EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: DL8RI on February 14, 2014, 11:13:11 am
-
Hi,
while I was testing my new Fluke-Scope, I came around a nice example why an analog scope is still quite important.
It's not about aliasing or stuff, but about seeing a error in a sine wave...
First let's look on my Hantek-DSO:
(http://abload.de/thumb/hantek_rawppjps.jpg) (http://abload.de/image.php?img=hantek_rawppjps.jpg)
Looks like a sine, doesn't it? But this Hantek is more a Noise-Generator... let's average a bit:
(http://abload.de/thumb/hantek_avjokh8.jpg) (http://abload.de/image.php?img=hantek_avjokh8.jpg)
Still looks good. But hey... it's a little portable scope and a rather cheap/crappy one...
So let's Use a Fluke 3370B:
(http://abload.de/thumb/fluke_rawaijeg.jpg) (http://abload.de/image.php?img=fluke_rawaijeg.jpg)
Less niose, and still looks good. Just for a check, with average:
(http://abload.de/thumb/fluke_avuxj29.jpg) (http://abload.de/image.php?img=fluke_avuxj29.jpg)
Maybe one could guess the distortion right now? Maybe one could not see it. I honestly didn't until i switched it over into analog mode:
(http://abload.de/thumb/fluke_analogqdk9s.jpg) (http://abload.de/image.php?img=fluke_analogqdk9s.jpg) (http://abload.de/thumb/fluke_detail3kk2u.jpg) (http://abload.de/image.php?img=fluke_detail3kk2u.jpg)
If you now know what too look for, you will clearly see it on the other pictures.
In reality the difference is bigger, because the digital screen is always wobbeling a bit and therefore hiding the distortion.
If you (for example) look at the averaged sine on the Fluke. The two maximums of the sine. One looks nearly perfect and one seems to show the distortion. So the Problem may be "averaged away".
Probably a modern, high-quality DPO will do better than this Hantek. But I don't own one to test it. Maybe if I win the MDO from the Tek-Advertising ;)
So for the conclusion:
Know the limits of a DSO, there is still a Place for the old CRO :) Even on some really old Hameg-Scope this distortion could be seen in one second, without using any average or whatever.
I think one should keep both, an old analog and a modern digital scope. And never trust them. ;)
-
if you need precise sinus waves, perhaps a distortion meter would do a better measurement system than any scope ?
but yes analog scopes still have a role to play.
-
I think you pretty much summed it up already: know the limitations of the tools you use.
-
Hi,
if you need precise sinus waves, perhaps a distortion meter would do a better measurement system than any scope ?
of course, I would never recommend (and use) a scope as a Distortion-Meter. If I can see a distortion on the screen it may be already at 1% THD, especially a high K2 can hide up to "enormous" values. Even when using a FFT, with 8Bit the scope is nearly deaf.
I just came around this phenomena when learning about the functions of this Fluke-Scope. When I randomly switched into analog-mode I immediately saw this strange distortion, I did not see using the DSO-Mode. It's just a good example about "being careful with measurement results" and I thought it might be nice to share :)
-
I miss my analogue scopes, you don't get the aliasing problem at any timebase setting. I miss the analogue display ability.
Analog aren't magic though, there are advantages to digital, they can display stuff you can easily miss on an analog scope too, so it's a two way street:
EEVblog #442 - Analog Vs Digital Oscilloscope Noise (https://www.youtube.com/watch?v=ImyUB3_n9fw#ws)
-
Hi Dave,
so it's a two way street
yes. And you showed a good example why a DSO is very useful too.
In general I'd say:
-An analog scope is good when you don't want to get "distracted" by (more or less) rare/single-shot signals or "noise". In my example this is masking the problem with the Generator. Unless you do some "tricks" (like ultra-long average).
-An DSO is good when you want to see this rare/single-shot/"noise" just like your example, where the switching-noise is masked by the "infinite" update-rate and low persistence of the CRO. There you have to do "tricks" with your CRO to get it.
One should always cross-check.
Not talking about stuff like Roll-Mode and so on... there it is clear ;)
-
Yup, with an 8-bit DSO your SNR/noise floor will be pretty 'meh' I'm not up fully on the math/actual noise numbers/calculations, but 10*ln(2^N) for 8 bits is only 55 dB, versus 110 dB for 16 bits. However I do know that your effective number of bits (ENOB) will be less than the full count on the DAC, so lets say 7 bits for a 8 bit scope, and 14.5 bits on a 16 bit system, gives 48 db and 100 dB, respectively. I'd expect even less for real world applications because of A) mathy stuff about digitization theory I'm not up on, B) Its real world, of course there is going to be noise and other errors causing problems.
You can hide a lot of stuff under a 50dB noise floor. Per http://www.sengpielaudio.com/calculator-thd.htm (http://www.sengpielaudio.com/calculator-thd.htm) you can get 0.3% distortion at -50 dB, and 1% at -40 dB So yes, its completely feasible that an 8 bit o-scope in the real world can not even see 1% distortion in a sine wave.
Since the noise floor/ENOB/SNR could vary greatly by the front end on a scope, it also makes sense that a with low end/entry level DSO vs a very nice, but very old analog scope, that the analog scope could have a higher SNR.
Ive worked with a data acquisition system which did 2MSps @ 16 bits over a large number of channels with a precision filter which did signal conditioning for it. The filters and ADC's were in seperate rack-mount chassis, with shielded differential pairs between them. The filter system had cards ~6" tall, and 24" long with only 2 channels per card, and it was stacked with mezzanine boards. And a ~4x6" card for each pair of ADC's. IIRC the ENOB was a touch over 14.5 bits, so yeah, it takes some careful design to get near your full bit-rate on high-speed systems.
Perhaps someone that knows more about noise/digitization could drop some real math.
-
You can still buy new low-end analogue scopes, but high-end ones are long gone.
That does however mean you can get some of the best analogue scopes made, like theTek 2465A/B series very cheaply if you want one.
-
10*ln(2^N) for 8 bits is only 55 dB, versus 110 dB for 16 bits.
Minor correction - should be 20*log10(2^N), so 48dB, and 96dB.
(CMIIW on the 20*log)
-
Jup, roughly 6dB per Bit.
But thats all theoretical, in the real world you don't get this dynamic of course. But on the other hand you can extend your dynamic range beyond the capability of one single ADC (e.g. non-linear quantizer).
Anyway, dynamic-wise, all scopes are quite useless. For a usable Distortion-Analyzer for Audio you need at least 100dB. But that's not their domain...
I personally like the Combiscopes of Fluke/Philips (later Hameg also made some).
By pressing one button I can change between DSO and CRO, Timebase and so on is kept. Quite handy :)
Of course 200MS/s are not very impressive nowadays. But it's still a good scope.
-
Iwatsu still make high end analogue scopes...
Also, on my old 8 bit Tek TDS2012 I can select FFT mode and look at LF waveforms and see down to about -60dBc for distortion terms. It's hardly useless in this respect and I use this feature a lot.
-
Interesting topic.
I have a 400 Mhz 2465BDM , a 500 Mhz 54111D , and a 300 Mhz Rigol DS2302A. I also have a Rigol DSA 815-TG Spectrum Analyzer.
Being a general class HAM here is my take on things.
The 400 Mhz Analog scope I still prefer for RF work. It responds the fastest to rapid changes such as modulation. The Rigol is close..but not the same.
The 500 Mhz (old) digital scope is awesome for checking rising edges, etc. It doesn't work all that well for modulation checks or the like.
The 300 Mhz Rigol just about replaces the other two in terms of visibility in to waveforms. The FFT is just about useless so that's where the 815 comes in.
My need to turn on the analog scope is less and less all the time and the old 54111D I'm trying to sell off. I think if I had a 2467 with it's superior tube I would use the analog scope more for checking digital signals. As it stands now if I want to do anything with a non repetitive waveform I head to a digital scope without a thought. For RF work I still head to the analog, partly because I feel it's less likely to die if I screw up. I tossed 600v p-p at the Rigol the other day just to see if it would survive, and it did without issue. This may give me some confidence to use it more for RF work.
Regardless of what digital scope I have I'll do everything in my power to keep my analog scope running. It's in excellent condition both physically and functionally at the moment. It's all been recently calibrated and in great working order.
Are analog scope obsolete, depends on who you are and what your doing. For digital work they are 100% dead IMO. For analog work such as RF they still more then service the need. Does the "right" new digital scope replace them, pretty much. Some of the low end digital scopes on the market are complete crap IMO.
Jeff
-
As an old guy, I'm 65, analog scopes are pretty much it for me. In fact I may pick up another on the www before things get too obsolete, LOL.
-
... so it's a two way street:
With one direction having much wider lane.
-
I have had the same experience resolving minor but significant signal characteristics on analog oscilloscopes versus digital storage oscilloscopes but with an additional twist; my Tektronix 7854 mainframe is a very early digital storage oscilloscope and it will display things that my newer but old 8 bit DSOs and current 8 bit DSOs that I have tested will not. The difference though is despite being one of the earliest digital storage oscilloscopes, the 7854 has 10 bits of vertical resolution and 10 bits of vertical and horizontal CRT resolution making the CRT effectively 1024 x 1024.