Analog scopes for noise measurements

either is OK. There is no preference, it depends upon the bandwidth of the noise, ripple, and the bandwidth of your scope(s). Digital may not sample fast enough to get the noise correctly, but that means the scopes bandwidth is too low.

I've had my share of problems looking at noise (especially wideband noise) on a digital scope. The major issue I found (for older and not so old units) was *aliasing* - this is especially an issue especially when you are trying to find a fast glitch amongst otherwise relatively low frequency signals, but is always an issue when you deliberately use the widest bandwidth the [digital] scope has to offer.

Aliasing is a well known issue in digital sampling, although Tek (in particular) has made great strides to fixing the problem.

I've even had that problem on a LeCroy (what a budget we had there....), although not as pronounced.

I found I was chasing aliasing ghosts on more than one occasion when using a (in this case high bandwidth) digital scope. After changing the timebase a few times, and noticing the ghost signals appear / disappear depending on timebase, I talked to the vendor (Agilent) who made triggering suggestions to minimise the issue.

Still, the bottom line is that digital scopes have their foibles that one should be aware of, especially when looking at wideband low level analog signals.

Cheers

PeteS

This is not necessarily the case, however the visible display of an analog scope includes wideband information that can be more readily perceived, characterized and usefully interpreted by human beings, with fewer sources for gross error.

Noise, produced by various effects that are not always coherent, can be missed by a digital scope, or be displayed too easily in a misleading or even random manner. Sometimes this can be avoided by making multiple measurements at differing time scales.

Wide-band digital scopes have traditionally been more expensive, with difficult or completely useless triggering. This situation is being corrected. Some of the newer scopes from tektronix and others do a pretty good job at duplicating analog display persistance effects, and provide reasonable triggering capability.

By far the greatest problem with power supply noise measurement is probing methods - common to both scope types.

Knowledge of the limitations and capabilities of any equipment is a powerful tool in avoiding error. Never buy or use (critically) a device you have not either proven or seen to be proven adequate for the job.

RL

Digital scopes are nice because of their variable/infinite persistance and ability to compute true RMS noise. And nice color displays. And data export.

Older plugin-type analog scopes (545-series+1A7, 7000-series+7A22) can have huge common-mode rejection, 10 uv/div sensitivity, and selectable high/low cutoff frequency, which are all very handy here. Tek does have nice external amp/isolator boxes that add this capacility to a digital scope, along with full ground isolation.

John

Then how would you have known to look for it on the analog scope?

What is cool about the DPO and it's sophisticated triggering settings is that you can set it up to look for a glitch that you hypothesize might be the problem, and then walk away for 2 hours rather than having to have your eyes glued to the screen.

Good day!

The new Agilent 6000 series does digital phosphor display even better than the Tek.

Good day!

Hi Ira,

the basic idea behind using a analog scope for noise measurements is: Their screens supply a three-dimensional information. Beneath x and y there is beam INTENSITY as the third information. Consider a noisy dc signal. Having set a analog scope's beam intensity and vertical amplification to the correct values you will see the dc component as a horizontal line and the noise as a 'band' centered around the line. The intensity of the band will resemble pretty well the noise's amplitude probability distribution and for that reason the band displayed on a analog scope's screen is good measure for noise.

Note that this effect is due to specific 'after glow' properties of the phosphor inside the display tube. Due to the after glow the phosphor performs kind of 'averaging over time' which translates probabilities into intensity. In contrast to that the raster screens of MOST digital scopes a basically television like and are optimized to have NO after glow because they are expected to display a lot of independend pictures per second. You will never get this 'band' display realized on a NORMAL digital scope.

I said 'MOST' and 'NORMAL', because TEKTRONIX have started to build what they call DPOs = Digital Phosphor Oscilloscopes. In these scopes intensity is again available as a third dimension of information, not by means of analog after glow but with lots of clever digital electronics. I do not know how well such a thing would perform on noise measurements but if you consider using a digital scope you should call for a DPO and nothing else. Because it is done with digital electronics, they can use color in stead of intensity and also reverse things: Signals having a high probability can be displayed dark and signals having a low probability can be displayed light. Pretty well suited to find 'glitches' in a otherwise repeated signal.

Best regards

Ulrich Bangert

"jadaha" schrieb im Newsbeitrag news:42e71755$0$6700$ snipped-for-privacy@authen.white.readfreenews.net...

for

ripple,

The RMS thing is the big reason why DSOs are neat for noise. But DSOs are inherently more noisy. So you can barely do sub-mV RMS noise measurements.

The new Agilent 6000 series has a 1 sample averaging function which effectively increases the ENOB of the sampling system for slower timescales (but works to much faster timescales than their older series).

This makes it possible to get down to about 50uV RMS resolution. The DC offset can be a pain, but the scope can be user-calibrated which brings it down, or you can measure RMS and average simultaneously, and correct out the offset.

My wife knitted me a pair!

?!?!?!?!?

The effective resolution of an analog scope would be the "fatness" of the trace divided into the height of the screen.

Hardly 2^22.

And DSOs are usually only 8-9 bits, with some using oversampling techniques at slower timescales to get effectively up to about 12 bits.

Good day!

That depends on how well the vertical amplifiers handle overload. If you can jack the peaks up to 20 or 50 times the visible part, and still retain good response for the stuff that is low enough amplitude to remain on the screen, then you are effectively pushing up the resolution. This can be a big difference between two scopes that otherwise have similar specs.

Hello Terry,

That is exactly why fast analog scopes have and will for a long time to come have their place in the lab.

Regards, Joerg

Hello Chris,

The nice thing about analog scopes is that you don't have to know. All you need to know is that something is wrong.

These are nice. However, if you don't know what to set the trigger for because someone else designed the circuit this may not help. You could end up with nothing in memory or a memory overflow of screen shots that don't tell you much.

In the same way I often use a communications receiver to diagnose tough EMI issues. It can find stuff that even a >50k a spectrum analyzer is unable to resolve. I used it so often that I wore down a pair of headphones. Now I have to find out where to get new foam pads :-(

Regards, Joerg

I once spent an afternoon in the lab, with a blanket draped over me and the 200MHz analogue CRO, intensity cranked right up, looking for a glitch I suspected was there. Eventually spotted on after a couple of hours, proving the problem lay with a piece of programmable logic. 10 minutes later we had a slow-scale example demonstrating the behaviour up and running, and a fix about 20 minutes later (turned out to be metastability).

A Tek rep came by with a DPO a few weeks later, and using some of the sexy triggering features we managed to actually trigger on the glitch. But had we not known it was there, we wouldnt have been able to set up the triggering.

Cheers Terry

In message , Ira Rubinson writes

Digtal scopes digitise with 12 to 16 bits which limits dynamic range. A good analog scope has an effective 22 bit range .

Hello Chris,

Thanks! Now that is a great idea especially since this headphone is from Europe and there are no parts available here.

Regards, Joerg

Hello Chris,

As John said you can overdrive a good scope to the hilt. Then reduce brightness, close the curtains or blinds, turn off the lights, put on the glasses and you might eke out another bit or two.

The DSOs I worked with including the expensive kind didn't seem to have anywhere near nine effective bits. Oversampling doesn't help if you are after a tiny runt pulse that happens once in a blue moon or randomly. For EMI work I hardly ever use a DSO.

Regards, Joerg

Hello Terry,

Chris' wife knitted him new pads and that seems like a great idea. But I wore something else down and that's a really nasty problem: The bearings of the encoder that dials in the frequency.

Regards, Joerg

bingo

If you know enough about what you are looking for, sure. In our case, we didnt know what the problem was, only that there was a problem. There was no way we could have set up the DPO triggering without this knowledge, although we could have used the DPO persistence to emulate what I did with the analogue scope.

nice.

Cheers Terry