possible to overclock the sweep speed on a hp70000 (swept) spectrum analyzer?

[CopperCone]

So the minimum sweep speed is 10mS. Is it possible to use a external ramp generator to increase the sweep speed past design limits?

What are the expected effects of going a order of magnitude or two faster then what it was designed for?

how much of a engineering constraint vs "we don't want to sell something that can mislead the customer with 30dB of gain error" is it?

how much more speed is possible before it starts just giving complete bullshit readings (i want to see if a powerful but short duration intermittent signal  can be captured)?

are these errors primarily magnitude or frequency errors or does the sweep speed increase eventually develop multiple errors? Can these be overcome by hard band limiting by use of signal filters (i.e. designed for the analyzer display) and lots of preamplifiers?

[KE5FX]

So the minimum sweep speed is 10mS. Is it possible to use a external ramp generator to increase the sweep speed past design limits?

What are the expected effects of going a order of magnitude or two faster then what it was designed for?

how much of a engineering constraint vs "we don't want to sell something that can mislead the customer with 30dB of gain error" is it?

how much more speed is possible before it starts just giving complete bullshit readings (i want to see if a powerful but short duration intermittent signal  can be captured)?

are these errors primarily magnitude or frequency errors or does the sweep speed increase eventually develop multiple errors? Can these be overcome by hard band limiting by use of signal filters (i.e. designed for the analyzer display) and lots of preamplifiers?

The resolution bandwidth filter will 'smear' the observed spectrum at higher sweep rates.  You can observe this yourself by overriding the automatic sweep time selection while looking at a given signal.  In general, as observation time decreases, frequency uncertainty will increase.  This is as true of an FFT-based SA as it is of an old-school analog model, but the FFT-based model has the advantage of hundreds of independent "filters" instead of just one.

With the older analog SAs, the usual workaround is to set the analyzer to zero-span mode, so that it doesn't sweep at all, and then feed the wideband IF output signal to a separate FFT analyzer.  This gets you the best of both worlds.  Back in the 1980s, Tektronix had an instrument called the 3052 that did just that:



I imagine HP did as well.  Today, you might achieve the same effect by connecting an inexpensive software-defined radio module to the 70000's IF output.  I think some of the 70K IF sections had 321.4 MHz output jacks with 10 MHz or so worth of bandwidth.  That would be a good match for almost any SDR, even the cheap USB dongles.

[CopperCone]

Hi, KE5FX, I'm actually confused about what you said. It is exciting to see you on the forum.  I tried to use your SSM software but my Japanese GPIB card is not supported despite playing around with the configuration utility. I am debating ordering a national interments one (I know the one I have works because I managed to hail my 34401A).

Can you clarify what the benefits are to connecting a FFT device to the IF output. I am waiting on a ultra wide bandwidth if section in the mail right now. It is the wideband section with the 324.1MHz output.

Does this give you real time performance over a non zero frequency band, or just the benefits or waterfall etc?  Like, do you see whats going on around the center frequency, or is it still zero span? I tried connecting a rigol FFT scope to my 2MHz IF output on the front of one of the IF cards, but I can't make heads or tails out of whats going on. Does adding a SDR give a "real time bandwidth" capability to the device? I phrased the question 3 ways because I am not sure how to ask it. (like, so you have a "real time probability of intercept of signals ZZZ uS").

I'm interested in hunting down a narrow bandwidth infrequent but powerful pulse.

The 3d mode on the 70000 is really nice BTW.. interesting to see the RF 'landscape'. Different then what your program displays since the traces smear off at an angle.

Also, how bad is the smear? Fast sweep of a 10GHz bandwidth connected to a bicone is different then what you propose, I think. Like, imagine trying to hunt down a frequency hopping device transmitting short bursts on different frequencies (or perhaps a digital logic chip (mcu, cpu, complex system) generating narrow band short duration interference infrequently cross a wide span). So the probability of intercept is basically 'blind luck' with conventional speed.  But lets say this might be happening all over the spectrum analyzers input range.
Based on what your saying, I imagine the HP70000 will degenerate to a cheap 20$ RTLSDR dongle at some point, if swept fast enough, with spurs, etc... but I am interesting in metering the instrument. Perhaps some 'happy range' can be found. And of course it will sweep to 26.5 GHz..

Is there a convenient port or modification point where a fast sweep signal can be injected? I am willing to break traces and reroute 'plumbing' or even build distribution circuits. I am quite apt at stuff thats not RF.


 am a noob at spectrum analyzers and RF. Normally I like precise low frequency stuff, control systems, protection systems, analog op-amp Jim Williams stuff.

[KE5FX]

Hi, KE5FX, I'm actually confused about what you said. It is exciting to see you on the forum.  I tried to use your SSM software but my Japanese GPIB card is not supported despite playing around with the configuration utility. I am debating ordering a national interments one (I know the one I have works because I managed to hail my 34401A).

The NI GPIB-USB-HS adapters being sold on eBay lately seem to be good units, from what I've seen. 

Can you clarify what the benefits are to connecting a FFT device to the IF output. I am waiting on a ultra wide bandwidth if section in the mail right now. It is the wideband section with the 324.1MHz output.  Does this give you real time performance over a non zero frequency band, or just the benefits or waterfall etc?  Like, do you see whats going on around the center frequency, or is it still zero span?

Basically, you'll be able to do everything shown in the Tek 3052 video.  When you set the HP 70K to zero-span mode, it will stop sweeping and act as a 'dumb' RF downconverter.  The center frequency will be translated directly to the IF.  If the IF output is 10 MHz wide, for instance, and you tune the analyzer to a center frequency of 1000 MHz, then the IF spectrum from 316.4 MHz to 326.4 MHz will contain a copy of the RF spectrum from 995 to 1005 MHz. 

I tried connecting a rigol FFT scope to my 2MHz IF output on the front of one of the IF cards, but I can't make heads or tails out of whats going on.

If you didn't put it in zero-span mode, it would have been very confusing indeed.  

Does adding a SDR give a "real time bandwidth" capability to the device? I phrased the question 3 ways because I am not sure how to ask it. (like, so you have a "real time probability of intercept of signals ZZZ uS").  I'm interested in hunting down a narrow bandwidth infrequent but powerful pulse.

Yes, you will get either the real-time bandwidth of the SDR or the IF bandwidth at the 321.4 MHz output jack, whichever is less.  POI will be 100% as long as the SDR's firmware and software is fast enough to process the entire IF bandwidth. 

The 3d mode on the 70000 is really nice BTW.. interesting to see the RF 'landscape'. Different then what your program displays since the traces smear off at an angle.

Yep, that's a real waterfall display.  Lots of people (including me) use the term "waterfall" to refer to a 2D spectrogram but that's not really correct.

Also, how bad is the smear?

A good article on the subject is in the June 1991 HP Journal, where they describe the strategy used in the HP 3588A to get the best performance possible out of a swept SA.

Fast sweep of a 10GHz bandwidth connected to a bicone is different then what you propose, I think. Like, imagine trying to hunt down a frequency hopping device transmitting short bursts on different frequencies (or perhaps a digital logic chip (mcu, cpu, complex system) generating narrow band short duration interference infrequently cross a wide span). So the probability of intercept is basically 'blind luck' with conventional speed.  But lets say this might be happening all over the spectrum analyzers input range.

If you are willing to sacrifice amplitude and frequency accuracy, you can sweep quite a bit faster than the default automatic rate selected for a given RBW.  But when you really want the best probability of intercept achievable, you don't want to sweep at all.

Based on what your saying, I imagine the HP70000 will degenerate to a cheap 20$ RTLSDR dongle at some point, if swept fast enough, with spurs, etc... but I am interesting in metering the instrument. Perhaps some 'happy range' can be found.  Is there a convenient port or modification point where a fast sweep signal can be injected? I am willing to break traces and reroute 'plumbing' or even build distribution circuits. I am quite apt at stuff thats not RF.   I am a noob at spectrum analyzers and RF. Normally I like precise low frequency stuff, control systems, analog op-amp Jim Williams stuff.

For the application you're talking about, it would definitely be best to avoid sweeping at all.  It is not worth hacking the 70K analyzer to go faster than 10 ms.  The wideband IF plugin combined with an external SDR is the way to go.

[CopperCone]

Ah, this makes sense to me now. Yes, I did not have it on zero span mode. Not to mention the rigol had triggering problems/fft delays,

So, this is also true of the other outputs right? Mine has a 2 or 3 MHz IF output, a 21MHz IF output and soon the 300MHz IF output.

So long the resolution BW is under the IF output bandwidth, I will get that much bandwidth around the center frequency. The 300MHz section has ~100MHz of real time bandwidth in 0 span mode.

I wonder if it is possible to use DSP + a sweeping signal to some how save time/make some sort of gain, rather then relying on zero span mode. I guess this translates to some kind of smear, but maybe there is a benefit?

I am still curious as to what it will do if 'overclocked'.. so long its not 

[CopperCone]

How does the quality of the SDR effect things at this point? Sample rate, noise floor and spurious emissions? Am I missing anything else to compare them by ?

[KE5FX]

Ah, this makes sense to me now. Yes, I did not have it on zero span mode. Not to mention the rigol had triggering problems/fft delays,

So, this is also true of the other outputs right? Mine has a 2 or 3 MHz IF output, a 21MHz IF output and soon the 300MHz IF output.

Yes, although the levels will be different at the various IF outputs, since they are connected to different IF stages.  The 21 MHz IF output might be wider as well; I'm not that familiar with the 70K. 

So long the resolution BW is under the IF output bandwidth, I will get that much bandwidth around the center frequency. The 300MHz section has ~100MHz of real time bandwidth in 0 span mode.

That should work very well, then.  The RBW setting will have no effect on the bandwidth at the wideband IF output jack, since the resolution filters are implemented at lower IF frequencies.  In the analyzer's block diagram, the 321.4 MHz wideband output will be very close to the front-end mixer.

I wonder if it is possible to use DSP + a sweeping signal to some how save time/make some sort of gain, rather then relying on zero span mode. I guess this translates to some kind of smear, but maybe there is a benefit?

That is basically what they did in the 3588A.  I edited my earlier reply to include a link to the June, 1991 HP Journal article that describes how it works.  Again, though, this was done at a time when realtime FFT analysis bandwidth was very limited.  If you were building the 3588A these days, you'd digitize the whole 150 MHz range directly.

For their new SM-200A, the Signal Hound guys may also be doing something similar.  They're advertising a sweep rate of 1 THz/second at 30 kHz RBW.  The particular FFT window type they use has mathematical properties that allow faster sweep speeds than would normally be possible.  If you wanted to optimize POI in a swept spectrum analyzer, I'd suggest reading the HPJ article to see what the fundamental compromises are, and then looking into the Nuttall window.  It's unlikely that just cranking down the sweep ramp time will be a good idea.  Microcomputer-based systems in the 1980s and 1990s were often designed with all kinds of timing dependencies and performance-optimizing tricks that don't always scale the way you'd like.

[KE5FX]

How does the quality of the SDR effect things at this point? Sample rate, noise floor and spurious emissions? Am I missing anything else to compare them by ?

Whatever the specs are, is what you'll see.  If all you're doing is looking for an intermittent signal, only the sample rate and real-time BW are likely to be of concern.

[CopperCone]

I will take a look at that. I also have a digitizer for my hp70000, but I am waiting on Tee-connectors to try it out. It is not very impressive (20MS/s), but I will see. I think its only 10 bits too. HP 70700A

[CopperCone]

Interesting feature of my 8559A is a 1uS sweep speed. I'm not sure if its locked out on non-0 spans..

[rickells]

hi,

Please note the start of a Thread for the HP 70000 or HP 70K MMS system & modules:
Future discussion hopes to include: Configuration, Use, repair, software, etc... to further understanding.
HP 70004a, HP 70205a & HP 206a Display units + the HP 70001 Mainframe & a host of HP 70xxx modules.
See:
        https://www.eevblog.com/forum/testgear/hp-70khp-70000-modular-measurement-system-thread/msg1420425/#msg1420425

thank you