EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: Dark Prognosis on October 17, 2013, 04:35:48 pm
-
So, I was having this argument with this guy on a review site and he brought up Dave Jones. He said that this scope really is only a 30 or 40mhz scope due to how it works and the industry knows it and Dave as well. Is that right?
How can Dave recommend this scope as a 100mhz dual trace scope if it only does 30mhz??? Something smells funny.
-
yeah, take a look at this photo
(http://johansense.com/bulk/boost/4phasev_1_10.JPG)
there should be 25 data points per trace per division, instead there is about 10. there's also a slope change once the blue trace gets to 48 volts, it starts climbing slower. it should reach all the way to 60 volts. the hook in the yellow trace should be lined up with the bottom most ring of the blue trace.. which has a real life frequency of 50Mhz.
instead the blue trace is delayed 5ns, and the scope shows that its frequency changes!... which i know is not the case, it should ring down like any other smps.
the only difference between my scope and the 1102E is the memory depth and the screen size.
another issue is crosstalk. note that the blue trace is lifted 1-2 volts while the yellow trace is falling.
but also note that the oscope isn't sure about that so it waits half a division before displaying this "lift"
in real life there is something there.. its on the order of negative 10 milivolts plus the inductance of the current sense resistor.
so you should see a spike on the blue trace on the order of -1 volt for like 5ns, then it should fall to zero, then stabilize at -10mv (these numbers are relative to the voltage drop across rds on before the gate is pulled to zero)
(one half amp of gate drive through a 20mohm resistor with an inductance of say 5nH)
another issue is lack of significant digits on such things as volts rms. (no surprise, 8 bits is only 8 bits.)
so you might see 60 volts rms displayed, or 61.44, or 60, or 61.44.. it doesn't know how to do a true rms over more than one line cycle i guess.. because that should average out to better than 8 bits.
another issue is the fft only calculates what is on the screen, not what is in the memory. the screen being 234 pixels high rather than 256 means that there's data loss somewhere between the 8 bit ADC and the screen. i've verified this by adjusting the signal so it is just barely clipping on the screen, but should still be below 256 bits wide, the fft and the rms measurements displayed will change.
there is also no warning about this data loss.
-
there should be 25 data points per trace per division, instead there is about 10. there's also a slope change once the blue trace gets to 48 volts, it starts climbing slower. it should reach all the way to 60 volts. the hook in the yellow trace should be lined up with the bottom most ring of the blue trace.. which has a real life frequency of 50Mhz.
instead the blue trace is delayed 5ns, and the scope shows that its frequency changes!... which i know is not the case, it should ring down like any other smps.
Please show us these signals with a good scope.
the only difference between my scope and the 1102E is the memory depth and the screen size.
I doubt that.
-
Then how could Dave have recommended this scope so highly for so many years now?
Something smells when Dave highly recommends a 100Mhz scope yet it supposedly only does about 30mhz.
-
I had an 1102E, before I broke it... :-\ It was definitely a proper 100 MHz scope. This guy is full of bullshit. And no, my Rigol didn't behave like that stinking turd of a Siglent.
-
yeah, take a look at this photo
(http://johansense.com/bulk/boost/4phasev_1_10.JPG)
there should be 25 data points per trace per division, instead there is about 10. there's also a slope change once the blue trace gets to 48 volts, it starts climbing slower. it should reach all the way to 60 volts. the hook in the yellow trace should be lined up with the bottom most ring of the blue trace.. which has a real life frequency of 50Mhz.
instead the blue trace is delayed 5ns, and the scope shows that its frequency changes!... which i know is not the case, it should ring down like any other smps.
the only difference between my scope and the 1102E is the memory depth and the screen size.
another issue is crosstalk. note that the blue trace is lifted 1-2 volts while the yellow trace is falling.
but also note that the oscope isn't sure about that so it waits half a division before displaying this "lift"
in real life there is something there.. its on the order of negative 10 milivolts plus the inductance of the current sense resistor.
so you should see a spike on the blue trace on the order of -1 volt for like 5ns, then it should fall to zero, then stabilize at -10mv (these numbers are relative to the voltage drop across rds on before the gate is pulled to zero)
(one half amp of gate drive through a 20mohm resistor with an inductance of say 5nH)
another issue is lack of significant digits on such things as volts rms. (no surprise, 8 bits is only 8 bits.)
so you might see 60 volts rms displayed, or 61.44, or 60, or 61.44.. it doesn't know how to do a true rms over more than one line cycle i guess.. because that should average out to better than 8 bits.
another issue is the fft only calculates what is on the screen, not what is in the memory. the screen being 234 pixels high rather than 256 means that there's data loss somewhere between the 8 bit ADC and the screen. i've verified this by adjusting the signal so it is just barely clipping on the screen, but should still be below 256 bits wide, the fft and the rms measurements displayed will change.
there is also no warning about this data loss.
Funny how it is 8bits of resolution while the lowly 8bit Arduino has 10 bits of resolution on the analog pins.
-
Funny how it is 8bits of resolution while the lowly 8bit Arduino has 10 bits of resolution on the analog pins.
You try to do 10 usable bits at 1 GSa/sec and get back to me.
-
I had an 1102E, before I broke it... :-\ It was definitely a proper 100 MHz scope. This guy is full of bullshit. And no, my Rigol didn't behave like that stinking turd of a Siglent.
Yeah, I think he was shill (was over on Amazon) but when I went to reply to him this morning the thread had been removed (404 error). So, I have no idea but what the guy was describing was not accurate nor correct. I called him out on it and he brings up Dave Jones as saying it had these issues and 30mhz was its tops. I told him I have sub'd to Dave for a long while now and I never heard him say that but I could have missed it as I don't follow these forums as much as I do his videos.
-
Funny how it is 8bits of resolution while the lowly 8bit Arduino has 10 bits of resolution on the analog pins.
You try to do 10 usable bits at 1 GSa/sec and get back to me.
No thanks I was just saying. Do any scopes sample at 1gsps, or higher, at more than 8 bits of resolution? I think I am seeing some of the reasons people say to keep a good analog scope around.
-
Not many. There are sampling scopes, which capture a repetitive waveform at high speed by taking many captures with a time offset - if I'm reading the specs correctly, the PicoScope 9200 (http://www.picotech.com/picoscope9200-specifications.html) pulls off 16 bits of resolution with an analog BW of 8/12 GHz, so presumably the overall sample rate, once multiple acquisitions are combined, approaches that or beyond. (Note that I have no experience with high speed sampling scopes...) It's also $10k.
Many of the Agilent scopes have a high resolution mode, which is also made by combining multiple captures, and gets you 12 bits.
-
That boggles my mind how you are oversampling to increase the bit depth. Interesting.
I am saving up for an 1102E then a signal gen then a proper PSU then the 2000 series so sometime in the next three years, lol.
-
Dave measured the bandwidth of his hacked DS1052E. It's a 100MHz scope. Skip to 16:10.
EEVblog #311 - Jim Williams Pulser Followup (https://www.youtube.com/watch?v=uBYMePUFinQ#ws)
-
No thanks I was just saying. Do any scopes sample at 1gsps, or higher, at more than 8 bits of resolution? I think I am seeing some of the reasons people say to keep a good analog scope around.
Yes, but they're pricey. The link below is to a LeCroy. Agilent has some, as well as Yokogawa. I'm not sure that Tek does.
http://teledynelecroy.com/oscilloscope/oscilloscopeseries.aspx?mseries=422 (http://teledynelecroy.com/oscilloscope/oscilloscopeseries.aspx?mseries=422)
-
Maybe those guys are using the scope only on the X1 probe setting and claiming this. If so then yes, it would appear that way, but so would most scopes. Now if they've been using a nicer Tek or Agilent scope with the expensive probes, those can auto-detect the probe and compensate so they may not be realizing that. (I hope they would)
-
Dave measured the bandwidth of his hacked DS1052E. It's a 100MHz scope. Skip to 16:10.
EEVblog #311 - Jim Williams Pulser Followup (https://www.youtube.com/watch?v=uBYMePUFinQ#ws)
I knew this guy was a shill and full of shit as I remember that episode and not once have I heard Dave bad mouth the 1102E (or even the 1052E). This guy got assed out and decided to try and bully me by throwing Dave Jone's name at me like a prize fighter throws punches. It didn't work so the last thing he told me (before Amazon removed the threads) was "I will satisfy you childish behavior by saying I am completely wrong." My childish behavior was calling him telling him that he was wrong and backing it up.
For future reference this joker's name (for any future reviews you might see from him) is Gordy.
-
Maybe those guys are using the scope only on the X1 probe setting and claiming this. If so then yes, it would appear that way, but so would most scopes. Now if they've been using a nicer Tek or Agilent scope with the expensive probes, those can auto-detect the probe and compensate so they may not be realizing that. (I hope they would)
Exactly what I was saying about it not being an issue with just this scope as he directly said. I told him not to single out this scope for the issues he was mentioning as it was all of them doing it as it was the nature of the beast with sampling etc...
-
No thanks I was just saying. Do any scopes sample at 1gsps, or higher, at more than 8 bits of resolution? I think I am seeing some of the reasons people say to keep a good analog scope around.
Yes, but they're pricey. The link below is to a LeCroy. Agilent has some, as well as Yokogawa. I'm not sure that Tek does.
http://teledynelecroy.com/oscilloscope/oscilloscopeseries.aspx?mseries=422 (http://teledynelecroy.com/oscilloscope/oscilloscopeseries.aspx?mseries=422)
That thing is a wet dream but I lost it at the starting from price of $14,900 for the 350mhz version.
-
Just a bit of history; Dave's posts about Rigol are as early as ~ 2007 but there was a lot of praise for it from others users since, you can find early discussions about this model in the newsgroups, before this forum existed.
http://www.electronicspoint.com/rigol-caught-their-pants-down-ds1052e-oscilloscope-t172654.html (http://www.electronicspoint.com/rigol-caught-their-pants-down-ds1052e-oscilloscope-t172654.html)
P.S. The newsgroups are notorious for its vitriol and flame wars; you'll get a bit of it from the infamous Paul A.
http://www.electronicspoint.com/emona-rigol-digital-scopes-t135569.html (http://www.electronicspoint.com/emona-rigol-digital-scopes-t135569.html)
If you read the long 1052e hacking threads on this forum, there was subsequent testing done by others to reconfirm Dave's impressions.
-
Maybe those guys are using the scope only on the X1 probe setting and claiming this. If so then yes, it would appear that way, but so would most scopes.
If anybody ever attempts to measure an oscilloscope's bandwidth like this, never, ever ever ever, believe anything else they say about a topic more complicated than a hamburger.
-
Please show us these signals with a good scope.
the only difference between my scope and the 1102E is the memory depth and the screen size.
I doubt that.
i believe they use the same firmware even!
the difference between the cml and the cnl is a 28$ memory chip.
-
We are talking about a Rigol DS1102E.
-
We are talking about a Rigol DS1102E.
And?
-
Then how could Dave have recommended this scope so highly for so many years now?
Something smells when Dave highly recommends a 100Mhz scope yet it supposedly only does about 30mhz.
It is a 100MHz (or more) bandwidth scope. Feed in a sine wave until it's 3dB down and measure the frequency yourself.
-
Then how could Dave have recommended this scope so highly for so many years now?
Something smells when Dave highly recommends a 100Mhz scope yet it supposedly only does about 30mhz.
It is a 100MHz (or more) bandwidth scope. Feed in a sine wave until it's 3dB down and measure the frequency yourself.
Exactly, Dave. I was taken back when the dude threw your name at me #1 and #2 he knows enough about you, and apparently electronics, yet doesn't know how crap works? Made me laugh, and sad, all at once.
I still say he was a shill and at least I was able to call him that. Either he was a shill or completely retarded but he can't be that retarded because he knew your name at least.
-
Then how could Dave have recommended this scope so highly for so many years now?
Something smells when Dave highly recommends a 100Mhz scope yet it supposedly only does about 30mhz.
It is a 100MHz (or more) bandwidth scope. Feed in a sine wave until it's 3dB down and measure the frequency yourself.
but that is only true if the front end amplifier doesn't have the uumph to feed the ADC, which is likely true, but somewhat unexpected.
if the processor has to do math on three samples to get a 7 bits out of 8 accurate that's a different issue.. which is what appears to be going on.
-
Then how could Dave have recommended this scope so highly for so many years now?
Something smells when Dave highly recommends a 100Mhz scope yet it supposedly only does about 30mhz.
It is a 100MHz (or more) bandwidth scope. Feed in a sine wave until it's 3dB down and measure the frequency yourself.
but that is only true if the front end amplifier doesn't have the uumph to feed the ADC, which is likely true, but somewhat unexpected.
if the processor has to do math on three samples to get a 7 bits out of 8 accurate that's a different issue.. which is what appears to be going on.
So, you are going to continually say that your Siglent is internally a Rigol?
-
but that is only true if the front end amplifier doesn't have the uumph to feed the ADC, which is likely true, but somewhat unexpected.
if the processor has to do math on three samples to get a 7 bits out of 8 accurate that's a different issue.. which is what appears to be going on.
Huh? A lot of the parts in the Rigol are actually 300~700MHz parts. Like the VGA and ADC drivers, from memory. It's just got a ~130MHz bandwidth filter so you don't get horrible aliasing at 1Gsps sampling rate. If you ripped out the filters it would probably do 200~300MHz. But it wouldn't be any good at that frequency due to the aliasing effects.
-
Then how could Dave have recommended this scope so highly for so many years now?
Something smells when Dave highly recommends a 100Mhz scope yet it supposedly only does about 30mhz.
It is a 100MHz (or more) bandwidth scope. Feed in a sine wave until it's 3dB down and measure the frequency yourself.
but that is only true if the front end amplifier doesn't have the uumph to feed the ADC, which is likely true, but somewhat unexpected.
if the processor has to do math on three samples to get a 7 bits out of 8 accurate that's a different issue.. which is what appears to be going on.
So, you are going to continually say that your Siglent is internally a Rigol?
i don't recall what the consensus is but afaik most siglents are identical to attens.. rigol is on their own but the history of these three corporations and others are so identical its difficult to tell if they all don't work out of the same office.
-
but that is only true if the front end amplifier doesn't have the uumph to feed the ADC, which is likely true, but somewhat unexpected.
if the processor has to do math on three samples to get a 7 bits out of 8 accurate that's a different issue.. which is what appears to be going on.
Huh? A lot of the parts in the Rigol are actually 300~700MHz parts. Like the VGA and ADC drivers, from memory. It's just got a ~130MHz bandwidth filter so you don't get horrible aliasing at 1Gsps sampling rate. If you ripped out the filters it would probably do 200~300MHz. But it wouldn't be any good at that frequency due to the aliasing effects.
see this thread. https://www.eevblog.com/forum/buysellwanted/which-oscilloscope-is-better/msg31932/#msg31932 (https://www.eevblog.com/forum/buysellwanted/which-oscilloscope-is-better/msg31932/#msg31932)
if the adc is actually good to 200Mhz then why do we have information loss at only 50mhz.
anyhow, i suspect all of these scopes are the same way, but people aren't looking at the waveform with a real time critical eye.
-
They do not work out the same office, that's nonsense. For one you can see completely different software and hardware design concepts in teardowns and demo videos/reviews. For example Atten use non-45deg traces on the ADCs. Rigol don't. And I believe Atten used 4 x dual 80Msps ADC OC @ 125MHz instead of 5 x dual 40Msps ADC OC @ 100MHz in their 50/100MHz product.
The reason you're seeing information loss is going to be down to probing and the characteristics of the circuit. I have used my 1102E up to 100MHz with no "information loss".
And, as stated, these models have a deliberate ~130MHz filter, so their input bandwidth is not 200MHz. But the design is (theoretically) capable of around that. It's just limited to avoid aliasing.
-
but that is only true if the front end amplifier doesn't have the uumph to feed the ADC, which is likely true, but somewhat unexpected.
if the processor has to do math on three samples to get a 7 bits out of 8 accurate that's a different issue.. which is what appears to be going on.
Huh? A lot of the parts in the Rigol are actually 300~700MHz parts. Like the VGA and ADC drivers, from memory. It's just got a ~130MHz bandwidth filter so you don't get horrible aliasing at 1Gsps sampling rate. If you ripped out the filters it would probably do 200~300MHz. But it wouldn't be any good at that frequency due to the aliasing effects.
see this thread. https://www.eevblog.com/forum/buysellwanted/which-oscilloscope-is-better/msg31932/#msg31932 (https://www.eevblog.com/forum/buysellwanted/which-oscilloscope-is-better/msg31932/#msg31932)
if the adc is actually good to 200Mhz then why do we have information loss at only 50mhz.
anyhow, i suspect all of these scopes are the same way, but people aren't looking at the waveform with a real time critical eye.
I own a Siglent SDS1102CNL, and can say that it is NOT the same as a Rigol. The Rigol is significantly more responsive, and updates faster.