EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: rbroders on May 31, 2016, 04:20:52 am
-
I'm new to DSOs, so forgive me if this is a stupid question.
I was using my scope for the first time today trying to measure ripple on a new 24V SMPS when switching on a 1A electromagnet load. I'm trying to decide whether adding a large capacitor (0.1F) to the output rail will help or hinder the system.
Things were going pretty well, until I ran into problems trying to zoom in on the signal by decreasing the vertical scale below 500mv/div. When at or above 500mv/div you can offset the signal +/-100V, but when below it you can only offset by +/-2V. What the heck! Anyone know the reason for such a drastic reduction in the offset level?
Thanks -- Bob
P.S. I have a current limiting power supply Meanwell SE-350-24 so it should be able to startup ok even with the giant cap on the rails. However, with no external load, it isn't stable with the cap: voltage wanders from 22-23V. I'm also driving a 3A actuator motor and thought the cap might help alleviate spikes. Ah, forget it. Still curious about the offset range limit thing though.
-
To measure small fluctuations on a large constant background (e.g. power supply ripple), the best solution is to set your scope's input to AC mode. That way, the input signal is coupled via a capacitor, and the DC offset is fully removed.
On the DS1054Z, AC vs. DC coupling can be selected by calling up the Input Channel menu (pressing channel selector 1-4); input coupling is the topmost softkey on the right.
-
Yeah, thanks, that is obviously the way to go. Doesn't it seem odd that there is a (nearly) 2 order of magnitude reduction in the offset capability when you drop below the 500mv/div level. Sheesh! Does the 1054 switch to a different amp or something? I mean seriously, what the heck!
-
I'm new to DSOs, so forgive me if this is a stupid question.
I was using my scope for the first time today trying to measure ripple on a new 24V SMPS when switching on a 1A electromagnet load. I'm trying to decide whether adding a large capacitor (0.1F) to the output rail will help or hinder the system.
Things were going pretty well, until I ran into problems trying to zoom in on the signal by decreasing the vertical scale below 500mv/div. When at or above 500mv/div you can offset the signal +/-100V, but when below it you can only offset by +/-2V. What the heck! Anyone know the reason for such a drastic reduction in the offset level?
Thanks -- Bob
P.S. I have a current limiting power supply Meanwell SE-350-24 so it should be able to startup ok even with the giant cap on the rails. However, with no external load, it isn't stable with the cap: voltage wanders from 22-23V. I'm also driving a 3A actuator motor and thought the cap might help alleviate spikes. Ah, forget it. Still curious about the offset range limit thing though.
Interesting... when I tried to duplicate your offset results I tried 1V/div and 200 mV/div, and in both cases the offset stopped at a maximum of +/- 20 volts. Then I realized you were probably having your channel probe setting at 1x rather than 10x.
So I tried 1x and did duplicate your results, with only +/- 2V available at the 200 mV/div and +/- 100V available at the 1V/div setting.
But why are you using 1x for the channel probe setting in the first place? 99.9 percent of the time we will normally use the 10x attenuation setting on the probe and the channel. Perhaps you were trying to get some extended range on the offset setting?
But the proper way to do the measurement you are looking for is to use 10x on probe and channel, use DC input coupling to measure the power supply's average or nominal DC voltage, then change to AC coupling to remove the DC component and measure the ripple, as eblaster says.
Why does the offset limit behave the way it does? I don't know but maybe it has something to do with protecting the inputs against user errors.
-
Does the 1054 switch to a different amp or something?
Yes.
Can you hear a clicking sound inside the scope when you turn the knob? That's a relay switching between different input circuits.
-
The idea of using 10x (div by 10 really) seems silly. Why attenuate the signal when you want a really good look at it? AC coupling I can understand, but attenuating seems wrong.
I still say +/-2V is lame at a 499mv/div. Its less than half the screen for crying out loud! Should be at least +/-10V I think. Are other scope better or worse in this regard?
Thanks -- Bob
-
The idea of using 10x (div by 10 really) seems silly. Why attenuate the signal when you want a really good look at it? AC coupling I can understand, but attenuating seems wrong.
Understand that for general measurement with a 1:1 probe connection of which can modify the signal in the DUT. It's called capacitive loading and it will affect the shown waveform by adding capacitance to the measured circuit.
For some indication of how much capacitance is involved your probe datasheet will have both 1:1 and 10:1 values listed. RTFM. :)
There are other probes namely active probes that have very low capacitance but are usually very expensive and worth more than any entry level DSO.
For these reasons it is normal practice to use 10:1 probes and adjust the DSO's channel input settings to match so that any OSD measurement are correct.
I still say +/-2V is lame at a 499mv/div. Its less than half the screen for crying out loud! Should be at least +/-10V I think. Are other scope better or worse in this regard?
Read the above again and have a re-think. AC coupling like others have mentioned is the correct method to display/measure a signal of interest riding on a voltage.
There was a thread quite recently discussing range offset limits, I'll find it and add as an edit. ;)
Here ya go:
https://www.eevblog.com/forum/testgear/ds1054z-distortion-issue/ (https://www.eevblog.com/forum/testgear/ds1054z-distortion-issue/)
-
The idea of using 10x (div by 10 really) seems silly. Why attenuate the signal when you want a really good look at it? AC coupling I can understand, but attenuating seems wrong.
Nope. Switch in 1x position is bad!
Why? Dave has done a video on this very subject:
https://www.youtube.com/watch?v=OiAmER1OJh4 (https://www.youtube.com/watch?v=OiAmER1OJh4)
-
Yeah, Dave's video makes it clear that you shouldn't use x1 mode when analyzing high bandwidth signals, but it still seems useful for the low bandwidth stuff I was looking at. I still contend that div10 on the probe means x10 on any internal errors/noise in the scope.
I didn't get the distortion issue posts. How are they relative to the Offset Range limit? Seems to me this is a flaw in the RIGOL DS1054Z design that the offset range is so very small (+/-2V or <1/2 screen at 499mv/div) once vertical range goes below 500mv/div. Again I am new to this. Perhaps other DSO brands have similar limits...
-
Yeah, Dave's video makes it clear that you shouldn't use x1 mode when analyzing high bandwidth signals, but it still seems useful for the low bandwidth stuff I was looking at. I still contend that div10 on the probe means x10 on any internal errors/noise in the scope.
I didn't get the distortion issue posts. How are they relative to the Offset Range limit? Seems to me this is a flaw in the RIGOL DS1054Z design that the offset range is so very small (+/-2V or <1/2 screen at 499mv/div) once vertical range goes below 500mv/div. Again I am new to this. Perhaps other DSO brands have similar limits...
Usually they don't.
DS1000Z:
1 mV/div to 499 mV/div: ±2 V
500 mV/div to 10 V/div: ±100 V
For comparison Rigol DS2000A or instek GDS-2000E have much higher offset range. Of course cheaper models are not as good as more expensive ones.
DS2000A:
500 uV /div to 50 mV/div: ± 2 V
51 mV/div to 200 mV/div: ± 10 V
205 mV/div to 2 V/div: ± 50 V
2.05 V/div to 10 V/div: ± 100 V
GDS-2000E:
1mV/div ~ 20mV/div : ±0.5V
50mV/div ~ 200mV/div : ±5V
500mV/div ~ 2V/div : ±25V
5V~10V/div : ±250V
As of the scopes in the same price range
GDS-1000B as example:
1mV/div : ±1.25V
2mV/div ~ 100mV/div : ±2.5V
200mV/div ~ 10V/div : ±125V
SDS:1000X
500uV ~ 150mV: ± 1V
152mV ~ 1.5V: ± 10V
1.52V ~ 10V: ± 100V
As you can see, it depends on the price. But DS1000Z with only 2 offset ranges doesn't seem to be well balanced in this regard even in it's price range.
-
Yeah, Dave's video makes it clear that you shouldn't use x1 mode when analyzing high bandwidth signals, but it still seems useful for the low bandwidth stuff I was looking at.
Until you probe a sensitive circuit that objects to the capacitive loding of a 1:1 probe.
I still contend that div10 on the probe means x10 on any internal errors/noise in the scope.
Watch these:
http://www.eevblog.org/video/EEVblog601-WhyDigitalScopesAppearNoisy-640x360.m4v (http://www.eevblog.org/video/EEVblog601-WhyDigitalScopesAppearNoisy-640x360.m4v)
http://www.eevblog.org/video/EEVblog610-WhyDigitalScopesAppearNoisyPart2-640x360.m4v (http://www.eevblog.org/video/EEVblog610-WhyDigitalScopesAppearNoisyPart2-640x360.m4v)
I didn't get the distortion issue posts. How are they relative to the Offset Range limit?
Waveform distortion in that thread arises from exceeding the offset limits.
Read it again and it may then be obvious.
Seems to me this is a flaw in the RIGOL DS1054Z design that the offset range is so very small (+/-2V or <1/2 screen at 499mv/div) once vertical range goes below 500mv/div.
Flaw? Maybe or a intended limit in design.
Again I am new to this. Perhaps other DSO brands have similar limits...
Keep asking questions.....that's what the forum is for.
-
Yeah, I guess flaw isn't the best word. Obviously for a budget scope, they have to allocate the dollars into different buckets. I like the way wraper put it in his informative post:
"But DS1000Z with only 2 offset ranges doesn't seem to be well balanced in this regard even in it's price range."
-
Any time I have to offset a channel baseline to an off-screen position the first thing I ask myself is "what am I doing wrong?"
-
Its nice that you talk to yourself ;-), but as I pointed out, you can't even move the trace 1/2 the screen height in 499mv/div (4.99V/div in 10x)! Clearly RIGOL made some compromises in the offset range limits that are a bit inconvenient...
-
Its nice that you talk to yourself ;-), but as I pointed out, you can't even move the trace 1/2 the screen height in 499mv/div (4.99V/div in 10x)! Clearly RIGOL made some compromises in the offset range limits that are a bit inconvenient...
Are we talking about the Rigol DS1054Z here? ???
On mine, in the first place the Vernier V/div scale only gives me 4.95 V/div, not 4.99 V/div in 10x, and the offset range is +/- 20 V, or slightly _more_ than 1/2 the screen height (total height 8 divisions, 1/2 height is 4 divisions @ 4.95 V/div = 19.8 V .
In 1x, I can only get 495 mV/div (steps of 5 mV) , not 499 mV/div, and again the offset range is +/- 2V, or slightly more than 1/2 screen height.
And of course at 500 mV/div the offset range is much greater.
So now I'm asking myself... what are _you_ doing wrong? ;D
That is to say, exactly what measurements do you need to make that require you to use 499 mV/div (or 495 mV/div) rather than 500 mV/div, and require really large offset settings? And require 1x probe settings? What is inconvenient for you about Rigol's choices in this matter?
I use my DS1054z every day for all kinds of different things, and I've certainly found some very real "flaws", but I have to say this issue has never inconvenienced me yet.
-
So now I'm asking myself... what are _you_ doing wrong? ;D
Doing wrong or not, it is not unreasonable to expect decent offset range. For example, now I'm measuring one circuit which has 1 V DC. I need oscilloscope being set to 10 mV/div (after 10:1 probe attenuation) to see what I want. AC mode has not much of use for me, because signal change can be rather slow and I want to see the DC voltage changes too. This measurement would be OK on DS1054Z, however if my voltages were higher and I needed to use higher voltage range and higher offset, that blind spot would make it pretty useless for me in that situation.
-
Hmmm, definitely got the Vernier wrong, mine is 4.95 as well (eyes not seeing good). Also, I just assumed there would be 10 divisions on the screen (what were they thinking using only 8? - are they octal rather than decimal?). Forgot to factor in the +/- doubling as well, so you can move signal slightly more than one full screen in worst case scenario. I guess that is okay - at least you can move two signals around enough so they don't overlap.
I was trying to measure noise on a 24Vpower supply. I was also getting some DC drift, so I wanted to use scope in DC mode but also enough gain to see the noise. Eventually had to use DMM to watch for drift and scope to see the noise. Disappointing first use of scope for me. Still learning though!
-
I was trying to measure noise on a 24Vpower supply. I was also getting some DC drift, so I wanted to use scope in DC mode but also enough gain to see the noise. Eventually had to use DMM to watch for drift and scope to see the noise. Disappointing first use of scope for me. Still learning though!
If that's the case your methodology and choice of tools was wrong.
Noise or ripple on a DC supply can be measured with scope or a DMM in ACV mode but only to the limit of the DMM's frequency response. DSO's OTOH are not so badly limited for this and it's also normal to engage the DSO's 20 MHz BW limiter to reduce displayed noise that is not of interest.
Drift OTOH is correctly measured with a logging instrument (bench or HH DMM) and displayed or uploaded to a trend chart that can be compared against other simultaneously measured parameters, say temp, line voltage, PSU load etc.
But you got the measurements you sought with alternative use of the tools on hand.
Welcome to EE. :)
-
I just assumed there would be 10 divisions on the screen (what were they thinking using only 8? - are they octal rather than decimal?).
They were probably thinking something like, "If we use a non-standard aspect ratio then we'll have to use custom-made screens instead of off-the-shelf. That will increase the price of everybody's 'scope just to make a tiny percentage of people a bit more happy. This means we won't be so competitive and we might go out of business."
I was trying to measure noise on a 24Vpower supply. I was also getting some DC drift, so I wanted to use scope in DC mode but also enough gain to see the noise. Eventually had to use DMM to watch for drift and scope to see the noise. Disappointing first use of scope for me. Still learning though!
You've got four channels. What was preventing you from looking at the drift on one channel of the 'scope and the noise on another?
-
You've got four channels. What was preventing you from looking at the drift on one channel of the 'scope and the noise on another?
Exactly. To wit, see the first scopeshot below. I am measuring the DC output from HP721A power supply on CH1, DC coupled at 10V/div, and on CH2 the ripple and noise, AC coupled at 20 mV/div. Both probes at 10x of course, connected to the same point.
(Note that my scope has about 4 mV positive offset error on the CH2 baseline that must be corrected for; this can be seen by setting the Input Coupling to "ground" as in the second scopeshot below.)
-
You've got four channels. What was preventing you from looking at the drift on one channel of the 'scope and the noise on another?
Exactly. To wit, see the first scopeshot below. I am measuring the DC output from HP721A power supply on CH1, DC coupled at 10V/div, and on CH2 the ripple and noise, AC coupled at 20 mV/div. Both probes at 10x of course, connected to the same point.
Usefulness of measuring DC voltage with such low resolution as minimum is dubious if the voltage drop is what you want to see.
-
You need more resolution than 1/10 of a volt, on a 24 volt nominal signal? Then perhaps the oscilloscope isn't the right instrument to use at all and the DMM is a better answer indeed.
-
You need more resolution than 1/10 of a volt, on a 24 volt nominal signal? Then perhaps the oscilloscope isn't the right instrument to use at all and the DMM is a better answer indeed.
And what if there are 2 Hz 200-300 mVpp voltage drops which you want to get rid of? How easy would it be to see on the multimeter what is going on? Your 0.1 V "resolution" actually is just not there. It is less than half of the least significant bit of the ADC resolution and is just below the noise and 0.2-0.3V won't be visible on the screen at all or look like a barely visible noise. Even 1V voltage drop will be barely visible.
-
You need more resolution than 1/10 of a volt, on a 24 volt nominal signal? Then perhaps the oscilloscope isn't the right instrument to use at all and the DMM is a better answer indeed.
And what if there are 2 Hz 200-300 mVpp voltage drops which you want to get rid of? How easy would it be to see on the multimeter what is going on? Your 0.1 V "resolution" actually is just not there. It is less than half of the least significant bit of the ADC resolution and is just below the noise and 0.2-0.3V won't be visible on the screen at all or look like a barely visible noise. Even 1V voltage drop will be barely visible.
I invite you to set up such a set of signals on the instrument(s) of your choice and display your measurement methodology and your screenshots here. Be sure to include some reference to the cost of the instrumentation required.
If you are looking for your 2Hz 200mV p-p voltage drops on the Rigol, would you perhaps use AC coupling and a vertical resolution of 100 mV/div, say, and a timebase setting of 500 ms/div? While simultaneously looking at the nominal DC voltage on another channel, DC coupled? Of course to see the higher frequency noise/ripple that I showed above you could use a third channel at even greater vertical sensitivity but you'd have to manually change the timebase, or use the horizontal zoom function.
I still don't see any need to set the vertical offset to some position way off screen to get a valid reading. Of course I suppose you could eventually make up some combination of signals that might require this. Again I invite you to do so, and show your measurement setup and screenshots of such a situation.
-
As far as I know, only keysight has a probe to measure this properly and directly, the N7020A. If anybody knows a similar probe for a sensible price I'd love to know.
-
They were probably thinking something like, "If we use a non-standard aspect ratio then we'll have to use custom-made screens instead of off-the-shelf. That will increase the price of everybody's 'scope just to make a tiny percentage of people a bit more happy. This means we won't be so competitive and we might go out of business."You've got four channels. What was preventing you from looking at the drift on one channel of the 'scope and the noise on another?
How about the fact that the screen is wildly rectangular and, by convention, reticle grids are square? The reticle is 8x12 because of the widescreen display. They could have gone 10x15 but 15 isn't an even number and the squares would be much smaller. They could have solved the x15 problem by just using larger menus. Heck, we could have had that stylish 10x10 reticle with a lot more room for menus.
BTW, the square reticle is required for XY mode.
All in, I can live with it the way it is.
-
If you are looking for your 2Hz 200mV p-p voltage drops on the Rigol, would you perhaps use AC coupling and a vertical resolution of 100 mV/div, say, and a timebase setting of 500 ms/div? While simultaneously looking at the nominal DC voltage on another channel, DC coupled? Of course to see the higher frequency noise/ripple that I showed above you could use a third channel at even greater vertical sensitivity but you'd have to manually change the timebase, or use the horizontal zoom function.
AC coupling, really? Don't you know that AC coupling does not work well at low frequencies? The fact is, low offset range is a serious shortcoming in certain (real) situations where another scopes would perform perfectly well.
Lets think about such situation, there is some load which have high current consumption a few times per second, say some heater, motor or power hungry relay connected to the power rail. You want to see what's going on on the power rail. As you can see from the following pictures, even at 10 Hz scope will be unable to properly display the waveform in AC mode. And even at 50 Hz there is noticeable waveform distortion.
(https://www.eevblog.com/forum/testgear/rigol-ds1054z-offset-range-limits/?action=dlattach;attach=229579;image)
(https://www.eevblog.com/forum/testgear/rigol-ds1054z-offset-range-limits/?action=dlattach;attach=229581;image)
(https://www.eevblog.com/forum/testgear/rigol-ds1054z-offset-range-limits/?action=dlattach;attach=229583;image)
-
As far as I know, only keysight has a probe to measure this properly and directly, the N2070A. If anybody knows a similar probe for a sensible price I'd love to know.
C'mon, you need to buy very expensive probe to make a such simple measurement? I guess you mean N7020A:
Low noise: 1:1 attenuation ratio probe adds only 10% to the baseline noise of the oscilloscope it is attached to
Large offset range: Has a large +/-24 V offset range, enabling users to set their oscilloscope at maximum sensitivity and have the signal centered on the screen
Low DC loading: 50 k? DC input impedance will not significantly load DC power rails
Large active signal range: Has a +/-850 mV active signal range in addition to its large offset range so users can measure large transitions of their power rails
High bandwidth: 2-GHz bandwidth makes it very useful for finding high-speed transients that can have detrimental effects on clocks and digital data
-
Yeah sorry N7020A.http://thesignalpath.com/blogs/2016/03/11/keysight-n7020a-2-0ghz-power-rail-probe-review-teardown-and-experiments/ (http://thesignalpath.com/blogs/2016/03/11/keysight-n7020a-2-0ghz-power-rail-probe-review-teardown-and-experiments/)
This will of course only work on a keysight scope with appropriate software support, not much help for you here sorry but I guess at some point other manufacturers will do something similar.
-
Yeah sorry N7020A.http://thesignalpath.com/blogs/2016/03/11/keysight-n7020a-2-0ghz-power-rail-probe-review-teardown-and-experiments/ (http://thesignalpath.com/blogs/2016/03/11/keysight-n7020a-2-0ghz-power-rail-probe-review-teardown-and-experiments/)
This will of course only work on a keysight scope with appropriate software support, not much help for you here sorry but I guess at some point other manufacturers will do something similar.
For 2 GHZ probe you need 2 GHz scope first of all. BTW Keysight makes also lower bandwidth version N7023A. But why would you need it anyway unless you are in serious business and have tons of money to blow. Main advantage of those probes is 1x 1:1 attenuation ratio which allows you to see very tiny noise.
-
You can use any BW probe on any scope and vice versa, but your only going to actually get whatever the lowest bandwidth part is.
The N7023A is just the 'prober' part, its included with N7020A.
-
You can use any BW probe on any scope and vice versa, but your only going to actually get whatever the lowest bandwidth part is.
Yes you can, but why would you blow a huge money away for the probe from which you cannot get an advantage from. That particular probe is certainly not for 99% of the oscilloscope users.
-
Your quite right, definitely in a niche, for people trying to accurately measure power rails with oscilloscopes. All niche products are expensive...and niche Keysight gear even more so :(
The advantage from the probe is it allows you measure exactly what your wanting to do, and no other probe I know of let's you do that.
-
Your quite right, definitely in a niche, for people trying to accurately measure power rails with oscilloscopes. All niche products are expensive...and niche Keysight gear even more so :(
The advantage from the probe is it allows you measure exactly what your wanting to do, and no other probe I know of let's you do that.
It is really useful only if you wan't to measure some very high frequency stuff, like FPGA power rail. For usual power supply measurements there is no value in it.
As far as I know, only keysight has a probe to measure this properly and directly, the N7020A.
About this again. This is completely wrong. That probe is for very detailed measurements with high bandwidth. For what OP wanted to measure, usual scope with decent offset range is completely adequate. 1x measurement without attenuation or fancy probes can be completely adequate as well if hundreds of MHz is not what you are looking for.
-
Haha, yes a big electromagnetic load certainly doesn't sound like a very sensitive load so the offset and range available on most scopes should be adequate.
-
If you need a much lower AC bandwidth, use your own capacitor, with the scope on DC coupled of course. :)
-
Sorry to butt in, but I'm not seeing what the Rigol is incapable of measuring here? The scope shots above show ~500 mv DC on both DC and AC couplings.
-
Sorry to butt in, but I'm not seeing what the Rigol is incapable of measuring here? The scope shots above show ~500 mv DC on both DC and AC couplings.
Those are from DS2000, not 1000Z. And DC voltage level is below what OP wanted to measure (limit of the signal gen). They are only for a reason to show that AC is not really a replacement for DC offset.
-
The offset range is multiplied by the attenuation before the offset signal is added. In the Rigol 1000Z series, there is only one switchable x100 attenuation stage yielding two offset ranges. Further, the dynamic range of the 1000Z input amplifiers is small limiting the range of the offset signal.
Some old analog oscilloscopes have huge offset ranges. A Tektronix 100 MHz 7A13 supports +/- 10 volts of offset at 1mV/div and +/- 100 volts of offset at 10mV/div. It is like having a position control with 10,000 vertical divisions of range but these were not primarily designed for this reason. What they allow is "slideback" precision DC voltage measurements at any point of a waveform using an analog oscilloscope. This is why the 7A13 is called a differential comparator and not a differential amplifier although it does that as well.
The modern implementation of the Tektronix 7A13 is the Preamble Instruments 1855 and from what I remember, the same designer worked on both. Preamble was bought by LeCroy but still produce this stand alone differential probe amplifier. If you have to ask the price, then you cannot afford it.
As far as the x1 versus x10 passive probe, the output impedance of a power supply or regulator is so low that the added capacitance of the x1 probe is insignificant but its relatively low bandwidth limitations still apply.
-
(https://www.eevblog.com/forum/testgear/rigol-ds1054z-offset-range-limits/?action=dlattach;attach=286036;image)
(https://www.eevblog.com/forum/testgear/rigol-ds1054z-offset-range-limits/?action=dlattach;attach=286038;image)
-
What, are you telling me there are better oscilloscopes than the Rigol DS1054z? Who knew? :horse:
-
What, are you telling me there are better oscilloscopes than the Rigol DS1054z? Who knew? :horse:
Micsig didn't do a very good job on this one:
https://www.eevblog.com/forum/testgear/dc-offset-wander/ (https://www.eevblog.com/forum/testgear/dc-offset-wander/)