Measuring Signal Generator Output Voltage vs. O-Scope Frequency Limit

[farlander762]

I'm scope dumb...VERY SCOPE DUMB.  I can turn one on and do mostly know how to perform the essential tasks of a scope.  I have been watching reviews of all kinds for various O-Scopes.  I'm probably going to buy a Siglent SDS1204X-E in the semi-near future.  It seems to be on the good end of entry-level scopes and is certainly more machine than I know how to use right now. 

Was looking on Amazon (dangerous) and saw this https://smile.amazon.com/Oscilloscope-Bandwidth-Sampling-Multifunction-Portable/dp/B08KVVXFHV/ref=cm_cr_arp_d_product_top?ie=UTF8.  Yes, I know to the guy with a real scope it's a toy, I like toys.  The reviews, including Dave Jones', pretty much all say it's really 20 - 30MHz and 100MS/s.  Fine.  But I saw this review and he mentioned signal generators dropping their output voltage and then linked to this video .

He's talks about a 3dB drop-off being the defining line for where a scope really performs to.  I don't know why that's the line but I can accept it is.  If the signal generator output also drops off, does this still mean the 20 - 30MHz top end is still valid?  My question isn't really about this particular scope, my question is about whether the reviews are really giving ANY scope a fair chance since some/many/most/all signal generators will have a falling voltage as frequency goes up.  Does he just have cheap/antique signal generators that aren't representative of better/modern ones?  Not badmouthing the man or his review, genuinely curious.

Thanks!
Lance

[TimFox]

Most modern signal generators should have a constant output up to their maximum frequency--there should be some kind of "leveling" circuit if the generator proper has a frequency dependence.
The analog part of the DSO has a frequency response, and that is followed by the sample rate of the digitizer, which should be high enough to see the -3 dB point of the response.
See any electronics textbook for the reason why -3 dB is the conventional measurement of response--for other applications, -1 dB or much higher attenuations are used for the spec.

[farlander762]

OOOOHHHHH!  Now I recall 3Db from audio amplifiers.  Been a long time since I thought about that! 

I guess his generators lack the leveling. 

[bob91343]

Yes -3 dB is half power.

[tautech]

OOOOHHHHH!  Now I recall 3Db from audio amplifiers.  Been a long time since I thought about that! 

I guess his generators lack the leveling.

Not quite.
A sig gen spec is generally well within 1dB out to max frequency and DSO vertical accuracy is generally no better 3% all the way close to rated frequency where after it rolls off generally 10-20% past rated frequency.
So yes the -3dB amplitude frequency is the real test of scope BW where we might use a 1Vp-p sinewave and find the frequency where the scope displays it as 0.707Vp-p which is the scopes true BW.

I'm probably going to buy a Siglent SDS1204X-E in the semi-near future.  It seems to be on the good end of entry-level scopes and is certainly more machine than I know how to use right now. 

Yes a well featured scope however the bit cheaper SDS1104X-E is easier on the budget and as a novice you'd be unlikely to need more than 100 MHz capability.
With the $ saved on the cheaper model consider also getting a SDG1032X to pair with it for the scopes Bode plot feature and ease of transferring captured waveforms to the AWG for it to replicate it.

[farlander762]

With the $ saved on the cheaper model consider also getting a SDG1032X to pair with it for the scopes Bode plot feature and ease of transferring captured waveforms to the AWG for it to replicate it.

I will!  I'm still tossing around the 100 vs. 200 MHz.  You're right, I'll probably never need the extra, but buying the better model just in case has served me well many times.  I probably don't need the 4 ch either but in the Siglent SDS1xxx series, 4 ch adds a bunch of features.  I had an analog 2 ch in the early 2000's, it wasn't a good scope, Channel 2 failed twice while under warranty  , but was fun to play with.  I know a lot more now and find myself less afraid to go probing than back then.  That may have to do with more disposable income and more space for projects...

[David Hess]

Does he just have cheap/antique signal generators that aren't representative of better/modern ones?

Modern ones of that type are no better, and they are the wrong signal source to use.

Most modern signal generators should have a constant output up to their maximum frequency--there should be some kind of "leveling" circuit if the generator proper has a frequency dependence.

He needs to use a leveled signal generator, old or new, but the example source is not.

A sig gen spec is generally well within 1dB out to max frequency and DSO vertical accuracy is generally no better 3% all the way close to rated frequency where after it rolls off generally 10-20% past rated frequency.

And a *leveled* signal source suitable for oscilloscope testing is generally well within 3% or 0.25dB over its entire range.  They are much more precise than a typical leveled RF source.  Vertical accuracy of the oscilloscope itself is irrelevant because leveling is a relative measurement.

An alternative to a leveled source of that quality is to use external means to level the signal source, like an RF sampling voltmeter or separate leveling circuit.  A true sampling system has a very well defined frequency response based on the sampling gate width which can be easily measured with an unleveled source by finding the first null in the frequency response.  Once that is done, the frequency response curve can be calculated to great accuracy.

[TimFox]

If you only want to verify constant output (up to a reasonable frequency), a simple "half-wave voltage doubler" with PIN diodes feeding a DC voltmeter is probably adequate.

[Paul Ed]

@tautech,

May I ask how to acheive this, please?

....
With the $ saved on the cheaper model consider also getting a SDG1032X to pair with it for the scopes Bode plot feature and ease of transferring captured waveforms to the AWG for it to replicate it.

i.e. The steps involved to transfer a waveform from the 'scope to the AWG - a tutorial or URL describing the process would be fantastic!

Secondly, can a SDS2352X-E when paired with a SDG2082X do this same trick as well?

Hope you don't mind me asking, if it's already been explained in a step by step form just point me to that, please.

[David Hess]

Incidentally when making leveled measurements of this type, the deficiencies of the cable and connectors used become significant.  When I used my sampling oscilloscope to check the performance of my 200 MHz leveled signal generator, I could see the effect of using different patch cables on the amplitude.  My leveled signal generator actually says that it was calibrated with a specific cable.

[TimFox]

My Marconi AC voltmeter, good up to 10 MHz:  in the manual, they remind me that the input capacitance, together with the 25 ohm combination of source and cable termination, is worse than the inherent response, compared with the actual voltage across the input BNC

[tautech]

With the $ saved on the cheaper model consider also getting a SDG1032X to pair with it for the scopes Bode plot feature and ease of transferring captured waveforms to the AWG for it to replicate it.

I will!  I'm still tossing around the 100 vs. 200 MHz.  You're right, I'll probably never need the extra, but buying the better model just in case has served me well many times.  I probably don't need the 4 ch either but in the Siglent SDS1xxx series, 4 ch adds a bunch of features.

Better model yes but only in rated BW as they are the same HW but one is SW limited and there's plently of info on this forum on how to overcome that. 
However it's good to see someone can read datasheets properly and spot the different capabilities between SDS1202X-E and its 4ch brothers. 
The SDS1202X-E 200 MHz 2ch X-E is a fine little DSO and we have sold lots of them but the 4ch variants offer so much more capability for owners to grow into.

@tautech,

May I ask how to acheive this, please?

....
With the $ saved on the cheaper model consider also getting a SDG1032X to pair with it for the scopes Bode plot feature and ease of transferring captured waveforms to the AWG for it to replicate it.

i.e. The steps involved to transfer a waveform from the 'scope to the AWG - a tutorial or URL describing the process would be fantastic!

Secondly, can a SDS2352X-E when paired with a SDG2082X do this same trick as well?

Hope you don't mind me asking, if it's already been explained in a step by step form just point me to that, please.

All Siglent DSO's and AWG's permit waveform captures to be downloaded/uploaded for replication via CSV or DAT file types. From the scope you use Save/Recall and save the correct file type to USB then upload it into the AWG.

From the SDG1000X User manual: P80
https://int.siglent.com/upload_file/user/SDG1000X/SDG1000X_UserManual_UM0201X-E01D.pdf
Data File
The SDG1000X can recall the data files in ―*.csv or ―*.dat format from the external memory and transfer them into ―*.bin format then store them in the internal memory. When it is done, the generator will enter the arbitrary
waveform interface automatically.
In addition, users can edit arbitrary waveforms with PC software — EasyWave, download them to the internal memory through remote interface and store them (in ―*.bin format) in the internal memory.

Alternatively if both instruments are connected to a PC with either USB or LAN and EasyWaveX is installed waveforms can be received and sent to either instrument from within the EasyWaveX UI.
https://int.siglent.com/upload_file/zip/Upper_computer/EasyWaveX_1.1.0.13_EN.zip
Unfortunately EasyWaveX needs the somewhat bloated NIVISA Runtime SW package for the connectivity drivers however it's a once only install if you are to use other Easy****X packages.

For multiple LAN connections you need a local LAN switch to connect all instruments to and maybe a USB hub if you're short of USB ports.

[antenna]

With respect to the signal generator in the second video, it looks like it has the exact same screen and buttons as my 60MHz Koolertron signal generator, except in a much different case. When I bought mine (Koolertron), I was disappointed because the output voltage reading on the screen is peak to peak, not RMS, and is only accurately representing the voltage when completely unloaded.  Connecting a 50ohm load cut the unloaded voltage in half, which suggests the output impedance is actually 50ohm, but the output voltage claims were as far to the dishonest side as possible without falsely advertising.  Mine runs about 1.78v RMS into a 50ohm load (3.53v RMS with no load ie. 10v pk-pk).  When I looked at the output voltage on my Picoscope 3206D (235MHz bandwidth) the voltage remained fairly steady until about 54MHz.  Aside for the shady voltage claims and frequency stability being a bit drifty, I was happy with my $60 investment! But be ready for that type of issue if you go looking at the voltages because, at first, I thought I had got a broken unit, turned out to just be dishonest advertising.

[Kleinstein]

With the frequency generators it is normal that the voltage drops with load. The output is with a defined 50 Ohms output impedance. So getting 50% amplitude with 50Ohms load is how it should be.

For the amplitude reading the peak to peak and open circuit number looks better for marketing. Both peak to peak and RMS are commonly used - just make shure to note what type of measurement is used.

The lower frequency generators and those generating different funktions usually don't have a seprate amplitude leveling loop. Expect that more from generators that go beyound 200 MHz.

[Paul Ed]

With the $ saved on the cheaper model consider also getting a SDG1032X to pair with it for the scopes Bode plot feature and ease of transferring captured waveforms to the AWG for it to replicate it.

I will!  I'm still tossing around the 100 vs. 200 MHz.  You're right, I'll probably never need the extra, but buying the better model just in case has served me well many times.  I probably don't need the 4 ch either but in the Siglent SDS1xxx series, 4 ch adds a bunch of features.

Better model yes but only in rated BW as they are the same HW but one is SW limited and there's plently of info on this forum on how to overcome that. 
However it's good to see someone can read datasheets properly and spot the different capabilities between SDS1202X-E and its 4ch brothers. 
The SDS1202X-E 200 MHz 2ch X-E is a fine little DSO and we have sold lots of them but the 4ch variants offer so much more capability for owners to grow into.

@tautech,

May I ask how to acheive this, please?

....
With the $ saved on the cheaper model consider also getting a SDG1032X to pair with it for the scopes Bode plot feature and ease of transferring captured waveforms to the AWG for it to replicate it.

i.e. The steps involved to transfer a waveform from the 'scope to the AWG - a tutorial or URL describing the process would be fantastic!

Secondly, can a SDS2352X-E when paired with a SDG2082X do this same trick as well?

Hope you don't mind me asking, if it's already been explained in a step by step form just point me to that, please.

All Siglent DSO's and AWG's permit waveform captures to be downloaded/uploaded for replication via CSV or DAT file types. From the scope you use Save/Recall and save the correct file type to USB then upload it into the AWG.

From the SDG1000X User manual: P80
https://int.siglent.com/upload_file/user/SDG1000X/SDG1000X_UserManual_UM0201X-E01D.pdf
Data File
The SDG1000X can recall the data files in ―*.csv or ―*.dat format from the external memory and transfer them into ―*.bin format then store them in the internal memory. When it is done, the generator will enter the arbitrary
waveform interface automatically.
In addition, users can edit arbitrary waveforms with PC software — EasyWave, download them to the internal memory through remote interface and store them (in ―*.bin format) in the internal memory.

Alternatively if both instruments are connected to a PC with either USB or LAN and EasyWaveX is installed waveforms can be received and sent to either instrument from within the EasyWaveX UI.
https://int.siglent.com/upload_file/zip/Upper_computer/EasyWaveX_1.1.0.13_EN.zip
Unfortunately EasyWaveX needs the somewhat bloated NIVISA Runtime SW package for the connectivity drivers however it's a once only install if you are to use other Easy****X packages.

For multiple LAN connections you need a local LAN switch to connect all instruments to and maybe a USB hub if you're short of USB ports.

Fantastic, thank you for making that clear, it just hadn't occured to me that it would be possible let alone work quite so easily
(Oh, and we don't use (any) MS-Windows here, and so hadn't looked in detail at EasyWave/EasyWaveX as we cannot easily run that software)

Like the comment about 2 vs 4 channel 'scopes too, more food for thought 

[tautech]

Like the comment about 2 vs 4 channel 'scopes too, more food for thought 

That doesn't apply to your SDS2352X-E though as it has the same feature set as the 4ch X-E's.

The webserver for it could offer some useful capability and it works with any browser and if you really know your stuff with IP's and router firewalls you can use it to operate your scope from 1000's of miles away.

[farlander762]

I've done a lot of reading and watching, learned a lot.  Some people swear that there is no reason for 4 channels, other people have stated how they wished for 4 or use 4 all the time...

One thing about the Siglent 4 ch, is that it then has 2 ea 1GS/s samplers, encoders, whatevers...the term is escaping me at the moment.  If I use channels 1/3 (or 2/4) I get 2GS/s worth of samples.  Seems sufficient.  Many of the others have 1 or 2GS/s for the whole rig.  If 2, great; if just 1 and it has to be divided by 4 channels, that gets to be what is considered low.  Also, 2 different units appears to be redundancy, I like redundancy.

Yes, I do read specs.  Don't really have to have a full grasp of the details to be able to compare the "industry standard" specifications of a given device.

[mawyatt]

Scope channels are like DMM digits, the more the better 

Also related, is having a Logic Analyzer as part of the scope allows one to view analog waveforms & digital "bits" together. Is is very handy for working with DACs and ADCs. 

Best,

[tautech]

I've done a lot of reading and watching, learned a lot.  Some people swear that there is no reason for 4 channels, other people have stated how they wished for 4 or use 4 all the time...

One thing about the Siglent 4 ch, is that it then has 2 ea 1GS/s samplers, encoders, whatevers...the term is escaping me at the moment.  If I use channels 1/3 (or 2/4) I get 2GS/s worth of samples.  Seems sufficient.  Many of the others have 1 or 2GS/s for the whole rig.  If 2, great; if just 1 and it has to be divided by 4 channels, that gets to be what is considered low.  Also, 2 different units appears to be redundancy, I like redundancy.

Yes, I do read specs.  Don't really have to have a full grasp of the details to be able to compare the "industry standard" specifications of a given device.

No, you only get the max sampling rate of whatever ADC is used. Typically in this industry a 1000 series scope has 1 GSa/s and 2000 series units have 2 GSa/s however as you've noticed the important spec is how many ADC's are used as to not divide down the sampling rate to have insufficient sampling/ch at higher frequencies and invite aliasing.

The other thing to consider is memory depth which again is generally divided down by the # of channels in use and this again can be managed by using 2 ADC's each with their own memory support therefore for a 4ch unit with dual ADC's the memory depth only gets halved instead of quartered as in single ADC scopes.

But this all comes at an additional cost, eg; SDS1104X-U vs SDS1104X-E $399 vs $499.....however there are some feature differences in that comparison.

[farlander762]

No, you only get the max sampling rate of whatever ADC is used. Typically in this industry a 1000 series scope has 1 GSa/s and 2000 series units have 2 GSa/s however as you've noticed the important spec is how many ADC's are used as to not divide down the sampling rate to have insufficient sampling/ch at higher frequencies and invite aliasing.

Yup, I wasn't clear in my statement.  That is what I meant.  It can have 2 channels each running at 1 GS/s due to the dual ADC's if selecting channels that are on different ADC's.  Most of the 1GS/s scopes don't have dual ADC's.  The Siglent does and that seems attractive to me.