Is my Siglent SDG1020 Square wave in spec? [solved]

[SFPL]

Hi everyone,

New member of the EEVblog here.   

I've been crawling the forums for a while and finally convinced myself into buying a Siglent SDG1020 to pair with my Philips PM3055.

It's a great device for the price, but I noticed one thing that I'm unsure whether it is to be expected or not (and I didn't find anyone else talking about it here by using the search function):

If I understood correctly, the SDG1020 declares to be able to generate a 20MHz square wave, with a 12ns rise and fall time but, while testing it, I noticed that the square wave begins to lower in quality above 5MHz; and the rise and fall time (at 10%-90%) are consistently at 20ns and not 12ns... (I will attach the pictures when possible).

Furthermore, both my oscilloscope and my V&A VA18B (low quality, I know  ) multimeter show that the frequency generated isn't equal to what is dialed on the Siglent by 0.5% (for example a 2kHz sinewave is measured to be approximately 1.99kHz).

Just to be clear, my oscilloscope is equipped with a 60MHz Lafayette probe (at x1 attenuation) but hasn't been calibrated in 10 years.

So, is the SDG1020 working as expected (and my analog oscilloscope showing its age) or is something wrong happening?

Thanks in advance!

[Macbeth]

Yes, if looking on a low bandwidth scope your squares will look closer to sines. Don't forget that a 20MHz square wave is made of the fundamental 20MHz sine wave and all its odd harmonics, so just to get the first harmonic you need a 60MHz bandwidth minimum. It isn't going to look very square without a lot of the other harmonics, 100MHz, 140MHz... etc...

[Macbeth]

60MHz Lafayette probe (at x1 attenuation) but hasn't been calibrated in 10 years.

x1 ... isn't that usually restricted to 6MHz, not 60MHz?

[TheSteve]

The frequency likely is off a little as Siglent uses a regular crystal for its time base in that model. I believe there is a DIY upgrade floating around to upgrade it to something with a little more stability.

[tautech]

Welcome to the forum.

AFAIK there have been no reported problems of this nature with the SDG1000 series.
Risetime is stated as 12ns or better for squarewave, however in pulse mode the risetime is 7ns or better.

The fundamental frequency of the leading edge is likely faster than your equipment can resolve, remember the probe and the scope have ristimes too.

Signal path termination also plays a part in results you might see and good examples of proper termination are shown/described in this thread:
https://www.eevblog.com/forum/testgear/show-us-your-square-wave/

And yes, pics are helpful for members to help.

[Macbeth]

The frequency likely is off a little as Siglent uses a regular crystal for its time base in that model. I believe there is a DIY upgrade floating around to upgrade it to something with a little more stability.

Yes, it is possible to upgrade the Siglent with an OCXO. It is also possible to use the 10MHz clock input to synch with all your lab instruments.

But somehow I think I would trust the SDG over a 4 digit multimeters "frequency counter", seeing as the primary job of a sig gen is generating frequencies, while a minor job on a DMM is a very poor low resolution frequency counter feature. Unless the DMM has an OCXO or GPSDO or something like that? 

[tautech]

The frequency likely is off a little as Siglent uses a regular crystal for its time base in that model. I believe there is a DIY upgrade floating around to upgrade it to something with a little more stability.

Yes, it is possible to upgrade the Siglent with an OCXO. It is also possible to use the 10MHz clock input to synch with all your lab instruments.

But somehow I think I would trust the SDG over a 4 digit multimeters "frequency counter", seeing as the primary job of a sig gen is generating frequencies, while a minor job on a DMM is a very poor low resolution frequency counter feature. Unless the DMM has an OCXO or GPSDO or something like that? 

+1
With my stock standard SDG1010 my 300 MHz DSO counter registers spot on frequencies, accurate enough for me but maybe not the frequency/time/volt nuts. 

[TheSteve]

I have demo'd an SDG1025 and it was off frequency, but not enough to be of a big concern(still bugged me though). Just throwing it out there as it is certainly possible his SDG1020 is off frequency. It is just as possible his other gear is off frequency of course too.

[Macbeth]

It is just as possible his other gear is off frequency of course too.

It isn't just possible, it is expected. I mean trying to attribute 1.99 vs 2.00 kHz - Even a cal lab HP3458A 8.5 digit can't guarantee its last digit. Most multimeter specs will include a +/- last count or range accuracy as well as the measurement accuracy . Those that don't, well, say no more...

[SFPL]

Thank you for the replies! You are all absolutely right.

The probe at 1x attenuation has a 23.3 ns rise time in the specs, which is consistent with what was shown on the PM3055; making *it* the bottleneck.

Changing to 10x brought the rise time down to 5.8 ns and everything is as I expected it to be.

What a rookie mistake 

Anyway cheers c(_)

(Tried to upload pics, but they're too large. I'll add them another day as a visual aid for anyone who committed the same mistake I did)

[Macbeth]

(Tried to upload pics, but they're too large. I'll add them another day as a visual aid for anyone who committed the same mistake I did)

Do what I do - use the Microsoft snipping tool to crop whatever is on your screen you like, then copy it. Use imgur.com to ctrl-v paste it for upload and then get all the URLs you could ever need

Only if you have something worth blowing up then upload it to imgur as a file and link it. Limit is much bigger than an eevblog attachment.

[tautech]

Everyone has their own method to compress pics for posting.

I use MS Office Pic Manager and Compress pics depending on the resolution required.
For normal "show & tell", just a few Kb's is all that's required, but where good resolution is required for members to download and enlarge, several hundred Kb's is best.

Often you may want to post a pic again and an EEVblog folder on your PC is handy to find pics again.

[SFPL]

Here are the pictures that I promised to upload.

Cheers c(_)

[tautech]

Here are the pictures that I promised to upload.

Risetimes would have been better but thanks anyway.

[SFPL]

Here are the pictures that I promised to upload.

Risetimes would have been better but thanks anyway.

Ok, will do that the next time I'll have chance, tautech 

Out of curiosity, aside from the obvious ease-of-comparison with the specs, is there any other reason to give more importance to the rise time? Isn't the time constant the same?

Thanks

[tautech]

Out of curiosity, aside from the obvious ease-of-comparison with the specs, is there any other reason to give more importance to the rise time? Isn't the time constant the same?

As much circuitry relies on fast risetimes, risetime is the spec most often valued and the hardest to get fast.

Therefore this is what most wish to see.
Often fall times are not as crucial and can be often much slower than risetime, but every Sig gen and circuit is different.
When wanting accurate timing from a falling signal, of course the reverse is true.