Used Spectrum Analyzers for under $1k

[jimon]

Looks like it's more sane to buy DSA815, then spend on used bulky equipment. I see only one issue with DSA815 : it starts from 9 KHz, so you can't really use it for audio applications =(

[KJDS]

Looks like it's more sane to buy DSA815, then spend on used bulky equipment. I see only one issue with DSA815 : it starts from 9 KHz, so you can't really use it for audio applications =(

I struggle to see how they could get the phase noise that poor.

[nctnico]

The Advantest R3131 could also be a good option (9kHz to 3GHz).

[Wuerstchenhund]

Looks like it's more sane to buy DSA815, then spend on used bulky equipment. I see only one issue with DSA815 : it starts from 9 KHz, so you can't really use it for audio applications =(

Well, I see two other issues: the DSA815 goes to 1.5GHz only when the OP wanted 1.8GHz, and the price is roughly $1500 when the OP's limit is $1k.

[KJDS]

The Advantest R3131 could also be a good option (9kHz to 3GHz).

They're good units, and the last few I had sold quickly, but I see very few of those for sale in the wholesale market.

[andrija]

How good are HP 8562A ? They appear to be about 15 years old and should still hold up well; they aren't quite boat anchors either.

[nixfu]

A Spectrum Analyzer is probably the last bit of bench gear, I have always wanted by have never gotten around to getting.

And, now I can't decide if I should get a used HP/Agilent or just get a RIgol.

The used HP/Agilents are going for about $1500, and you can get a new Rigol for the same thing with the tracking generator included, lots of computer connectivity etc..

I think the prices on the HP/Agilents need to come down.  It wont be long before people realize that there is a better option at that price level than the old HP boat anchors.  If the prices come down to $500-$1000(with a TG included) maybe I might consider a good used one, otherwise I will plan on getting a new rigol. 

Heck in a coupe of year, Rigol will probably have new models that blow the existing ones out of the way and the used gear will look even more expensive in comparison.  You might even start to see some used Rigols on the market too pushing the prices down.

[Rupunzell]

Very little wrong with the hp 141T system, had a system decades ago complete with storage normalizer, tracking gen, preselector and several RF plug ins including the LF plug in. It served fine as the lab SA for a very long time.

It is so very easy to discount and discard the capability of older SA systems like this when in fact the still are extremely useful and good instruments in every way. One thing memorable about the hp 8555A, 0.01 to 18 Ghz plug in was low noise and the signal ID switch, even if one had to keep track of the numerous spurious response blips that appeared which had to be verified as real.

While newer SAs offer better performance in some areas and automation with abilities to interface to a computer, these features do not make them automatically better than older SAs like the 141T system. It depends on what the measurement needs are.

As for fragile SA inputs, user needs to be aware and careful to prevent input mixer and related damage. Often this is just not that difficult to prevent zapping the SA input.


Bernice

Dang, I feel old.. more like a dinosaur, having a HP 141T 

[Rupunzell]

Curious, how bad is the phase noise (real world conditions, not as spec'ed) is the DSA815?

Sweep-ed spectrum analyzers are not the ideal choice for audio-low frequency work. There are far better ways and instruments for that measurement need today.


Bernice

Looks like it's more sane to buy DSA815, then spend on used bulky equipment. I see only one issue with DSA815 : it starts from 9 KHz, so you can't really use it for audio applications =(

I struggle to see how they could get the phase noise that poor.

[Electro Fan]

Thanks for the good discussions – it’s very informative and helpful.  I appreciate all the posts and hopefully they will help others here read or search in the future for spectrum analyzer advice.

Regarding the budget, this is for an amateur/educational endeavor so the budget is driven by what is feasible/practical and not by an absolute figure ($1k is just a benchmark).  The budget could be higher and of course it would be nice to go lower but it isn’t going to provide a financial return on investment (as a professional application would) so this spectrum analyzer goes in the nice/want to have bucket vs. the need to have bucket. - So maybe we could set the price aside for a while and look at possible specs and features.  Plus, others might have a different budget so I’m inclined not to rule out anything on price just yet – at least until I can better understand the price/performance tradeoffs.  Having said that, in my case it’s hard to imagine spending over $2k and at that I would have to have my head examined. 

Regarding bandwidth, 1.8 GHz is not an absolute, it could be 1.5 GHz or elsewhere in the 1-2 GHz range.  However, if it were possible to get to 2.5 GHz that would reach the Wifi range and if BW could get to 5 GHz that would be even better for Wifi.  (Although perhaps with harmonics even more BW is needed?) Obviously, as the bandwidth goes up so does the price – but let’s set the price aside for the time being.

Regarding RBW, it would be good to get smaller increments than 9 kHz.  Getting to below 100 Hz would be great to cover audio.  On the other hand, if the price uplift (I know I said we were going to set that aside) is too great to cover audio then maybe a separate solution (PC sound card?) might make sense for audio.  I had a sense for these tradeoffs before my original post but all the discussions have now surfaced these tradeoffs and more with better clarity.

Regarding possible equipment it seems there isn’t a lot confidence in some of the vintage models.  The Tek 49x series looks cool but it doesn’t get a lot of love here.  Likewise for the Tek 27xx series?  Perhaps as avvidclif said "Tek make's scopes, HP makes everything else".  I’m not ready to rain on Tektronix’s parade (I love their analog scopes) but I get the idea that not everything has to be Tek.

The HP units seem to come in at least 3 possible series:  856x, 859xE, and 859xL. 

There is also Wuerstchenhund’s Agilent E7595A/B which looks very interesting; I’m inclined to set it aside for a moment and declare it a finalist that can be compared against one or a few other finalists.  I’m also inclined to put the Rigol DSA815-TG in the final consideration bucket too.  Let's put Rupunzell's HP 141T on the finals table too - it looks gorgeous in a vintage kind of way.   So the question is what other used equipment might outperform these two models on value (ie, price, functionality, and performance; reliability will also ultimately become an important consideration to be made – but for now let’s assume that a good working condition used unit can be found if someone searches hard enough).

In the 859xL series are some contenders:
8590L 9 kHz – 1.8 GHz
8594L 9 kHz – 2.9 GHz dc coupled (100 kHz – 2.9 Ghz ac coupled)  - probably going to stress the budget
According to the HP specs both offer RBW of 1 kHz to 3 MHz in a 1,3,10 sequence.

In the 859xE series are some contenders:
8591E 9 kHz - 1.8 GHz
8594E 9 kHz - 2.9 GHz
8595E 9 kHz - 6.5 GHz
According to the HP specs all three offer RBW of 30 Hz to 3 MHz with 10 Hz Nominal.

In the 856X series are contenders:
8560A 50 Hz – 2.9 GHz  - probably going to stress the budget
8561B 50 Hz – 6.5 GHz  - probably going to break the bank
According to the HP specs both offer RBW of 10 Hz to 1 MHz in a 1,3,10 sequence and 2 MHz.

Any thoughts on the tradeoffs among the above models?  (The 8591E, 4E, and 5E look like very good price performers.)

Beyond frequency range and RBW are a bunch of other specs and considerations.  Phase and SFDR measurement performance seem to be worthy considerations.  Any other considerations on buying criteria and potential models are very welcome.

Thanks, EF

PS, I'm kind of in the same boat with what nixfu says in his post above - I'm betting there are others in this boat too so the wisdom from the experienced users is definitely appreciated.

[nctnico]

For audio I'd get an audio analyser or a scope with a good FFT function. An audio analyser is likely to include features like determining harmonic distortion (THD) etc.

[Wuerstchenhund]

I think the prices on the HP/Agilents need to come down.  It wont be long before people realize that there is a better option at that price level than the old HP boat anchors.  If the prices come down to $500-$1000(with a TG included) maybe I might consider a good used one, otherwise I will plan on getting a new rigol.

The Rigol DSA800 is primarily one thing - cheap. If you want a brand new device then the DSA800 is certainly worth considering. But aside from the price there's nothing spectacular on the Rigol. It's specs are somewhat OK, and feature-wise it's pretty simple. As others said any of the old HP boat anchors will (when working) easily out-perform the Rigol DSA.

Heck in a coupe of year, Rigol will probably have new models that blow the existing ones out of the way and the used gear will look even more expensive in comparison. 

I don't think so. Rigol's main feature has always been the price, and they are pretty successful with their kit. It never has been performance, though, aside from the fact that most Rigol kit suffers from some or several firmware problems that seem to be pretty standard with Chinese brands' kit. Just look at their pathetic offerings on mid-range scopes. Or the DSA1000 Series of Spectrum Analyzers.

I can see Rigol pushing down prices somewhat for other entry level gear but not for second hand mid-range and high-end kit.

[Wuerstchenhund]

For audio I'd get an audio analyser or a scope with a good FFT function. An audio analyser is likely to include features like determining harmonic distortion (THD) etc.

Indeed. An audio analyzer can also be found in some Communication Testers. For example, R&S CMU200 I have came with the optional two channel audio analyzer which also includes two separate LF generators (multi-tone capable).

Below are some screenshots showing the main menu, the analyzer with a 10kHz sine on applied (via a crappy BNC lead), the multi-tone analyzer and the THD analyzer.

Much more useful for audio stuff than a SA in my opinion.

[Fraser]

Late to the party but a comment on the Tek 2700 series mentioned by the OP.....

I own a Tektronix 2710 SA and was initially pleased with it. My Advantest SA's at that time were the R4131D units and I used the 2710 for TV signal work. The 2710 then suffered corruption in the character generator so some display text was unreadable.Such a fault is common on these units. This failure was soon followed by PSU failure. After a Re-cap and the failure of several PSU key components, I have to recommend against buying an elderly 2710 SA.

My 2710 is presently sat under my bench in disgrace as I can't get the enthusiasm to dive into the PSU to repair it. It will likely end up sold on ebay as a project for someone else.

Aurora
UK

[G0HZU]

Don't forget to include the old TR4172 in the mix... This was the mega expensive flagship analyser from Advantest/TR that could compete with the best from HP in the 1980s. It covers up to 1800MHz and has a tracking generator. It also has an internal preamp option (not a particularly good one though), a 1000 x 1000 data point display (via CRT) and it has a 'real' frequency counter built in at the IF and this measures with up to 10digit resolution. (eg 1Hz at 1GHz).

It's OCXO frequency reference has less ageing/drift per YEAR (at +/- 0.02ppm) than the basic short term accuracy spec claimed for the Agilent E7495A even with help from GPS.

It's mixer IP3 is typically about +22dBm (compared to about +13dBm for a HP8568B or +9dBm for the HP8566B)

It's phase noise at 100kHz offset is about -128dBc/Hz. This is about 13dB better than a HP8568B.

It also has a factory option to allow impedance measurents if you use the tracking gen and an external return loss bridge from Wiltron. So it can be used as a basic VNA for impedance measurements once calibrated. I don't think any other analyser mentioned so far on the thread can measure impedance with a TGen and external RLBridge. This level of performance still outclasses most modern analysers.

The major downside is that it is a huge and heavy boatanchor and mine lives on a trolley as it is bigger/heavier than the HP8568B. The major plus point is that they are often very cheap to buy if you can find one. I recently paid just £250 for one

[orin]

Depending on how the upcoming new software release holds up, I'd be inclined to consider the Signal Hound SA44B at $919 over the Rigol 815.  The SA44B goes up to 4.4GHz, but does need a PC to run its software.  The associated tracking generator is $599.  You'd want the $40 accessory kit whatever - I'd want the DC block and an attenuator permanently connected.

I actually considered a Signal Hound when I got the 8568B.  I don't regret getting the HP.

[Codemonkey]

For what its worth, I've got a HP8595E which I obtained for a stupidly low price when my employer was selling some surplus kit. I don't know a great deal about them not being an RF boffin but wanted one and this was too good an opportunity to turn down. Speaking to the engineers that used to use it, they all praised it and said it still compared well performance wise to the newer equipment that had replaced it.

I also got a HP ESA-L1500A at the same time, also for a stupidly low price (I got both together for probably less than you're budgeting for one!).

They're both a bit bulky, they're both a bit noisy (fan wise), but they look like they'd survive the apocalypse unlike my Rigol scope which despite being awesome for the price just seems so much more fragile.

[Fraser]

I managed to buy a pair of Advantest R3132 SA's for GBP600 each so there are good buys to be had out there. The R3132 makes my R4131's seem very antique and basic but I still love their ease of use and the capabilities that I have when remote controlling the R4131 via GPIB 

[Wuerstchenhund]

It's OCXO frequency reference has less ageing/drift per YEAR (at +/- 0.02ppm) than the basic short term accuracy spec claimed for the Agilent E7495A even with help from GPS.

The TR4172's frequency stability is pretty good and as you say could easily compete with HP's best SAs back in the days. The E7495 can't reach that but still comes close (<+0.03ppm with GPS locked), but the R&S CMU200 with option B12 (Reference OXCO) does quite a bit better short-term (+0.005ppm).

But I'd say the TR4172 is one of the few old timers that may be worth investing in if you can use it's advantages, even though the problems with almost 30yr old kit are still valid. And because it's not a HP or Tek they are usually pretty cheap.

[G0HZU]

Yes, the OCXO inside the TR4172 is a huge thing and is very expensive.

The ageing spec over 1 day is +/-0.0005ppm according to the OCXO manufacturer's data sheet. i.e. < +/- 5E-10. This is pretty good even by today's standards.

[Wuerstchenhund]

Yes, the OCXO inside the TR4172 is a huge thing and is very expensive.

The ageing spec over 1 day is +/-0.0005ppm according to the OCXO manufacturer's data sheet. i.e. < +/- 5E-10. This is pretty good even by today's standards.

It is. I guess it's part of the reason the TR4172 was excessively expensive back in the days, even compared with its HP counterparts.

[Rupunzell]

The Tektronix 492 was originally designed and intended for a US military contract. Tek eventually got this contract and became sole supplier for this specific type of SA for the US military. Ailtech-Eaton (Cutler-Hammer) filed a protest to zero avail.

Standard US military issue for SA back in the 1970's was hewlett packard 141T system (any wonder why there were SO many of these around) and the Tektronix 491.

http://www.gao.gov/assets/430/421442.pdf



Bernice

[Electro Fan]

What is the meaning of the letters (A, B, C, L, others?) in the HP 859X series?

What are the key tradeoffs between the 856X series and the 859X series?

What is preferred for a tracking generator?  An integrated TG, or for some reason an external TG (such as the 85640A?); I'm inclined to save space an go with integrated...

How valuable/nice/useful is the split screen feature?

Anyone have a link to something that shows the history/evolution/production dates of the HP spectrum analyzer products?

Thanks!

Update:  answered part of my own question here:
http://www.testunlimited.com/pdf/an/5968-2602E.pdf
https://groups.yahoo.com/neo/groups/hp_agilent_equipment/conversations/topics/27032[/url

[Electro Fan]

Keeping in mind this application is for "learning", here are some more thoughts and questions:

What defines "real time" in a spectrum analyzer? 
http://www.radio-electronics.com/info/t_and_m/spectrum_analyser/analyzer-types.php

According to this article SA's can be classified as:
Swept spectrum analyser
FFT spectrum analyser
Real-time spectrum analyser

Is the Rigol DSA815 a "real time" SA?
Are the Agilent 8560E and 8590E "real time"?

The article states:
Both swept / superheterodyne and FFT analyzer technologies have their own advantages. The more commonly used technology is the swept spectrum analyser as it the type used in a general-purpose test instruments and this technology is able to operate at frequencies up to many GHz. However it is only capable of detecting continuous signals, i.e. CW as time is required to capture a given sweep, and they are not able to capture any phase information.

FFT analyzer analyser technology is able to capture a sample very quickly and then analyse it. As a result an FFT analyzer is able to capture short lived, or one-shot phenomena. They are also able to capture phase information. However the disadvantage of the FFT analyzer is that its frequency range is limited by the sampling rate of the analogue to digital converter, ADC. While ADC technology has improved considerably, this places a major limitation on the bandwidths available using these analyzers.

In view of the fact that both FFT and superheterodyne / swept instrument technologies have their own advantages, many modern analyzers utilise both technologies, the internal software within the unit determining the best combinations for making particular measurements. The superheterodyne circuitry enabling basic measurements and allowing the high frequency capabilities, whereas the FFT capabilities are introduced for narrower band measurements, and those where fast capture is needed.

An analyzer will often determine the best method dependent upon factors including the filter settling time and sweep speed. If the spectrum analyser determines it can show the spectrum faster by sampling the required bandwidth, processing the FFT and then displaying the result, it will opt for an FFT approach, otherwise it will use the more traditional fully superheterodyne / sweep approach. The difference between the two measurement techniques as seen by the user is that using a traditional sweep approach, the result will be seen as sweep progresses, when an FFT measurement is made, the result cannot be displayed until the FFT processing is complete.

Swept or superheterodyne spectrum analysers:   The operation of the swept frequency spectrum analyzer is based on the use of the superheterodyne principle, sweeping the frequency that is analysed across the required band to produce a view of the signals with their relative strengths. This may be considered as the more traditional form of spectrum analyser, and it is the type that is most widely used.
 
Fast Fourier Transform, FFT analysers:   These spectrum analyzers use a form of Fourier transform known as a Fast Fourier Transform, FFT, converting the signals into a digital format for analysis digitally. These analysers are obviously more expensive and often more specialised.

Real-time analysers:   These test instruments are a form of FFT analyser. One of the big issues with the initial FFT analyser types was that they took successive samples, but with time gaps between the samples. This gave rise to some issues with modulated signals or transients as not all the information would be captured. Requiring much larger buffers and more powerful processing, realtime spectrum analyser types are able to offer the top performance in signal analysis. 

How relevant are these distinctions and considerations to the selection of an "entry level / first-time / learning" SA?
(On the one hand it's hard to imagine ever buying a second SA so it would be good to get a unit that will go the distance (5-10 years or longer?) but it's also hard to justify extra $ for something that is a nice-to have rather than a need-to have).

Other considerations:

If you have a SA, a Tracking Generator would be pretty useful, yes?

It seems like the 8560E will outperform the 8590E in various specifications but the 8590E might have more automated capabilities to facilitate testing.  This might not be a perfect analogy but I've found that a Tek 2247A scope is sometimes more enjoyable to operate because of the way it employed "early" microprocessor technology and therefore measurements and readouts are sometimes more accessible and better labeled on screen than it's more highly specified big brother 2465/7B.  I'm wondering if the user interface on the 8590E might make it a better "learning" tool?  On the other hand the RBW and many of the specs on the 8560E are superior.  Either way, one of these (actually a 8560E or a 8594E, either with a built-in TG) would seem to be a contender.

The downside of the 8560/90 is that for a rookie there could be any number of issues I wouldn't find until after purchasing a supposedly good unit.  Might be too much complexity/reliability risk and cost to invest in a "learning" SA.  They aren't "vintage" but they might be just slightly long in the tooth.

Having said that, I am enamored with Bernice's much older 141T (it seems like it just barely escaped the vacuum tube era to make it into the transistor era - just kidding, sort of).  It probably has the simplest user interface and it's the oldest and maybe the least likely to keep working but it seems like they can be found for $500 or less and it has a ton of bandwidth - but no Tracking Generator.

Then there is still the Rigol.  And the E7495B.  And the 8568A/B.

I know, it would be good to start by pinning down some constant:  price, functionality, performance, ease of use, reliability, etc. 


 

[joeqsmith]

Keeping in mind this application is for "learning", here are some more thoughts and questions:

How relevant are these distinctions and considerations to the selection of an "entry level / first-time / learning" SA?
(On the one hand it's hard to imagine ever buying a second SA so it would be good to get a unit that will go the distance (5-10 years or longer?) but it's also hard to justify extra $ for something that is a nice-to have rather than a need-to have).

Other considerations:

If you have a SA, a Tracking Generator would be pretty useful, yes?

The downside of the 8560/90 is that for a rookie there could be any number of issues I wouldn't find until after purchasing a supposedly good unit.  Might be too much complexity/reliability risk and cost to invest in a "learning" SA.  They aren't "vintage" but they might be just slightly long in the tooth.

Having said that, I am enamored with Bernice's much older 141T (it seems like it just barely escaped the vacuum tube era to make it into the transistor era - just kidding, sort of).  It probably has the simplest user interface and it's the oldest and maybe the least likely to keep working but it seems like they can be found for $500 or less and it has a ton of bandwidth - but no Tracking Generator.

I know, it would be good to start by pinning down some constant:  price, functionality, performance, ease of use, reliability, etc. 
 

Good article BTW. 

There were a few different tracking generators for the 141T.   I had one that was good to a little over a GHz. 

A tracking generator would be useful and allows you to make scalar network measurements.   Looking at a filters transfer function for example.   There is no phase information.

The 3589A is a single channel vector network analyzer (phase information is added).  There is an S-parameter set available for it but I don't have one.   I have played with using an external directional coupler with it.  Limited because the software needs to detect the test set before some of the features are enabled.   

If you are buying used, your point about not finding a problem until after the purchase is valid, rookie or seasoned.  Maybe you could get some help checking out a used one if you go that route.     

Personally, if learning is the goal, the suggestion of making your own a while back is a good way to go.  The comment about using a TV is tuner can make it really cheap.   The cost will be next to nothing and you will understand how it all works when you are done.   Maybe then look at buying an OTS system.   

http://www.qsl.net/n9zia/spec/w6hph.pdf