I own an DSOX2012A. I had as many others Agilent customers the same problems with big noise over the signal, or without input signal. The noise level in my oscilloscope was not exactly so big as described in another thread (about another type scope), but was big enough (3-4mV and also a clear frequency/pulses) to get me frustrated after I`ve paid the price for this device.
I started to take a close look and examine this problem. I found out the follow:
A big noise source in this oscilloscope is its main switching power supply. Is quite surprisingly that Agilent chosen a such PSU type for theirs products.
Another main noise source is the display and its internal power supply.
The important noise sources are the small power units too, spread on the entire main PCB and on the power PCB. All those small ferrite trafos should be shielded with appropriate magnetic shield material. Agilent used only aluminium plates in between the internal devices/PCBs in their oscilloscopes...
The input signal process/amplifier stage is very poor/inefficient shielded. The fan ( in DSOX2012) which run very near those input circuits is a quite important noise source for it.
My taken solutions: removed the main Agilent power supply, and replaced it with a serial regulator PSU (an transformer is now in its place and the oscilloscope GND is completely isolated from the main). Shielded the most of the small power supplies on the main PCB and power PCB. Efficient shield placed over the input circuits. Place a magnetic shield on the back side of the display. Ferrite beads on the all wires between the power PCB and the main PCB.
The result: 1mV noise (pure noise and no any frequencies or pulses) on the horizontal range down to 200ns. 800µV noise on 200 - 50ns range. 300 - 500µV noise in the last range of 50 - 5ns. This with a probe 1:1 on the input, shorted with spring GND. No any other interferences/frequencies from outside get in now. I do not like at all this resulted noise level, but is better than before, in my case...
I should expect that Agilent should think more professional in designing their products, specially about those quite simple solutions to minimise the noise. Should definitively not be the customer/the buyer who have to improve an Agilent product after one has already paid quite expensive price for it.
I started to take a close look and examine this problem. I found out the follow:
A big noise source in this oscilloscope is its main switching power supply. Is quite surprisingly that Agilent chosen a such PSU type for theirs products.
Another main noise source is the display and its internal power supply.
The important noise sources are the small power units too, spread on the entire main PCB and on the power PCB. All those small ferrite trafos should be shielded with appropriate magnetic shield material. Agilent used only aluminium plates in between the internal devices/PCBs in their oscilloscopes...
The input signal process/amplifier stage is very poor/inefficient shielded. The fan ( in DSOX2012) which run very near those input circuits is a quite important noise source for it.
My taken solutions: removed the main Agilent power supply, and replaced it with a serial regulator PSU (an transformer is now in its place and the oscilloscope GND is completely isolated from the main). Shielded the most of the small power supplies on the main PCB and power PCB. Efficient shield placed over the input circuits. Place a magnetic shield on the back side of the display. Ferrite beads on the all wires between the power PCB and the main PCB.
The result: 1mV noise (pure noise and no any frequencies or pulses) on the horizontal range down to 200ns. 800µV noise on 200 - 50ns range. 300 - 500µV noise in the last range of 50 - 5ns. This with a probe 1:1 on the input, shorted with spring GND. No any other interferences/frequencies from outside get in now. I do not like at all this resulted noise level, but is better than before, in my case...
I should expect that Agilent should think more professional in designing their products, specially about those quite simple solutions to minimise the noise. Should definitively not be the customer/the buyer who have to improve an Agilent product after one has already paid quite expensive price for it.
As I wrote in previous post in my work I have Tek and HP oscilloscopes and hasn`t the noise level from my home agilent. I read and notice about how reduce that noise but when I use the same methods in the other oscilloscopes the noise it´s reduced to imperceptible levels. In my scope the noise continues big but better that the first day thanks to Frank_BR. All of changes that you explain are reasonable and surprise me that Agilent not put almost the toroids and shieldings. I can understand the use of an switching power (a good one I presume), today electronic equipment comes with these power sources in 95% of cases, but not understand WHY agilent not shield properly the circuits than can generate noise. It´s a shame, I need to improve this isolation by myself????? I need to wait that the warranty expires, then I have a long time of use a noise instrument that can be much better with some aluminium inside!!!!!?????
Several years ago, when you buy an HP or TEK oscilloscope (even the most cheap of the models) you can rely on the price performance of the unit and buy confident in the first brands! Next time I buy a toy brand, pay much much less and expect the same results? The rumour of Rigol making the scopes to Agilent it´s finally true?
A VERY SAD FACT, blame Agilent. Maybe somebody from Agilent can argue why with two or three dollars in aluminium can be reduce the noise in the scope and Agilent not do that?
(I put the same comment in the previous post)
First of all to use a switching power supply in an oscilloscope is a non sense. I understood also that this is done only for the commercial reasons. Very cheep and not so heavy for the final product. But an oscilloscope is an measurement device, which have to be very precise. This is quite basic! The PSU Agilent use in theirs oscilloscopes, can deliver on its output a quite clean power. I don`t know, I did not measured yet that PSU output. The big problem in using switching PSU is not exactly its noise on output delivered power, but that it radiate very much electromagnetic noise around it. Agilent used aluminium shield between that PSU and the main board, where the signal input amplifier/processors are few cm away from that PSU... No any electromagnetic shield in between! I could detect clear frequency and spikes pulses generated/radiated of that PSU, with the shorted probe, approx. 20 - 30 cm away from both the back and front of the osciloscope... The same problem with the generated noise from the oscilloscope display its self... This can very hardly be understandable as professionalism in a such product design...
It seems to me that Agilent sell the DSOX2012 for approx 1500$ ( and pardon, but this is not a very cheap price...) as a toy or learning object for the very young students in the schools. It could be very well like this, but Agilent should specify/precise in the specs of this osciloscop model (as in others too) that is not about a precise measurement device, but a treinig object for the beginner students... Or it have to be précised by Agilent that this kind of oscilloscope is not meant to work or be trusted in measurements below 5mV signal levels...
I should want to have a clear answer from Agilent about how one can measure or visualise events in the range of 100µV - 3mV using the DSOX2012, or another one of their Infiniivision series. And please do not tell me stories about intensity or Z axis of the displayed signal. In today electronics it is quite much about low level signals. Right?
Thanks for the recommendation. It seems that reduce quite much the noise level... but in the lower range of the spectre. In the higher range is not so effective. Interesting also the "Average" filter... I will test more those filters in the future.
I think is not very important which company designed in fact the oscilloscopes for Agilent (if is the case...). It could be a different one (is quite normal) or can be Agilent it self. Is not this matter. As long as this product is sold under Agilent name/logo, then is an Agilent product. It is also very clear that an Agilent specialists team has reviewed, controlled, tested and approved a such design/product. As those oscilloscopes have design problems (very simple details which could improve enough theirs performances) is a quite big shame at last for the Agilent company. Else, nothing to blame (yet...) about their software, functions, and measurements facilities.
As I can easily see, the input amplifiers and the rest of the processing of the analogue signal was designed to go down in resolution in µV range (the circuits inside can do it well). I just wander what for if the noise level exceed this µV range, and the user can not trust very much to visualize/measure something under 5 mV level... Is right that one can turn on some built in software filters to get rid a part of the noise, but I personally can not agree that this is the right way to do it. Those filters have limitations, can alter the real signal, and is to be used in specials circumstances. But not as a standard measurements/visualize procedure.
As Agilent specialists could not see/think that applying very simple and just common sense measures in designing of this product, will make it quite easy more competitive on the marked, it is a mystery for me... Maybe the finally product would costs with 100$ more... So what? At the prices Agilent sell their products, a such costs/price increasing is nothing... A position in the marked is not only about prices and costs...
I started to use real oscilloscopes while performing my career in physics, using a HP or Tek guaranted me that I not have unwanted surprises when measuring and worry only in did the right thing, not think or worry about if the manufacturer of the oscilloscope could make the things right inside!.
A couple of months ago I was surprised with the excessive noise in my new very basic oscilloscope (buyed for my home use), and is true that if extreme care in measure is applied the noise is reduced, but even is much larger than in HP or Tek models of 10 years ago, even if I compare with oscilloscopes with higher bandwidth and features, and then more sensitive to noise. Actually I need to take precautions in my oscilloscope measuring 60Mhz signals that I take with 700 or 1200 Mhz signals in other scopes!
What happens with your model has caused me a shock. If in your model (by far superior to mine) Agilent committed these "oversights" I hate to think that was implemented in mine. Until your post and after my original post, I convince myself that in new oscilloscopes the inputs are more sensitive to noise or simple has less noise stabilizing circuits that the company uses 10-15 years ago in their oscilloscopes. I know too that these oscilloscopes was comparatively more expensive than the current equivalents. When the warranty expires I add the shields and review the isolating of input channels, but really, I buy Agilent ( and not Rigol, Owon or Atten for example) to avoid having to do this and feel confident with the equipment.
I agree too in I do not care who makes the Agilent equipment, but I assume that if I buy the stuff of a company of this trajectory, the characteristics of the stuff are keeping inside that I expect from an equipment with the Agilent logo. I not expect anything outside of the range of my oscilloscope, I like and want better performance but I not pay for it, but assume that the noise it´s the less possible for my input channels quality and at least lower than the minimum scale of the equip.
Now I really do not know what to think. I love my oscilloscope, I´m still confident in Agilent but I really hesitate about how different it´s a cheap Chinese oscilloscope from the cheap oscilloscopes of the big´s companies and if the next time the extra money matter.
The issue with display noise is often confusing: certainly there is some noise from to the scope (the fact is a scope is a wide-bandwidth instrument); another reason is most scopes are 8-bit, so there's 256 states (about one quantised level per pixel on the display)... however interpretation of the display and has to be considered. Old analogue scopes make the signal look nice and "clean", however they tended to be much lower bandwidth than modern digital scopes (so there's much less noise to detect, amplify and display), and the Gaussian nature of the phosphor display hid the noise in the signal - simply, the electron beam cannot deflect fast enough to represent the noise on a signal. Now add-in the fact that you had tiny 3" or 4" displays and modern scopes have 9-12" displays (which result in 10 times the screen area, so everything looks 10 times bigger).
Early generation LCD displays, which use data-point compaction, made the signal look really noisy. This is because 100s or 1,000s of data points along the horizontal are being represented by just one pixel "wide", but can be many pixels "high" on the LCD screen (e.g. 1,000,000 data points of memory when displayed on a 320 x 480 screen would mean over 2,000 datapoints per pixel - any difference in the amplitude of these datapoints results in a "thick" trace on the display with equal intensity, which looks like noise, but may actually be real data). Ways around this is to use a different display algorithm which parses or decimates the acquisition data or varies the intensity depending on the frequency of the signal (a patent protected by one manufacturer), or if you can set some parameters of the scope yourself: use less memory which reduces the sample rate and hence the digital bandwidth; or reduce the analogue bandwidth; or average the display.
Thanks for your deep and extended good explanation. You have for sure right. But I can not agree with your opinion about deflection in an old analogue scope (its cathodic tube) which has not enough speed to show the eventual noise. The phosphorus on the screen could be a problem, the deflection amplifier too, but the beam inside the tube is certain to be very fast. I can hardly agree with this story about old scopes which can not show enough the noises... There was old analogue scopes with many hundreds Mhz bandwidth, where the noise was lower then in those modern scopes... I think is quite normal that noise occur in a digital device. The designer of that device has to take in to consideration this fact and find solutions to minimise this. This is all about in this case.
What I actually meant in my previous post was about noise which come apart from the display too, from its internal switching power supply. This small PSU inside the display unit it self it provide power to the back light device of the screen, or so. This is quite noisy in the same way the main power supply in those oscilloscopes is: its radiate much enough electromagnetic field because missing or inefficient shielding. All those radiated frequencies and noises are induced in the nearby very sensible circuits. Like the signal input one, like the processing devices, processors, and so on. Very bad filtering/shielding, and missing ferrite beads on the power wires inside those Agilent oscilloscopes leads to quite high noise level over the analysed signal.
If Agilent should care more about its prestige and about they who buy theirs products, to put in place very simple measures to minimise the normal noise which occur in a digital machine, the consumers would not have to struggle whit this noise problem in a sold as a professional precision measurement device, which is happen the consumer pay a quite high price for...
I agree again with Coris. I use a Tek TDS 784d and it´s a VERY more sensitive oscilloscope with more bandwidth (it´s digital too) and has less noise than I see in the previous print screens. Off course it is by far costly than this scopes then I not compare them because it´s impossible, but I not see the noise that see in my cheap scope and in the screen copy of other new models like 2XXX and 3XXX series. I can remember the model of an old 400Mhz analog oscilloscope from HP that I use for years, but remember that screen it´s reasonably noisy and the bandwidth and sensitivity was higher than in this new scopes. This scope show correctly a 350Mhz signal, but you say that it´s not capable to show a 40-80Mhz noise?. I think that say that "Old analogue scopes...they tended to be much lower bandwidth than modern digital scopes..." it´s a totally unhappy affirmation.
In these old scopes (and others) I can connect a probe to it and observe less noise than in my 100Mhz Oscilloscope and again less than I see in the pictures from DSO2XXX. Where are noise come from? From outside? Sure, I can probe that in other post, but it´s impossible that ALL noise came from exterior because then in other oscilloscopes with similar characteristics I would to see the same noise with the same probe and that not happens.
If the internal noise it´s impossible to shield, well, that´s it. The noise level it´s one of the facts that separate the prices of scopes and put a margin between a cheap amateur equipment (like mine DSO1014) and a very professional scope. The misunderstanding (by me almost) fact it´s that apparently with few dollars in shielding you can reduce significantly the internal noise.
That´s the only fact here. WHY Agilent not put this shielding?. Only this, other comparisons or resolution and representation arguments are out of this question I think and only unclear the thread. Please, Agilent, explain why the PSU not have ferrites in the cables, not has shielding to avoid noise or LCD has not proper shielding (like my laptop do for example).
And a last question and only important to me:
Then DSO1XXX yes? I pay for a badged chinese oscilloscope and not an Agilent product?
http://www.youtube.com/watch?v=fGxusFBM ... ure=relmfu
It is very easy to remark that the main power supply has a very poor shielding, and so on...
To be honest with Agilent the oscilloscope seems to be correctly shielded and the owner of the blog says that quality it´s remarkable. But it´s true that the PSU cables hasn´t ferrites and that it´s bad, the input channels has separate shielding but the rest of board not and It´s not visible if the LCD has proper shieldding but you say that hasn´t. And more important, you can reduce the internal noise (the only that matters for me).
I´m wait any from the proper sector of Agilent that explain why miss this improvements or if this happens in first models and corrected in other models.
A question to you, you replace the power source with a new power source with a transformer?
I was a little confuse when I couldn`t see this thread earlier in the forum. I just thought that was deleted, but now I see that is again in its place...
It is right that in Agilent scopes the input channels are shielded in one way. In my case too.The point (my point) is that this is not enough for the input circuits sensitivity. On the other side of the board (back side of the channels) is nothing. Only the coper (GND) plane inside the board... I could see quite clearly that the channels took noise from the nearby fan, and from main power supply. Those noises was variable/modulated as one moved f. ex. the hand over that area... Only this phenomenon is just unacceptable for a such (priced) measurement device... I have now an 0,8mm steel plate over the original shield of the input channels, and the heat sink of the processors are now grounded too. The noise I have seen before is gone... I had a clear frequency noise/pulses, which I detected to be from the display it self (its internal switching PSU). I placed a steel plate 0,3mm under the display. That frequency noise was gone... And the most important, I`ve replaced the main switching PSU with an serial regulator feed it from a transformer. All the frequencies and good defined pulses (with shorted probe) are now gone... Before replaced the original PSU, I`ve placed ferrite beads on all the wires between PSU board and the scope main board. An important decreasing in channels noise it was clearly to be noticed...
What I have now, after those simple improvements it looks like a pure noise, which could come from the electronics it self, amplifiers, and so on. So, is nothing noise from outside. My next step is to use µmetal shield material and improve the placement of it. The actual noise in my scope is now decreasing as one goes high in frequency range (down in period of time base). Before I had a constant 2-3mV or even more noise over the whole range with a shorted probe. Now I have in the period upper range 1mV white noise, and in the lower period range 300-400µV noise. The conclusion is enough clear for me...
To answer your question: Yes, the replacement of the original PSU is now an quite simple (diy made...), but very effective serial regulator. This is connected to an transformer with grounded shield between windings. I`ve used the original main connector and its original set up/filter for the low AC in power (24v) from the transformer. The scope is now completely isolated from the main power/GND in the wall. Maybe in some circumstances this completely isolation is not quite well, but I do not intend to use the scope on high voltage measurements. As I know from very long time, an scope have to have the GND of the measured device. This is the case now with this my improved Agilent scope. By the way, I could see that 2 small caps are connected (inside original PSU) between the PSU isolated GND and the scope/main GND of the power input plug. This standard GND connection (through GND caps) for a switching main PSU I never liked...
http://www.eevblog.com/2011/02/15/eevbl ... -teardown/
http://www.eevblog.com/2011/02/21/eevbl ... -teardown/
I just found out in the last that Dave Jones in his blogg have noticed too that the Agilent scopes of 2000/3000 series consume 6W of main power when are off... Now I understand that this 6W goes in to that capacitor to get it hot at more than 40 degree when one not use the scope... This is just stupid!
So, the users of those type scopes have to be aware about this detail...
Attachments
cp.literature.agilent.com/litweb/pdf/5989-3020EN.pdf
By the way it's also interesting to note that on 2000 and 3000 series oscilloscopes the vertical gain ranges of 1mV/div and 2mV/div are simply digital zoom functions and do not change the input channel sensitivity (at least that's the way I read the data sheet). So it would seem that the 1mV/div and 2mV/div settings don't have much real value. Too bad.
I'm wondering if someone could please post a screen shot of what the noise floor looks like with vertical gain set to 4mV/div and no scope probe attached. Also, state what model scope is used for the test and make sure bandwidth limiting and any filtering is turned off. Thanks.
I`ve read earlier that note you mention. My personal opinion after reading this Agilent text is that is a kind of excuse, explanation/story to get the users/customer used with the idea that Agilent have quite high noise level on inputs of their devices, and is nothing wrong with this... Unfortunately, I haven`t read this and haven`t been deeper into this "story" before buying my Agilent DSOX2012...
Else, I can not fully agree with your opinion about a digital zoom used for the 1mV vertical range. As the noise increase on this range, comparing with the 5mV range, have to be about a changing of input sensitivity. Could be usefully that Agilent him self should clarify this (and the rest too...).
Unfortunately I didn`t think to take some screen shots before improving the shielding and the PSU in my scope. But I will come in the near future with same shots about how it looks the noise now in my device.
Overall, the worst situation now (at least for me) is that I can not fully trust the measurements (in low level signals) with this Agilent scope. I have to think twice about what I see on the screen... It is the signal or it is some (internal) noises? This is very bead when is about a measurement device! Don`t you think Agilent?
Before be personally aware about noise problems in low level signal in this my DSOX2012, I just used my bren new trusted scope to measure the parameters of some power supplies. I didn`t knew at that moment that the noise, high frequencies and all kind of spikes I`ve seen on the output of the PSU`s I`ve measured, was indeed coming from inside of my own scope... I concluded that was wrong with that PSU... Was actually not the case, but I was put in a quite delicate (unpleasant) situation...
In old days, when one had a quality scope to work with, one had fully trust in that HP, Tektronix, or what it was made by that device. Nobody couldn`t even think that a HP company, or Tektronix could be on the marked with a non high quality and very trustfully measurement instruments. That because the customers paied the price to have full trust in theirs measurements using a device made by a highly trustfully company. Now those days it seems that this trust is broken down... Like in this Agilent (noise level) case...
" ** 1 mV/div and 2 mV/div is a magnification of 4 mV/div setting. For vertical accuracy calculations, use full scale of 32 mV for
2 mV/div sensitivity setting."
I take this to mean that 1mV/div and 2mV/div settings is digital zoom. This makes sense because they say to use 32mV full scale for vertical accuracy calculations. 32mV is 8 vertical divisions times 4mV/div.
I was just about to purchase one of these scopes. Perhaps I will try to get a demonstration first.
Yes... it looks like you have right... :?
I fully agree: yes, we need answers and some explanations!
They should watch this forum I suppose, as a serious company which prioritise its reputation.... I can see in the forum that it come out about problems in another fields of measurement devices too...
I'm an apps engineer with Agilent. I support scopes. I won't comment on anything internal to the scopes.
A few points and 1 thing for everyone to try.
The 2000X and 3000X scopes are 100% designed in Colorado Springs by Agilent. They are made in Malaysia, by Agilent. Each is tested and calibrated on Agilent's factory floor.
The 4 mV/div setting is the most sensitive hardware setting. Anything below that is magnification in software.
averaging and high res acquisition modes can reduce noise, when used properly. They also have 25 MHz bw limit filters behind each channel that will reduce noise.
1 reason these scopes "look noisey" is because of the update rate: more acquisitions per second -> more apparent peak - peak noise, as seen by a a human eye. I know this is hard to swallow. I think I have a nice, simple way to prove this. Further, as compared with analog scopes, modern scopes typically have more bw, though certainly there are analog scopes in the 100s of MHz range. All else being equal (which is not the case), more bw = more noise.
Please try this:
Get a BNC shorting cap (NOT a load), and stick it over one of the channels. Using a BNC short prevent external noise from enteringt eh scope's front end and all that is left is the raw performance of the scope. Set the scope to 1 mV/div (or whatever setting you want). Make sure the attenuation factor is set to 1:1. Turn on a Vrms measurement. how much noise do you see? do you see anything that should not be there?
To get to 4 mV/div, set it to 5 mV, then press the knob in for Vernier/fine control.
Thanks for answer, I really appreciate your effort to answer us, BUT, none that you answer respond to our questions.
Why when the oscilloscope´s internal shielding it´s improved the noise is reduced? Agilent not optimize this point