Spectrum analyzer vs scope for radio experimentation

"John Larkin" wrote in message news: snipped-for-privacy@4ax.com...

Something you see often in their writings is how low-noise the things were, despite using obviously noisy components. Early schottky ("hot carrier") diodes, planar triodes, plasma tubes to switch the waveguides and protect receivers, etc. For all the noise that those components add, it's only a few dB total, and there isn't really any other source of noise. The big difference is basically no atmospheric noise in the GHz, so either you're seeing signal or your amplifier sucks and should be fixed. And the engineers of the day were already quite knowledgable about building low noise receivers.

Frequency drift wasn't even that bad; there are plenty of things today which still use resonators, which would've been one of the primary references available. All that takes is a good silver plated, say, invar cavity, machined to exacting specifications.

Things we take for granted were invented back then; PLLs would've been easier if they had dividers, since reflex klystrons are just big, hot VCOs. They might've done harmonic lock to crystal oscillators using harmonic multipliers alone; it's a lot of stages between 10.000 MHz and a few gigs, but the last few stages can be passive diode multipliers, and there are buckets of tubes that'll do VHF no problem. The last stage need only be a diode mixer to generate the DC offset to feed back into the error amplifier and klystron.

Yeah, that's not even terribly hard: start with a 10MHz reference oscillator, two 5x multipliers (not very good efficiency on each, but a tight filter on each will keep them harmonically clean), then amplify, filter and buffer the 250MHz a little bit, then send it into a pair of diode doublers, first 500MHz then 1GHz. You could even push 2 or 4 gigs with another stage or two, but as the remaining power drops exponentially with subsequent stages, you're asking a lot for more. All this was easily possible with '40s technology: I've seen 6BQ5s used in ARRL Handbook exciters and up-converters as high as, I think ~300MHz with deratings.

Tim

I just happen to be reading Tek "Fundamentals of Spectrum Analysis". On Pg 14: Another source of error when making noise measurements occurs in the detector and logarithmic circuitry. These two errors cause the measured noise to appear lower in level than the actual noise by the following factors: LIN display mode: 1.13 dB LOG display mode: 2.5 dB The DVM, being a linear device, will have the 1.13dB error.

There's also a graph for correcting measurements that are within 10dB of the spectrum analyzer noise floor. The measurement will appear higher than expected because the detector is measuring mostly system noise, not signal. I vaguely recall reading that some computerized spectrum analyzers compensate for this effect, but I couldn't find any references.

Good find, congrats. I have a small pile of various Fluke 4 1/2 digit meters, none of which are true RMS. However, I do have an HP 3400A analog RMS meter, which keeps me out of trouble.

On a sunny day (Sun, 21 Apr 2013 15:43:07 -0400) it happened Jamie wrote in :

A increasing HV wil usually leave focus and intensity pretty much normal. It is a common fault, measure the HV. I even had it with a new scope, just from the factory. Fixed it and reported it back, they were happy...

Problem is that a changed HV also changes your vertical sensitivity as Mike already pointed out. Needs recalibrating. Of course it could be anything else in the H amp, or some other supply.

The 8920A uses a thermal RMS converter, which is a lot cleaner than they solid state approach.

I used them to align video filters for Telemetry systems. The relative dB function made it easy to get accurate -3 dB points, and the HP 3325B programmable function generator made it very easy to get the exact frequency. I wanted to write software to automate the testing, but the department head couldn't see spending about $1200 for the hardware. It would have saved about 10K the first year in labor. :(

I have four working 8950A. I'd like to get hold of a working Tek

was

Sure they were. Have you seen a high end, modern receiver?

It took a lot of time to build and align those early systems. I've serviced '50s era RADAR systems, and they took a lot of work to get acceptable performance. The early '70s Westinghouse 2 MW systems at Ft Rucker had a staff of four EEs to keep them running, and they were only a couple years old. Tubes had to be sorted for performance, and everything was pushed to the limit to get acceptable performance. The transmitters & receivers could be swapped between systems, in the event of both systems being down.

If early designs were so damned good, nothing would have changed since the '20s

That's a really great app note. I should link to it on my website someplace.

A sine wave with RMS value 1 is

v(theta) = sqrt(2)*sin(theta).

Rectifying and averaging this from 0 to pi gives sqrt(8)/pi ~= 0.900, so an average-reading meter calibrated for sine wave RMS has to multiply the average value by a Finagle constant

F_sine = pi/sqrt(8) ~= 1.111 .

Gaussian noise with RMS value 1 has a probability density of

p(v) = 1/sqrt(2*pi)*exp(-v**2/2) .

Rectifying this gives

p(v) = sqrt(2/pi)*exp(v**2/2) H(v)

where H is the unit step function. To average this, you multiply by v and integrate from 0 to infinity. This has an average value of sqrt(2/pi)), so the required Finagle constant is

F_noise = sqrt(pi/2).

The error committed by using the sine-calibrated meter for measuring noise is the ratio of these

F_sine/F_noise = sqrt(pi)/2 ~= 0.886 .

To correct for this, you add 20*log(2/sqrt(pi)) ~= 1.049 dB .

I don't know where they get the 1.13 from, but it isn't quite right for pure average-reading meters.

Cheers

Phil

(...)

The current version of the app note is now 120 pages long: "Spectrum Analyzer Basics" The noise RMS/average correction factor is mentioned on Pg 66 and is now 1.05dB, which coincides with your calculations.

Drivel: I printed a copy which I suspect will make useful reading next week, should I be selected for jury duty. Sigh.

already pointed out.

I did some checking with it last night, the service manual is 670 pages and it is hard to look at on the screen when the schematic you need is towards the end and windows is dogging.

The vertical level is as close as I need it to be. I don't think it could be any better than it is. But what I did find was the H-blanking signal isn't correct and in that section, where it passes the H-scan gain circuit. There Is a PNP transistor there and I have my suspicions that it may have lost some beta, because it acts like it has very low gain in the circuit.

I also noticed the menus on the CRT are showing H retrace lines so this still leads me back to the sweep card.

This is a IFR 1500 service monitor. I'll check the HV, I'll have to scroll a few hundred pages to find ot what it should be so I can probe it. I am not really that concerned about it since it isn't worth much these days but It does make a good boat anchor in case you are in a hurricane!

Jamie

I finally twigged: that 1.13 is the ratio rather than the dB value--2/ sqrt(pi) ~= 1.1284 .

Don't worry, they never pick engineers for juries.

Cheers

Phil Hobbs

Except when the lawyers run out of preemptive challenges.

I was on a federal jury for just that reason, along with three medical types (RNs mostly). The issue was medical in nature, so this violates the first law of such things, that the jury cannot be too educated in the issue at hand.

While the plaintiff won, the award was a tiny faction of her dreams.

Joe Gwinn

That's true. I've sat on one jury[1], but was dropped for cause from three others. I think I've found the trick on how to do it. The attorneys always ask prospective jurors a series of questions, usually read from a script. I noticed that the attorneys were not paying attention or making notes until the last question, which was "what are your hobbies". I kept score and found that if a prospective juror answered something that takes minimal brain power, such as sports, couch potato, hunting, gardening, racing, etc. they were usually selected. However, if the answer was something more cerebral, such as ham radio, building anything, writing science fiction, computing, home nuclear physics, kitchen chemistry, etc., they were dropped for cause. The courts really doesn't want anyone that thinks, especially that thinks independently.

Makes sense, doesn't it? They want a jury that can be manipulated, influenced, scared, into whatever opinion they need. Independent thinkers will see through such schemes and might come up with original viewpoints, leaving the attorney speachless. Everyone will agree that a speachless attorney has a serious problem...

Jeroen Belleman

On a sunny day (Tue, 23 Apr 2013 10:36:03 -0700) it happened Jeff Liebermann wrote in :

Does etching your PCBs in the kitchen count as kitchen chemistry? If so I have th'm all:-)

???? Strange I would expect the opposite.

Yes, and I have the damaged brass drain plumbing to prove it:

Nope. An attorney friend once mentioned that he could predict the outcome of the case by the jury distribution. They made a big mistake when I served. I was foreman (because nobody else wanted to be it). The charge was voluntary manslaughter. The initial vote of the jury was almost unanimously for guilty. I thought differently and went through the evidence item by item demonstrating that everyone (including the police) either were lying or were hopelessly confused. I also demonstrated that the homicide did not occur in the manner contrived by the coroner. I had the entire jury turned around, except for one individual, who decided that the defendant was evil, a menace to society, and based on his convictions, should not be allowed to walk the streets. He even conceded that he had lied during jury selection about his preconceived opinions of homeless people. I pounded on this one person for about a full day, and gave up. Hung jury. The assistant district attorney was totally surprised, but eventually agreed with my analysis of the evidence. This jury was selected to be easily swayed, but they never considered the possibility of having it swayed by someone on the jury.

On a sunny day (Tue, 23 Apr 2013 13:33:10 -0700) it happened Jeff Liebermann wrote in :

Yes, good work. I sometimes read about people being innocent in jail for 10 years or something like that. No financial compenation can make that right, undo that wrong.

'Justice' is indeed often only for the rich or those who can afford it...

I did my own defence once, as described here in my reply to the late Jom Timpson:

PostPosted: Fri Apr 23, 2004 2:32 am Post subject: Re: Zap RFID tags? On a sunny day (Thu, 22 Apr 2004 09:48:12 -0700) it happened Jim Thompson wrote in : Quote: Looks like there's some *serious* brain-washing going on in Euro-land.

Particularly got a kick out of "without Europe no CD no DVD" and "we have our own oil / natural gass" ;-)

ROTFLMAO!

I think I can safely nominate Jan Panteltje as Best Village Idiot of

asked what I was arrested for... I will tell you: Killing a drunken Usenet poster.... While I was standing there with the knife, blood dripping from it, my suit all messed up of cause, the police came screeching in, sirens and all, and told me to reach for the sky. Decided to play along, and while the gallows were being put up, in that old courthouse, judge Killdare was at it. All 12 members of the jury had seen it happen too. Now you may wonder how I got away with it, as I am now writing this. Well, I did the Perry Mason voice thing... and did my own defense of cause. It looked really bad and people were crying, and this guy came up to me to get my size, for the coffin, you know. But fortunately there was an Internet connection in that court, and my request to use it was sustained. I did a quick printout of some of that victims (Think his name was Tom Jimpson or something), and handed it to the jury. They took only a short time, and unanimously decided I was not guilty. The judge looked a bit surprised: 'Ladies and gentlemen of the Jury, HOW did you come to that conclusions?' he asked... 'Well, from these postings we can see that the victim was brain dead. Since he was brain dead, he was already technically dead, and so was not killed'. The judge fumbled through his books for a long time, and then finally said: 'Justice has been done', and also had the relatives of that guy pay for cleaning my suit, as I was merely doing a service to the community, when I tried to dispose of the body of the braindead.

So there is a lesson here...

Copyright Jan Panteltje 2004 All Rights Reserved.

A defendant is considered innocent until he runs out of money for the attorneys.

Incidentally, your GPS jammer is obsolete. We now have MEMS "Jamming Dust". If your receiver seems to be picking up more noise than expected, try cleaning off the dust.

On a sunny day (Wed, 24 Apr 2013 08:19:02 -0700) it happened Jeff Liebermann wrote in :

Well for some reason it does not really worry me, Sorry non-disclosere in effect.

yes.. :-)

Much of modern hightech and modern hitech ideas are effectively killed by very simple means,,,,

Some old technology is insensitive to all that (old diesels engines are insensitive to EMP pulses used to stop cars etc).