Analog scopes for noise measurements

Hello Frank,

Even more amazing: I almost pulled my hair out on an EMI case although I am pretty seasoned in the art. What on earth...? Then I had a hunch since I was using the client's spectrum analyzer and that one just looked too glitzy to me. A newfangled thing with flat panel display and the whole enchilada. It probably even had some function to pre-order a Starbucks Cappucino via the web.

Took the EMCO sniffer to it and, bingo. All the stuff I saw came out of that display. So, as weird as it may sound the drill changed to "eat chocolate and clean foil wrapper", "cover screen with foil wrapper", "measure and store", "uncover screen to look at results". I was thankful that the store button wasn't a touch screen function so I didn't have to do the "poke tooth pick through hole" routine :-)

Regards, Joerg

Most, if not all analog 'scopes have a dynamic range of more that one screen height, usually several. Trace width is usually quoted as 0.5mm

High end, specialized Y amplifiers, such as the Tek 7A13 plugin, can have an effective screen height of 10,000 divisions at 1mV per division.

--
"Electricity is of two kinds, positive and negative. The difference
is, I presume, that one comes a little more expensive, but is more
durable; the other is a cheaper thing, but the moths get into it."
                                             (Stephen Leacock)

Hello Fred,

Even not so high end Y amps do. My first scope was a Hameg 207 from the early 70's and I still have it. 8MHz BW, no trigger (!) but a "synchronizer". It's Y amp was very saturation proof. You could crank it up and almost see the electrons dance. The only downside it the teeny tiny CRT in it but hey, it still works like new.

Regards, Joerg

Interesting.

Thanks for the input!

--
_______________________________________________________________________
Christopher R. Carlen
Principal Laser/Optical Technologist
Sandia National Laboratories CA USA
crcarleRemoveThis@BOGUSsandia.gov
NOTE, delete texts: "RemoveThis" and "BOGUS" from email address to reply.

Did you ask him if it had ventricles?

Thanks! Rich

It used to be called "windowing". Manufacturers made a sales point of it.

The first digital 'scope I saw demonstrated, I asked the shiny sales guy how far I could window the display. He didn't know what I was talking about. He didn't seem to understand quantization noise, either, just kept going on and on about memory depth and features I didn't want.

--
"Electricity is of two kinds, positive and negative. The difference
is, I presume, that one comes a little more expensive, but is more
durable; the other is a cheaper thing, but the moths get into it."
                                             (Stephen Leacock)

No, but my final comments had quite a lot to do with testicles :-)

-- "Electricity is of two kinds, positive and negative. The difference is, I presume, that one comes a little more expensive, but is more durable; the other is a cheaper thing, but the moths get into it." (Stephen Leacock)

In digital scopes, a relatively new specification that says a lot about the error sources in each vendor's scope is "jitter noise floor".

Comparisons of high end (high BW - 8 GHz or >) scopes today have some with fractions of pico-second values. They all make measurements automatically with statistics. So X and or Y scale variances can be quantified and bounded with high degrees of confidence.

JS

Some/most digital scopes don't try to synchronize the ADC sampling clock to the trigger, so there's always 1 sample peak-peak time jitter inherent to the architecture. Fancier digital scopes measure the time between the trigger and the sample clock and compensate.

Equivalent-time samplers have had picosecond jitter almost from the very first. The HP185 had 4 GHz bandwidth and tens of ps of jitter ca.

1962, using tubes. The newer ones measure RMS jitter in fs.

You can get a working 1990-vintage Tek 11801 on ebay, with a sampling head, for around $2k, for 12-20 GHz bw and jitter of a few ps RMS. I sometimes buy broken ones for a couple hundred bucks if the seller reports the right diagnostic error, one we know how to fix. Somebody could make a nice living buying the duds for, say, $200 and fixing them and selling for, say, $2k.

John

Dead batteries.

John

What's an example of the easier diagnostic codes to fix?

Robert

Seriously? I've been known to rescue TVs and microwave ovens that only needed a fuse. Once I got a window air conditioner that just had sticky bearings - a little WD-40 and I was cool all summer! I had to get a fat extension cord, so there's something electronical about it. ;-) Once, I saw a 27" console TV in a dumpster. I went to take a look, and see if it was only the fuse, but the idiots who had tossed it, tossed it - literally - and it had landed on the neck socket. "FSssssssssssss....." )-; So I left it there.

Thanks! Rich

My understanding was that even the cheapest Tek and Agilent scopes you can buy today do this (compensate, in software, for the difference between the sample point and the trigger point). Anybody know for certain?

Neither my TDS2012 (100 MHz) nor my TDS3052 (500 MHz) does. I checked the current 3052B datasheet, and they don't mention timebase jitter at all, so I suspect it's still doing asynchronous sampling like my original 3052.

My 2012 samples at 1 Gs/s and has 100 MHz bw (risetime is just about

3.5 ns) but the fastest timebase is only 5 ns/cm. That trick hides the sample splatter pretty well.

The really high-end scopes, the $15K and up things, usually fix this and spec jitter.

John

Hi John,

Wow, that's a little disappointing! I stand corrected; thanks for the info.

DSOs sample continuously so they can display pre-trigger samples, so it's too late to change the true sample times once the trigger arrives. All you really need to do is quantify the delta-t between the trigger and the next (or in fact any known) sample, and fudge the display accordingly. That's not real hard to do.

John

Hello John,

I wonder why. I mean, couldn't they have used a smart oscillator design and then yank the oscillator into phase when trigger cometh? It'll mess up the first sample but the others would be in sync. We have done stuff like that in ultrasound more than 10 years ago.

Regards, Joerg

And you usually don't want the error sources of the scope in your measurements - so modern methods collect absolute values (periods, for instance) and simply present the statistics. Timebase error creeps in with ADC effective bits, preamp gain and offset, temperature, and other gremlin noise sources.

JS

[...]

Or better yet, record the phase of the sample clock at the time of triggering.

--
--
kensmith@rahul.net   forging knowledge