every couple of seconds or it is something else. I'd bet on "something else".

The generator has a time base in it. It is software corrected for frequency offset. That calculation likely drives some math in the DDS that gives you exactly this or that frequency. The arbitrary wave side of the generator's brain knows that 1 billion samples must each go out on a specific clock edge, no drops, no adds no goofs. The two approaches will eventually collide with each other. If the trigger pulse lives in the "arbitrary waveform" side of the brain and the sine wave sweep lives in the "corrected" side .... you get what you see.

I have *absolutely* no clue if that is even close. The only way I can think to test it would be to somehow zero the calibration coefficients. I'm equally clueless to how that might be done.

Bob

The answer is yes : if i purposely set a "not in ratio" period for the pulse function,  like 4.000001ms with  sweep time = 12ms  (so one sweep time quarter plus 1ns), i can see the two channels drift away as expected.

Notice that the previously reported drift ratio is from a quick & dirty estimation, let's say that is this scale order, i observed it for some minutes, it's not fine smooth but it step's sharply, i could expect also that it has some periodic characteristic, like coming back to zero at some point and restart to drift in fixed cycle.

Another thing that i noticed, that could be usefull to understand how the generator works, is that the sweep function does not have a fixed phase, the signal restart with a different angle every sweep, so you cannot rely on averaged acquisition for the produced sweeped signal.
Days ago i gave up with the second channel pulse time reference,  that was intended to trigger a sampling system placed on the output of an logarithmic amplitude detector, and i'm working only with the generator trigger out signal that so far seemed reliable.

[uncle_bob]

The answer is yes : if i purposely set a "not in ratio" period for the pulse function,  like 4.000001ms with  sweep time = 12ms  (so one sweep time quarter plus 1ns), i can see the two channels see drift away as expected.

Hi

So indeed you have a math function that can demonstrate the same sort of behavior. With finite precision math a discrete hop is not un-expected. I also believe that the giga sample end of the generator is driven from some sort of interpolator off of a 250 mega sample base. That would also give you 4 rather than 1 ns steps. If the base is lower still 125 MHz (as it very likely is under some conditions ) we are right back at your original 8 nanoseconds.

Bob

[markone]

That would also give you 4 rather than 1 ns steps. If the base is lower still 125 MHz (as it very likely is under some conditions ) we are right back at your original 8 nanoseconds.
Bob

Tomorrow i will do some more precise assesment on that, let's say anyway that if the phase error restarts from the very same value with a sweep stop/restart command sequence, as it appears to be, for single shot or short term acquisitions sure it's not a deal breaker.

[tsaG]

Hi,
Is it possible to output an arbitrary wafevorm signal which I created in National Instruments Multisim?

[uncle_bob]

That would also give you 4 rather than 1 ns steps. If the base is lower still 125 MHz (as it very likely is under some conditions ) we are right back at your original 8 nanoseconds.
Bob

Tomorrow i will do some more precise assesment on that, let's say anyway that if the phase error restarts from the very same value with a sweep stop/restart command sequence, as it appears to be, for single shot or short term acquisitions sure it's not a deal breaker.

Hi

Let us know what you find !!

Bob

[markone]

Let us know what you find !!

Bob

Well, my yesterday's estimation was way off ...   actually we have a drift rate of about 2.14ns/s, measure also along 1 hour period, in the form of one 21.4ns sharp drift step every ten second.
The generator trigger output rising edge is always 535ns ahead the actual sweep start, regardelss the sweep settings.

The generator was always set for :

CH1 : SWEEPed Sine Waveform 1Mhz to 120Mhz, SWEEP Time 16ms, Trig Out ON
CH2 : PULSE function, PERIOD 4MS (16/4), PULSE Width 200us, Delay 0

[uncle_bob]

Let us know what you find !!

Bob

Well, my yesterday's estimation was way off ...   actually we have a drift rate of about 2.14ns/s, measure also along 1 hour period, in the form of one 21.4ns sharp drift step every ten second.
The generator trigger output rising edge is always 535ns ahead the actual sweep start, regardelss the sweep settings.

The generator was always set for :

CH1 : SWEEPed Sine Waveform 1Mhz to 120Mhz, SWEEP Time 16ms, Trig Out ON
CH2 : PULSE function, PERIOD 4MS (16/4), PULSE Width 200us, Delay 0

Hi

Another fine set of theories blown away by real data

21.4 ns is a really weird number. It's roughly 46.7 MHz. That does not fit into *any* clock scheme I know of in the generator.

So off to a whole new theory ...

These guys play games with DDS stuff to get a better signal than they could otherwise. The DDS process generally has some sort of phase accumulator. For a sweep, you are constantly messing with that accumulator. I'd bet that you do some rounding in the process. The theory of the day: They sum up the round off error with a good level of accuracy. They put it in every 10 seconds. They got the sign on the number backwards ....

Bob (the guy who hasn't blown a simple math problem for ... errr ... hours).

[markone]

They sum up the round off error with a good level of accuracy. They put it in every 10 seconds. They got the sign on the number backwards ....

A similar trick during fast sweeps does not make a lot of sense, i guess we are watching to a phase error accumulation / compensation from a PLL mechanism implemented inside the FPG, from which are derived the two channels master clocks.

I saw something similar in a erroneously implemented PLL, again inside a FPGA, that had the task to lock some ADCs data samples batches to power line cycle, so IMHO my bet is on that     

[tsaG]

Hi,

I have a problem using the arbitrary function of this generator.

I created a waveform using the EasyWave software.

New -> ticked the SDG2000X box, set it to 8000 samples and 100Hz Frequency

I created this [1] Waveform and uploaded it to the generator. On picture [2] you can see the waveform on the oscilloscope. The odd thing is, that the oscilloscope picture looks exactly like the preview picture on the signal gen! Somehow it doesnt transfer or load the waveform correctly... Or did I do something wrong?

Im using Easywave 1.1.34. The Siglent is connected over Ethernet.

EDIT: I created a Sine wave in easywave, imported it into my signal gen which looks like [3]. Normally it should be +-7V.

EDIT2: A readback of the Sine wave outputs the exact wave of the oscilloscope (and preview on the screen of the siglent) in easywave (see readback screenshot)

EDIT3: It doesnt work as well if I connect the Siglent using USB but it works like a charm if I import the csv file from easywave into the Signal gen. with a USB stick. So it looks like there is something wrong with the easywave transfer (?)

[CustomEngineerer]

tsaG

I had the same issue when I first tried the EasyWave software. I tried sending a file named wave1 to channel 1. I tried to create another file, also named it wave1 and sent it to channel 1. The best I could tell it looked like it was combining the files to make 1 waveform, and just kept adding to it. I was about to give up on the EasyWave software since I didn't really need it at the time, but the next day I decided to try 1 more time. This time I used different names for each waveform file, and sent some to channel 2 and others to channel 1 and it worked like expected every time. I'm not really sure what caused it to start working correctly, but it has ever since. Now whenever I send a waveform to either channel, it always overwrites whatever is already on that channel, and displays exactly what it is supposed to.

Maybe try rebooting your generator, and trying different names for your waveform files (not suggesting you weren't doing this).

[Salas]

The fan in my unit started making a slight annoying buzzing noise, so I opened it up to see if I could manipulate the fan to get rid of the noise (which I was able to).

While I had it open I noticed it had the Siglent signature rust on the edges of some of the metalwork. It was only slight and am sure it won't be a problem but does show poor attention to detail.

Its not a bad quality fan but there could be less air cooling circulation noise in this generator (which mainly comes from the box working like a speaker cabinet)

I don't know if it is calculated for worse case scenario when installed in a rack as it is, but for home lab use maybe a Sunon fan (HA60251V4-999 looks good) with some 3V drop resistor inline could suffice for safe cooling and about half the final audibility. Maybe Siglent can advise us here about which is the max CPU sink's temperature for long term reliability?

[uncle_bob]

The fan in my unit started making a slight annoying buzzing noise, so I opened it up to see if I could manipulate the fan to get rid of the noise (which I was able to).

While I had it open I noticed it had the Siglent signature rust on the edges of some of the metalwork. It was only slight and am sure it won't be a problem but does show poor attention to detail.

Its not a bad quality fan but there could be less air cooling circulation noise in this generator (which mainly comes from the box working like a speaker cabinet)

I don't know if it is calculated for worse case scenario when installed in a rack as it is, but for home lab use maybe a Sunon fan (HA60251V4-999 looks good) with some 3V drop resistor inline could suffice for safe cooling and about half the final audibility. Maybe Siglent can advise us here about which is the max CPU sink's temperature for long term reliability?

Hi

I'd bet the fan was picked on the basis of "this is what we stock for the other 20 instruments". Purchasing guys love that approach. The stock room guys love it as well.

If the heatsink is below 50C, it's a good bet that all is well. If it''s above 70C things may be ok, but life of the device is not what it might be. If it's over 90C, you need more air. All based on the "most chips / most places  / most ratings / most likely" guess. 


Bob

[markone]

The fan in my unit started making a slight annoying buzzing noise, so I opened it up to see if I could manipulate the fan to get rid of the noise (which I was able to).

While I had it open I noticed it had the Siglent signature rust on the edges of some of the metalwork. It was only slight and am sure it won't be a problem but does show poor attention to detail.

Its not a bad quality fan but there could be less air cooling circulation noise in this generator (which mainly comes from the box working like a speaker cabinet)

I don't know if it is calculated for worse case scenario when installed in a rack as it is, but for home lab use maybe a Sunon fan (HA60251V4-999 looks good) with some 3V drop resistor inline could suffice for safe cooling and about half the final audibility. Maybe Siglent can advise us here about which is the max CPU sink's temperature for long term reliability?

No (at least not yet) buzzing noise from mine generator's fan, i can hear a moderate air flux sound that maybe is one-tenth of the of DS1074Z's fan noise, so i can say i feel good with it.

What about the warranty sticker placed on the generator box ?
I guess we have to cut it to get access to instrument's fan and does not seem a smart move.

Anyway, talking about DC fans, i feel we have a very high unit to unit performance/quality variance, ie my original GW Instek scope's ADDA fan (60x60x25 too) is a piece of cra@p, dunno if  it's a fake device or actually ADDA make crap, i wonder why GW Instek uses this crap for its instruments, i swapped it with a 60x60x15 Sunon Maglev with surprisingly better result for 4 Euro 1 Pz street price.

GW Instek also made a mechanical design mistake putting the fan intake side too close to circuit metal shielding case (8-9mm) simply suffocating it, indeed the slimmer version i used leaves 5mm more air space and works waaay better.

Here Siglent seems to have done a much better job, at least with mine 2042X.

[markone]

I'd bet the fan was picked on the basis of "this is what we stock for the other 20 instruments". Purchasing guys love that approach. The stock room guys love it as well.

I think the very same thing, and they are not alone.

Anyway it's a decent solution and looking at where the fan is placed in respect to CPU, its temps should lay on the low side, conversely the PSU is probably working with lower temp's margin.

[uncle_bob]

Hi

There are an enormous number of fans out there. If you can get close to anything that gets used on a normal PC case ... dirt cheap is a typical price.

My preference would always be for a fan that does not need a dropping resistor. Dirt / dust / wear / crud tend to make the poor little motors start harder as time goes by. They actually put "stuff" in some to give it a startup boost and then cut back after it's running. A dropping resistor can degrade that performance. How common "chips in fans" are for non tach readout fans ... no idea.

Bob

[Salas]

I find the 2042X fan and wind noise constantly noticeable and distracting in a quiet room when standing at 1m near. My DS2072 scope with a quieter medium quality Scythe fan emits 1/3 the generator's noise subjectively. The gen sounds more like the original 2072 noise level albeit with a bit less annoying lower pitch hum.

That Sunon Maglev is two wire like the stock 2042X fan and has significantly less nominal noise (13.8dBA vs 26dBA), still good CFM (13.8 or 0.391m3/min), while having a 60000 hours life spec @ 40C. I proposed using it at 9V via a 100 Ohm resistor because there is noise augmentation by the box so using it at its max will probably not bring as a drastic cut in sound level as expected by its dB spec. Voltage range is 4.5 ~ 13.8VDC so no problem starting at 9V. That Sunon can be found for about 5 Euro before shipment. Digikey 259-1618-ND. Typically best PC industry silent fans like Noiseblocker or Noctua will be much dearer.

About the sticker, maybe Siglent will be kind enough to let us know here if it will be safe we change the fan for the above spec without warranty loss

[markone]

My preference would always be for a fan that does not need a dropping resistor.

I agree and there is an additional reason to drive at full power the above mentioned Sunon HA60251V4-999, its static air pressure is already quite low, 0.04 inch-H2O / 1.02 mm-H2O and the fan is quite close to case exhaust holes pattern.

Anyway i would check PSU critical parts temp before to cut down case internal airflow.

Has anyone succeded in reading original fan's brand & P/N ?

[Salas]

PAAD16025SL 0.15A 12VDC. Visible in Dave's video. The question is does the gen need a really strong airflow in home lab use? Or is it just a cautionary measure.