HI:
I buy few days ago a small and basic oscilloscope Agilent DSO1014a. When I connect it and begin to measure I notice that the channels are too noisy when I connect the 200Mhz agilent included probes even connected to GND in the internal square wave generator or in other channel BNC connector. Noise it´s 50mV pp (or worst) 95Mhz aprox modulated by a 10-15Mhz signal. If I coupled to GND or disconect the probes the noise dissapear or it´s very reasonable (<<<20mV). I change the location of scope, try it in other building and the noise it´s the same.
This is normal that even connected to GND the osciloscope be too noisy?
THANKS IN ADVANCE and sorry for my bad english.
I buy few days ago a small and basic oscilloscope Agilent DSO1014a. When I connect it and begin to measure I notice that the channels are too noisy when I connect the 200Mhz agilent included probes even connected to GND in the internal square wave generator or in other channel BNC connector. Noise it´s 50mV pp (or worst) 95Mhz aprox modulated by a 10-15Mhz signal. If I coupled to GND or disconect the probes the noise dissapear or it´s very reasonable (<<<20mV). I change the location of scope, try it in other building and the noise it´s the same.
This is normal that even connected to GND the osciloscope be too noisy?
THANKS IN ADVANCE and sorry for my bad english.
Today in the morning I test the scope with a lot of cheap probes and think that the problem it´s solved, but when I do the printscreens I doubt again.
The next screen it´s with the scope without probes
With probes connected to DSO but left unconnected of GND or any circuit
Probes connected to the each self GND
Probes connected to the "Probe Comp" GND . Magnitudes are variables depending relative position of probe coaxil
If I disconnect some probes increase the amplitud of noise. In the next screen I put in channel one the N2863A probe and in channel 2 a cheap probe of """""100 Mhz""""". I think that magnitude diference would be attribute to brutal attenuation of cheap probe.
Same that previous screen BUT the probe used it´s and "not so bad" 100 Mhz probe. I use it in TEK scopes with minor differences with TEK original probes.
Idem previous but with other time base showing that noise it´s a stable sine modulate by other 10 or 15 times slower.
I put the other probe in X1 and notice that noise greatly reduce that it´s correct because it´s normal that bandwith of probes reduces in x1 range.
I've missed something, it´s correct that the probe connected to the self ground would be so nosy? I assume that "Probe Comp GND" it´s a digital GND and it´s noisy, but the probes self attached to GND I think that I see less than 2mV of noise and I see 25 times it!!!
Any help?
By the way, do you knows why sometimes in the screen copy it´s included the right menu and in others not? I always do the same procedure and don´t understand why sometimes shows the menu and sometimes not.
The parameter files of each screen are in
http://www.lorien-sistemas.com/dso1014a/NewFile0.png
http://www.lorien-sistemas.com/dso1014a/NewFile1.png
http://www.lorien-sistemas.com/dso1014a/NewFile2.png
http://www.lorien-sistemas.com/dso1014a/NewFile3.png
http://www.lorien-sistemas.com/dso1014a/NewFile4.png
http://www.lorien-sistemas.com/dso1014a/NewFile5.png
http://www.lorien-sistemas.com/dso1014a/NewFile6.png
http://www.lorien-sistemas.com/dso1014a/NewFile7.png
http://www.lorien-sistemas.com/dso1014a/NewFile0.txt
http://www.lorien-sistemas.com/dso1014a/NewFile1.txt
http://www.lorien-sistemas.com/dso1014a/NewFile2.txt
http://www.lorien-sistemas.com/dso1014a/NewFile3.txt
http://www.lorien-sistemas.com/dso1014a/NewFile4.txt
http://www.lorien-sistemas.com/dso1014a/NewFile5.txt
http://www.lorien-sistemas.com/dso1014a/NewFile6.txt
http://www.lorien-sistemas.com/dso1014a/NewFile7.txt
But, a question, if I join the GND of each probe with his point the interference must be reduce and I observe the oppposite. The shielding of the probe it´s so weak? How I can measure something in a circuit if I have 50-100 mV noise in the 20mV scale? I work with signals near the volt range, 100mV it´s not allowed. I´m very confuse, never happens to me that an oscilloscope introduce this level of noise through the probe. I see this when I measure a wire without shielding but no with a coaxial designed for a 300Mhz probe!
I think that the GND of the probes aren´t conected in any form and then the probe is not shielded and when I join the GND with tip enhance the antenna not isolate the signal path, but, when I connect the GND of probe to a DC battery and put the tip in the positive I measure the corrected potencial. If I float the GND that measure must to fail and that isn´t that happens.
Any help please? All suggest it´s welcome!
Sorry for my bad english!
Sometimes an oscilloscope probe appears to behave strangely, especially if one is not aware of the probe’s parasitics. For example, when you connect the ground lead to the tip of the probe, you create effectively a loop antenna! Because of its inductance, a ground lead is not a short-circuit for high-frequencies. Consider that a 20 cm long wire has inductance of about 200 nH, so the ground lead will resonate at 90~100 MHz with the typical 12~15 pF capacitance of the probe. The resulting tuned loop can work as a reasonably efficient antenna for the FM broadcasting band.
A good discussion of oscilloscope probes can be found in the old ABC’s of Probes, by Tektronix. The ABC’s is reproduced as Appendix A of this excellent article by Jim Williams:
http://cds.linear.com/docs/Application%20Note/an47fa.pdf
Two days ago, on Thursday, I received my long waited oscilloscope Agilent DSO-X 3014A but... tragedy! It exhibits a so strong "noise" to make even impossible the probe compensation procedure.
Here it is the picture of how I see the probe compensation signal and in this picture, being static, it is impossible to see how the trace jumps left and right because the OS is not able to trigger!
Dear Nestor, the "noise" on my expensive toy looks much the same as that illustrated in your pictures!
I don't know if it is the same in your case but, in mine, the "noise" amplitude varies according to the position of the probe cable, i.e. when the cable is fully extended the "noise" is much stronger then when the cable is winded in a few turns (I apologize for my English, in this language I only have a 4 bits resolution...); it doesn't change changing the position of the loop formed by the probe ground strap connected to the probe tip.
These latter facts, together a few others and some past experiences, make me to believe that the problem is actually caused by a very bad case of self-oscillation in the vertical preamplifier (this is why I write "noise" with the two "") and not by externally picked noise.
For instance, my 24 years old Tektronix 468 has developed, over the years, the same problem (but at a lot less extent) and this, according to Tektronix, is due to the components aging; strange enough, my vintage, more then 40 years old, Tektronix 547 with the 100 MHz 1A1 plugin doesn't have this problem and all the traces are perfect as they were the first day!!
Dear Frank, I have made a FFT analysis of this oscillation and this if the screen shot:
The screen shot has been taken with the probe tip and its ground clip both connected to the ground point on the oscilloscope front panel.
If it was an externally picked noise, I believe, it would be attenuated by the band limiting filter hidden somewhere in the vertical amplifier chain (my OS is the 100 MHz base model) while if it is internally generated it could be present everywhere and measured by the FFT (unless the limiting filter is digital and in the sampled signal chain, but who knows?); and this make me to suspect that this oscillation is somehow related to this filter.
Anyway, I have already informed the Agilent vendor about this problem and if they are not able to solve it I will return the instrument because in this condition it is unusable!
Ciao, so far and hoping to find a solution because, under all other aspects, this instrument is a wonderful beast.
Franco
Web: http://www.flanguasco.org
Attachments
Yes, with NO DOUBT noise it´s inyected at probe tip (I didn´t post the screen but if I unscrew the pointy tip of the probe the noise it´s reduced) but the fact that surprise me it´s that with one old Agilent (HP) and with three Tek´s with more bandwidth (250-500 Mhz) I haven´t this problem, the noise it´s of 10% of minor graticule max, typically 2-5%.
I think that 2-5 mV of noise with the probe not connected to any circuit it´s normal, but 50-100 mV even connected to the GND of the scope I think (may be wrong) that´s it´s not usable. I read several years ago the ABC of TEK (even I have the old printed free versión and the PDF file) and I think that I understand the problems of work in """"Hi frequency"""" circuits and took measures, but in other scopes when I connect the probe tip to the scope GND the noise it´s greatly reduced and it´s not appreciated or may be discarded. When I work with signals of 1volt magnitude, 100 mV captured by the probe it´s a lot of noise and blocks the ringing that I need to see.
When I says that "shorcircuit" the GND with the tip wasn´t with the crocodile tail, I use a small wire from the GND top of the probe to tip. I not have a loop antenna, I have a coaxial with the end shortcircuit. It´s an antenna too, but it´s so sensitive?.
I would take tomorrow some probes of my work to test the scope. If the problem disappear, then I need to buy 4 new probes better that Agilent but I think/believe that probes includes with the scope are better than the half of probes that I have at work. I not believe that a 300Mhz passive probe of agilent would be so noisy.
Really I not understand that happens at all, I see noise inyected in the probe but is not like this.
Dear Franco:
Sorry that I wrote to you in english but my Italian it´s worst. I can write in my brutal "argentinoliano" but believe me, Dante himself would rise from his grave to condemn me to some of his underworld circles!. Anyway it´s rude and definitely not polite to write in other language that the forum uses and I´m grateful with all of answer and read that support my very bad english!.
Thanks for you long answer and the scope that you buy IT´s a BEAUTY!!! I´m tenting for it but the budget for my home "toy" not permit spend more than 1000 dollars and I really needs the logic analyzer but if I sum to the scope cost the modules that need to look at serial protocols the cost really escape of my personal budgets! Really a wonderful scope!
I think that it´s reasonable that you suggest, something in the pre-ASIC circuit that oscillate when I inyect noise near the bandwidth limit. I look the same that observe you when I move the probe coaxial and assume at the beginning that it´s EMI noise but it´s too much noise for a oscilloscope under the giga band. Because that I put at the beginning if this is normal because I see something like EMI but two orders of magnitude greater and suspect that the scope may be has another problem and ask in the forum to see that Agilent experts say.
Unfortunately I didn´t buy the scope in Argentina and I didn´t know if I have some kind of guarantee service to carry it in my country for control, but I think that if this amount of noise it´s normal to this scope I made a mistake buying it. I use Agilent (HP´s) scopes for almost 20 years and never has a problem, always works like a charm, comfortable and precise and I not believe that something be changed with this model.
Actually I´m disoriented with this problem and I really really appreciate to KSmith, Frank_BR and you to answer this post because help a lot to me! May be in the next days I resolve the problem in any form.
I take a screen of the FFT to post, maybe this help to solve this.
http://www.lorien-sistemas.com/dso1014a/NewFile8.txt
Detail in the FFT to view the FM zone.
http://www.lorien-sistemas.com/dso1014a/NewFile10.txt
Probe shot-circuited to GND not by a plain wire but with a resistance of 47 ohm. I´m impressed, the noise it´s gone, but if I use a simple wire not!
http://www.lorien-sistemas.com/dso1014a/NewFile9.txt
I wouldn't be so sure: if you connect the probe tip to the clip of its ground lead AND you wrap all inside a grounded bag formed by aluminum foil, the oscillation remains exactly the same.
This, to me, should exclude the possibility of externally picked noise.
Believe, believe
This is true, but it is also true that this is the frequency at which a 100 MHz, band limiting filter, would oscillate.
On the other hand Frank wrote:
Right but then, if you double the length of the ground lead, the "noise" frequency should halve and this doesn't happen: try you too.
Ciao and let keep crossed our fingers...
Franco
Web: http://www.flanguasco.org
I say that the noise it´s inyected because if I disconnect the probe the noise practically dissapear, BUT, I agree with you about something happens between the probe and the AIC that amplifies or generate (I don´t know) this noise.
If I put the probe and almost all the coaxial in a shield made with aluminium foil and put to GND the foil, the noise it´s reduced but with random noise peaks
This is the signal without foil
http://www.lorien-sistemas.com/dso1014a/NewFile11.txt
And this with the foil and a random peak captured
http://www.lorien-sistemas.com/dso1014a/NewFile12.txt
VERY VERY BAD NEWS!
That it´s true too, I notice at the beginning that noise it´s distributed central to a frequency near the bandwidth limit of the scope.
I´m doubt above that. In a loop antenna I´m not sure that the frequency change, I think that only changes the sensitivities but I HATE all that concern radio stuff and resist to think about it
Maybe somebody from Agilent can put some light in this topic. Thanks for your answer
The sources of the interference, as I said before, appear to be the FM radio stations in the area. I would like to remember that the frequency response of an oscilloscope doesn’t fall abruptly after the pass band. For example, the response of a 100 MHz oscilloscope drops only 3dB at 100 MHz. That means that a typical 100 MHz oscilloscope has significant response at frequencies as high as 150 or 200 MHz.
The picture below, taken from a 200 MHz digital-storage oscilloscope with a 400 MHz Tek probe, shows that the interference is a sine-like wave with a frequency of about 90~100 MHz. (Please ignore the wrong reading of the internal counter. The correct value can be inferred easily from the sweep speed.)
The second picture shows the same interference wave seen in first picture, but now at different sweep speed. At the first sight, the interference appears to be amplitude modulated, but the “AM look” is due to the “beating” of several waves of different frequencies that compose the interference.
The third picture shows the same interference as above, but captured now by a 200 MHz Fluke Scopemeter with a Fluke 200 MHz probe.
Maybe tomorrow I will show a way to get rid of the interference from FM radio stations…
Attachments
PLEASE! I don´t blame Agilent! In the last two decades I think that Agilent (HP) and Tektronix are 2 of (perhaps) 3 or 4 best oscilloscopes´s makers. In fact, I´m surprised that an Agilent probe inyect more noise than a Tek.
I´m concerned because with the oscilloscopes that I used before this, the noise is less relevant. Maybe they have better internal filters or perhaps this oscilloscope be more sensible, I don´t know. I´m happy with the people at Agilent and in worst case only my unit eventually has a problem.
The only thing that I need now it´s know that this noise it´s normal and then works to reduce or eliminate it.
I really appreciate your suggestions and wait for your method to reduce FM noise.
Today I´m looking about this noise in the Agilent site found this articles
http://cp.literature.agilent.com/litweb ... 3020EN.pdf
Franco, look at the screen in the article and it´s the same noise that we see on ours oscilloscopes. Off course, I haven´t the same oscilloscope that uses as example in the article
Willco:
If you look at Image 9 I put the print screen of probe short-circuited with 47 ohm (I don´t have near a 49.9 ohm resistor but it´s close) and the noise it´s almost gone. Why when I short circuit with a wire observe noise and with an R no, escape my knowledge.
Thanks to everybody, really This post was very useful to me and I´m very glad if something of Agilent confirm that all is normal. THANKS AGAIN and sorry for my bad english.
I previously say
And thinking after post in that Franco says about internal oscillations that If part of the noise it´s the internal noise of analog front end, when I use a 1x probe the final noise it´s reduced because it´s not "amplified". Id est, the noise floor of the front end it´s the same in all probes, for example 2mA, but when I use a 10x probe this noise it´s show as 20 mA. If that it´s true I need to use the 1X probes to do the noise test not the 10x ones. It´s that correct?
THANKS!
while I can't dismiss at all the possibility that "some" of the noise is coming from external sources (FM broadcast?), I still believe that the majority of it is responsibility of the scope itself : have you seen the first picture on my first post? with 0.4 V of noise over the 2.4 V Probe Comp square wave signal? with the clip of the probe ground lead connected to the front panel ground? with the "noise" going drastically down when I disconnect the ground clip? wen the scope is not even able to trigger a so common signal? do you really think that all of this is normal?
The following picture shows how I see a 1 kHz, 1 Vpp sine wave generated by the internal Wave Generator and with 50% of screen Intensity; the probe pin tip is inserted directly into the WaveGen BNC central hole and the probe ground lead clip is connected to the front panel GND point: can you make any measurement out of it?
And I don't live in a zone particularly crowded of FM stations, I am at home, not in an industrial plant and NONE of my other scopes suffers this problem!!
Come on, Agilent, if this is the normal behavior of this scope, I prefer to stay with my old Tektronix 468 which, at least, allows me to make some significant measurement and I will return this beautiful, expensive toy to the vendor.
And, believe me, I will be very sad to do this because after five days of "betrothal" I am already fond of it and I would love to find a good excuse to "marry" it!!
> Maybe tomorrow I will show a way to get rid of the interference from FM radio stations…
Please, please do that because this could offer to me the excuse I am looking for...
Ciao and many thanks for your interest in this matter.
Franco
P.S. could you please tell me how, in your pictures, were the probes connected? i.e. probe tip open/grounded, ground clip floating/grounded/shorted to the probe tip, etc...
Thanks.
Web: http://www.flanguasco.org
Attachments
while I can't dismiss at all the possibility that "some" of the noise is coming from external sources (FM broadcast?), I still believe that the majority of it is responsibility of the scope itself : have you seen the first picture on my first post? with 0.4 V of noise over the 2.4 V Probe Comp square wave signal? with the clip of the probe ground lead connected to the front panel ground? with the "noise" going drastically down when I disconnect the ground clip? wen the scope is not even able to trigger a so common signal? do you really think that all of this is normal?
The following picture shows how I see a 1 kHz, 1 Vpp sine wave generated by the internal Wave Generator and with 50% of screen Intensity; the probe pin tip is inserted directly into the WaveGen BNC central hole and the probe ground lead clip is connected to the front panel GND point: can you make any measurement out of it?
And I don't live in a zone particularly crowded of FM stations, I am at home, not in an industrial plant and NONE of my other scopes suffers this problem!!
Come on, Agilent, if this is the normal behavior of this scope, I prefer to stay with my old Tektronix 468 which, at least, allows me to make some significant measurement and I will return this beautiful, expensive toy to the vendor.
And, believe me, I will be very sad to do this because after five days of "betrothal" I am already fond of it and I would love to find a good excuse to "marry" it!!
> Maybe tomorrow I will show a way to get rid of the interference from FM radio stations…
Please, please do that because this could offer to me the excuse I am looking for...
Ciao and many thanks for your interest in this matter.
Franco
P.S. could you please tell me how, in your pictures, were the probes connected? i.e. probe tip open/grounded, ground clip floating/grounded/shorted to the probe tip, etc...
Thanks.
Web: http://www.flanguasco.org
Attachments
I beg your pardon!
Franco
a few comments:
1. this forum is not "official" support... to get real support, please call your local contact center... in the top right corner of the agilent webpage, set your country, and then, nearby, you should see a link for "contact us." Sounds like everyone's scope is new... and therefor covered under warranty... we'll be happy to fix/replace the instrument(s) if they are not functioning properly.
2. Nestor: Based on the first screenshot, where no probes are connected, I think the inputs on your scope are fine. I don't see any measurements there, but the screenshot looks rather typical. When you hook up probe, yep, it becomes a great antenna. There is a chance that the probe comp output is bad, and that is whqt you are seeing, BUT, the frequency of the noise is so high and amplitude so large I rather doubt it, since it is a 800 Hz or 1.2 kHz or other slow frequency. As far as external interference... the aluminum foil would need to be thinker than the skin depth of the signal for the aluminum to effectively shield it... Also, the ground leads on these probes ARE NOT shielded. Further, on the DSO1000A series, you need to manually set the probe attenuation, but it does default to 10:1.
2. Franco: the noise on your scope is maybe ok. I am nowhere near one of my scopes and won't be for a while... call your contact center to find out what is normal... however, here is what bugs me: your trigger. you SHOULD be able to trigger on an edge, of course. Specifically, when you show the probe comp signal (square wave), that edge looks pretty clean and I would expect the scope to trigger on that in auto mode just fine. But when I see the output of the wavegen, the sine wave, the noise on that sinewave, that poor trigger doesn't surprise me. You can try doing a noise reject under the trigger menu... try increasing holdoff.... setting the trigger mode to normal... all the standard stuff. Further, it likely makes sense to try triggering on another channel, or other scope, to try and determine if the trigger circuit on CH1 is damaged.
3. Why do I see more noise on my oscilloscope when a probe is connected?
http://www.home.agilent.com/agilent/edi ... &id=744743
The typical noise floor of a digitizing oscilloscope is a few millivolts peak-to-peak. You can measure the noise floor of your oscilloscope by disconnecting all cables and probes from the inputs and changing the volts per division setting to the lowest possible value. Set the scope to Auto trigger and you will see a baseline of the scope's internal noise.
All currently shipping Agilent scopes have an "Auto Probe" feature that automatically identifies connected probes. When a probe is connected, the scope changes its attenuation setting so that it correctly displays the voltage coming from the probe. For example, a typical passive probe has a 10:1 attenuation ratio, which means that the probe divides the signal by a factor of 10 before presenting it to the scope. The scope compensates for this attenuation by multiplying the input signal by a factor of 10.
Since the oscilloscope cannot distinguish between the external signal and the internal noise floor, the noise floor is also multiplied by the attenuation factor. Therefore with a 10:1 probe connected to a scope channel you can expect the noise floor to increase by a factor of 10. For this reason it is best to use a 1:1 probe, an active probe, or a simple coaxial cable to view low voltage signals.
4. all these probes you folks have came with probe-tip-to-BNC-adapters (little barrel looking do-dads). These can help remove external noise... try them with a function generator, and set the f-gen output to be High-Z.
I´m really grateful with your post, really I appreciate that you spend time on this post. I know that this not replaces the official support, but the two words "see normal" it´s the only that I need. Measures have a lot of noise like I show in the screens and it cause that I open the post, to check it´s that is normal. When I use the aluminium foil put several sheets and grounded, but the noise persist. Anyway, the only noise it´s in the FM band and then I´m going to probe with a bare coaxial to Probe GND to see the noise and compare with the probe as antenna noise. The noise floor of the DSO with one 50ohm specific terminator it´s really negligible.
Exist any test that we can do with the DSO to discard that Franco suggest about an internal oscillation originated in the external noise. It´s only for discard all of the possibles issues exposed along the post.
PM: A question off this context. Why in some screens copy I see the lateral menu and in others no? I not like to see this menu in the images. It´s a minor detail but in the users manual not see anything about take the image with or without the menu. THANKS AGAIN AND SORRY!
KSmith:
I've connected today the Channel 1 with the Probe comp with a coaxial of 50 ohms and the noise persist but never greater than 10 mV (I think originated in the conection to oscilloscope clips of probe comp). I've connected the center of coaxial to the square signal and the shield to the GND. If I connect the center of coaxial to the GND of probe comp the noise continues almost without change. I think that the GND at probe comp it´s very noisy, more than other oscilloscopes that I use, but I don´t know if that it´s really a problem or only a little annoyance.
Can I do other tests that discard problems in the analog front end of oscilloscope?
THANKS A LOT FOR YOUR INFINITE PATIENCE
Franco,
I saw the pictures you posted. You are experiencing a little more severe interference than me but the essence of the problem appears to be the same.
In my test, the tip of the probe and the crocodile of the ground lead were connected to the ground point of the oscilloscope case. If the crocodile was kept connected to the probe tip, but both disconnect from the case ground (“floating ground”) the observed noise was a little lower, but not too much.
When the probe tip was connected to the CAL point (the output of a 1 kHz square wave for probe calibration), the oscilloscope displayed a noisy waveform, as shown below.
Again, the culprit was the interference by FM broadcasting. Note the thickening of the trace caused by interference from FM radio stations. Not surprisingly, the trigger was unstable. The trigger HF reject function had to be used to get a stable trace.
Attachments
My location is also very “quiet”. The most powerful FM station is about 15 km from here. Below you can see the FM band spectrum, as measured by a spectrum analyzer connected to a 1m long wire “antenna”.
The difference of frequency between the two strongest signals is 800 kHz, what is in agreement with the period of 1.2 microseconds of the envelope of the “AM-like” wave I presented in one of my previous post (see the third picture). This is strong evidence that the interference shown by the oscilloscope was indeed caused by FM radio stations.
Note also that the strengths of the signals from the two strongest stations are measured by the SA are only about 1 mV. How, then, an oscilloscope can display a interference much stronger than 1mV? This is the puzzle I will try to explain better in my next post.
Attachments
first of all many thanks for your kind and interesting reply.
Off course I have informed the Italian contact center, even before joining this Forum, and they have routed my problem to the Agilent Scope Support on Monday this week.
However they have forwarded to me a couple of pictures of how they see the Probe Comp signal on a DSO-X 5014 and the displayed waveforms look a lot better then mine (despite the relation wider band -> more noise)
It looks quite clean at 0.2 ms/Div but if you go down to 50 ns/Div (next screen shot) you see that the noise is causing multiple crossing of the trigger level: hence the poor trigger becomes very confused
Off course I have already tried all of that and it works on the Probe Comp signal but, you understand, this is not a viable solution for the problem: what about triggering a signal of, let say, 1 MHz?
The HF Rejection filter has a cut off frequency of 50 kHz and then will not allow to trigger on this hypothetical signal and the Noise Rej. additional hysteresis is not enough on a such noisy signal: the following screen shots show a 1 MHz, 500 mVpp sine wave, as picked out from the Wave Gen BNC, with the Noise Rej. and HF Reject options turned ON and OFF; by using the Holdoff it is, obviously, alway possible to find a position that stabilize (more or less) the displayed wave but it is so buried within the noise to make impossible any measurement.
This is, approximately, my noise too, good to know!! If you can, could you please be so kind to run a measurement (possibly on a OS similar to mine) of the noise floor with a 10:1 probe having both the tip and the ground clip connected to the instrument front panel GND? This will help me to dissipate any remaining doubts about the health of my beloved animal!
Dear Frank:
I have just seen your two last posts and ...
I start to believe that my problem too is mainly due to FM interferences: I will now run a few more experiments, e.g. comparing the position of the spectral lines, as displayed by the OS FFT function, to the frequencies received by a FM radio with digital counter; you know, I had hard time to get convinced because I NEVER had such a problem and (past) experience not always is a good advisor but often make me to stuck on old solutions.
I am, of course, anxiously looking forward to know your "way to get rid of the interference from FM radio stations".
Ciao and many thanks for your help.
Franco
Web: http://www.flanguasco.org
Attachments
The key word is RESONANCE!
Let’s take a look at the circuit of a 10X oscilloscope probe with a 15 cm long ground lead. If the probe tip is connected (grounded) to the crocodile clip of the ground lead, the probe can be modeled as below:
R1 - resistor internal to the probe
C1 - capacitor for probe compensation
R2 - oscilloscope input resistance
C2 - coaxial cable plus oscilloscope input capacitance
L1- ground lead inductance
V1 – voltage induced in the ground lead
A real probe has some a few components than showed in the diagram above. These components are used mainly to correct certain transmission-like effects. However the circuit above is still reasonably good to our purposes.
Note that L1 and C1 in series with C2 make up a serial resonant LC circuit. The effect on the resonance is to give a “gain” for certain frequencies. It is unfortunate coincidence that the resonance occurs in about 100 MHz, just in the FM band. The resonance can be seen more clearly if the probe is simulated with PSPICE. The simulated frequency response is as below:
As expected, for DC and low frequencies the gain is -20 dB (10X attenuation). However, after 30 MHz the response starts to increase and for frequencies around 100 MHz, the response is 20, 30 dB, even more higher than for low frequencies. It can be shown that the peak “gain” is equal to the Q of the circuit. The Q of a real probe depends on how the probe was made but make sure the Q can be pretty high.
The enhancement of the response for frequencies around 100 MHz explains therefore why the scope probe with a ground lead can be so sensitive to interference from FM stations.
You may also be curious to know why a spectrum analyzer doesn’t suffer from the same “noise enhancement”. The reason is that a SA has input impedance equal to 50 ohms, what is sufficiently low to damp any possible resonance. In the next post I will discuss what can be done to eliminate, or at least reduce, this annoying interference from FM stations.
Attachments
Well, I understand why in a SA the noise is not amplified, but in old oscilloscopes? Different internal impedance?
THANKS A LOT!
Enough of problems; let’s talk now about solutions!
As I tried to show in the previous post, the interference from FM radios is the result of induction in the ground lead of the probe. After the voltage is induced, it is then enhanced by the resonance between the inductance of the ground lead and the input capacitance of the probe. Let me say in passing that this resonance also causes ringing when the oscilloscope is used to see fast digital signals.
To eliminate, or at least, mitigate the effect of the interference, the ground lead should be made as short as possible. A practical way to minimize the inductance of the ground lead is to use the so-called ground spring. Many passive probes are sold with the ground spring accessory. If you never saw a ground spring, take a look at the picture at page 3 of this Agilent datasheet:
http://cp.literature.agilent.com/litweb ... 7111EN.pdf
Of course, the ground spring can be considered as a ground lead, but of very small length. This reduced length brings two benefits. First, the induced voltage in the ground spring by the electromagnetic field is much smaller than in the normal ground lead. Second, the reduced inductance puts the resonance at a frequency much higher than the FM band. This way, the enhancement of the interference from FM broadcasting is virtually non-existent.
To use the ground spring, the probe cap must be removed first, and then the ground spring in inserted firmly on the probe. The picture below gives an idea of the probe after insertion of the ground spring. As a reference, the TEK probe at bottom is with the normal ground spring, whereas the two others (a TEK and a Fluke) are with spring loads.
IMPORTANT NOTE: It is possible that even with a probe with ground spring you experience interference from FM stations when the probe is connect to the CAL terminal. A possible explanation would be that the CAL circuitry itself is suffering from induced interference, and transferring it to the probe. To solve any doubt, short-circuit the ground spring to the probe tip with an aluminum foil, and see if the noise seen on the screen is only the normal internal noise of the oscilloscope.
Attachments
Nestor,
Consider, first, that input capacitance and the inductance of the ground lead, and so the frequency of resonance, vary considerably from one probe to another. That makes that some probes picks up more interference than others.
Second, as I noted in my previous post, there is the possible that the calibration circuit itself is picking up interference, and transferring it to the probe. Did you notice that all the old oscilloscopes had a metallic case whereas the new ones are plastic encased? Probably the old oscilloscopes had better-shielded calibration circuits than many "modern" plastic oscilloscopes!
That said, the final test of an oscilloscope vertical amplifier regarding noise/interference susceptibility is to insert the ground spring on the probe, short-circuit the tip to the ground spring with a aluminum foil (or a metallic plate), and see the result.
It's true, all the oscilloscopes that I've used have metallic box connected to GND acting as Faraday cage. Sounds logical for me.
My use of oscilloscopes since beginnings was to observe degradation in digital signals and for years I use the GND spring only because it´s more practical when I test IC pins! MY MISTAKE!!! I not associate the short "antenna" of the spring with better performance of measure.
I do the test and all works like a charm.
Probe connected to the probe comp using the GND spring instead of crocodile. Look the noise at square bottom and it´s VERY reasonable. The trigger now acts perfectly and stable.
The same but connected the BW of channel to on, NOW the things are clear
And the final test, aluminium foil joining and shielding the GND spring and tip. Without BW limit the Vpp of noise it´s 5.6 mVolt and with BW 2.4-3.2 mV. Perfectly reasonable with a 10x I think.
GOOD!!! I´m totally convinced now! THANKS
I have tested your suggestion of using the ground spring instead of the ground lead with crocodile: following are the screen shots that I got.
Not knowing the Nestor screen Intensity I used 100% just to show all the noise.
Full Probe Comp signal with no BW limit (Vertical setting 500 mV/Div).
Enlarged Probe Comp signal with no BW limit (Vertical setting 20 mV/Div).
Enlarged Probe Comp signal with BW limit (Vertical setting 20 mV/Div).
The Base line, with probe tip short-circuited and wrapped to the ground spring with a aluminum foil, screen shot is attached to the following post.
My waveforms appear a lot more noisy than the Nestor ones.
What do you think? Can them be considered "normal"?
Ciao and many thanks for your help.
Franco
Web: http://www.flanguasco.org
Attachments
Base line with probe tip short-circuited and wrapped to the ground spring with a aluminum foil (Vertical setting 20 mV/Div).
Ciao again.
Franco
Attachments
Did you analyze the noise with FFT? After interference (e.g. from FM broadcasting) is eliminated, the spectrum of the scope's intrinsic noise should be reasonably flat (white noise).
I noticed that the vertical sensitivity was set to 20 mV/div. As a 10X probe was used, the sensitivity at the vertical input (at the BNC connector) should be 2 mV/div, right? Wrong! In the 3000 X-series, when the sensitivity is set to 1 or 2mV/div, the actual sensitivity is 4 mV/div! If you don’t believe me, read the note marked with **at the bottom of page 17 of:
http://cp.literature.agilent.com/litweb ... 6619EN.pdf
The trick used by Agilent is: when the sensitivity is set by the operator to 1 or 2 mV/div, the oscilloscope processor sets the sensitivity to 4 mV/div, and does a 4X or 2X zoom in the waveform acquired. Obviously, the perceived noise on screen increases by the same factor 4X or 2X.
This application note has a good discussion about noise in oscilloscopes:
http://cp.literature.agilent.com/litweb ... 3020EN.pdf
From the peak-to-peak noise measured for several commercial oscilloscopes, it appears that, indeed, the noise in your oscilloscope in the 20 mV/div is above average, even after you consider that the actual sensitivity is 4 mV/div. The discussion about “fat waveform” is also interesting.
Old analog oscilloscopes also suffered much from noise. This is a good reading on the subject:
http://www.national.com/rap/Story/0,1562,18,00.html
Yes, I did, this is the screen shot (probe tip short-circuited and wrapped to the ground spring with a aluminum foil) and the emerging noise around the 100 MHz frequency line is almost null (compare with the FFT that I posted before): what about the base line noise still present?
Yes, I knew that, before buying I read quite carefully the data sheet, but I wanted to compare, at least visually, my noise to the Nestor one (even if, I think, his scope doesn't have this zooming "feature").
On the other hand, if I look at the perceived noise amplitude shown in the above screen shot (taken at 50 mV/Div) I don't see it much different from that measured at 20 mV/Div: why?.
While I knew the two Agilent .pdf documents, I found this very, very interesting: thank you for linking!!
Ciao and many thanks again for your help.
Franco
Web: http://www.flanguasco.org
Attachments
Franco,
At least the shape of the FFT seems okay, now. The spectrum is reasonably flat and the interference has gone.
I would like to inspect the microstructure of the noise. Would you please post a screenshot of the noise alone in the following conditions?
Probe: 10X, short-circuited to the ground spring
Vertical sensitivity: 50 mV/div
Sweep speed: 50 ns/div
Sweep mode: single shot
Display brightness: normal
Here it is.
Thank you for your comments!!
Franco
Web: http://www.flanguasco.org
Attachments
I took a look at the screenshot you posted, but, sincerely, I couldn’t see anything especially wrong with it. I am inclined to say that the amplitude of the noise you measured was normal for a 100 MHz oscilloscope.
To have a basis of comparison, I measured the noise in my Fluke Scopemeter with the same settings you used. Let me say first that the difference of bandwidth (200 MHz for the Scopemeter vesus 100 MHz for the Agilent) doesn’t appear to be important, since when the BW in the Scopemeter was set to 20 MHz, the noise intensity changed practically nothing.
Note in the picture above that the noise peak-to-peak measured with cursors was about the same as in the screenshot you posted.
I took the opportunity to measure the noise when the vertical sensitivity was set to 20 mV/div (second picture). The noise peak-to-peak and RMS were 11 mV and 3 mV, respectively. I believe these values aren’t much different from what you would get with your oscilloscope.
To be sincere, I never liked much the waveforms displayed by digital oscilloscopes. I was shocked the first time I saw a “digital phosphor” Tek scope. The Tek’s trace was fuzzy, noisy, and much uglier than what I was accustomed to see with analog scopes. Digital oscilloscopes have improved since then, but in the “trace beauty” department, they still let a lot to be desired. Of course, digital oscilloscopes are invaluable for many applications, but they aren’t ideal when one wants to analyze certain details of complex signals. If you have a good analog oscilloscope, don’t disregard it.
Back to the original question, if you want a really clean trace, use averaging. Also, avoid setting the display brightness to 100%, or persistence to long/infinite, as these settings exaggerate the effect of noise, visually at least. After a while you’ll get accustomed to your DSO-X 3014A (a truly beautiful instrument), and you will stop paying attention to its normal noise. Of course, if you still have some doubt, take the scope to an Agilent representative, while in the warranty period, to see if the instrument is working as it should.
Attachments
I am now (97%) convinced too that there is nothing wrong with my scope: to clear even the remaining 3% I asked yesterday the local Agilent Service to visit their labs and to compare my scope to one of theirs: they have kindly accepted and I'll go to Milan Tuesday this week. I let you know the come out.
I won't!!! As I wrote in my first post the traces and the waveforms details of my, more than 40 years old, Tektronix 547 are sharp as razor cuts but it is just "a little bit" impractical to use today (and I have to repair it quite often); the other one, the Tektronix 468, although quite good in the analog mode, is too limited so far as Digital Storage.
Then I decided to buy this last one and true, it really is a beautiful animal, that kind of things that make to ask yourself how you could live before without
Ciao, Frank, and many, many thanks for all the time you spent helping me.
Franco
Web: http://www.flanguasco.org
OK, now I have to learn how to cope with it and I even have to thank Agilent because they made me learning something that I never encountered before!
Ciao and many thanks to all of you.
Franco
Web: http://www.flanguasco.org
I'm really grateful to all that answer and help in this post. THANKS