Best USB stick/device as Spectrum Analyzer for <= US$ 500

[Gribo]

Ok, here is my actual measurement data:
30MHz to 1GHz, at 100KHz RBW takes 4 seconds, for 33950 data points on my machine (Windows 8, Core i5 4460). I use this unit mainly in ATE, so I don't really care for this.
There is no logarithmic frequency view, however, in LV it is quite easy to set your plot X axis to logarithmic.
Regarding EMC precomliance - It is as good as any low range spectrum analyzer. It is not an EMC analyzer, with all the limits, masks and pre amplifiers, but that is not the scope of this device.
Also, take into account that the SA44B gets up to 4.4GHz vs 3GHz for most low end analyzers.

[davorin]

Hi Gribo

Ah great...thanks for the feedback....

Have you used the new Spike software? Can you maybe post the 30MHz - 1GHz plot with no input signal here?

Well...I'm already saving money for a SA44B now...the TG44 comes later.....might come handy also doing repairs of old radios :-)
Sad the TG44A can't be used as a signal generator with AM/FM modulation....


thanks in advance
richard

[Gribo]

Here you go. I don't know what that dip around 150MHz is, it exists also in a lower span. This is without anything connected to the input.

[Fraser]

Something worth checking on the SignalHound and being aware of, IIRC it has a DC coupled input  from memory the signal passed straight into the first active stage without decent protection. this avoids issues with the input protection degrading the frequency response but it carries risk with it.

Forgive me if this is duff information, I looked at the design a long time ago and I may have it confused with another USB spectrum analyser. I did photograph the whole input stage though and will have to see if I can find the images. Worth checking though as I know my friend fried his SignalHound input on more than one occasion. If it is DC coupled, I strongly recommend the use of a DC block on the input port.

Aurora

[Mr Simpleton]

The dip you see, is due to internal band switching, signals do have an alternative path inside, below 150 MHz.
Now the SA44B do have a 250 kHz wide IF where it does its sampling, i.e. one have a real time spectrum which is up to 250 kHz wide. If one want to sweep wider, the  SA44B is stepped, which accounts for the slightly long sweep times vs "normal" spectrum analyzers. The new software "Spike" has improved the UI and sweeptimes so its almost like a new instrument. Having used simple TV dongles and the similar Fun Cube and they are not in the same class as the Signal Hound.

As far as I know the Signal Hound offers a 30 day money back... and they do have an european dealer so this might be a way to try it out!

I find that the Signal Hound compliments rather than tries to replace my big HP analyzer, at sweeps less than 250 kHz it is a lot faster than any of my HP analyzers... especially when having very narrow RBW.

And the SA44B is calibrated and offers traceability to NIST....

[Mr Simpleton]

1 Hz to 4.4GHz, max +20dBm and 0 VDC.... Have been lucky and my input stages are still unharmed!

Agree a DC block do come in handy... and watch the RF input level too

[Fraser]

Yep, just found a pic of warning on unit...... Unusual and some users are used to <=25V DC offset capability on conventional analysers so kill their new toy with a DC offset  A case of RFM 

Aurora

[Fraser]

Also BB60 front end block diagram may be of interest. I will see if I can find the exact one for the '44'

[Mr Simpleton]

This is from the manual, a very simplistic block diagram. I remember someone in Russia tearing the analyzer apart... see if I can find the thread. It was all in russian though

[Fraser]

I took one apart as well. From memory I was impressed with the unit (apart from the lack of front end resilience) When it came down to it, it was a SDR with wide window sweep and it performed as such. i.e very much dependant upon the capabilities of the software. This was seen by me as a positive as the software provided was Ok and there was room for greater capabilities in the future without a hardware change. I was not a great fan of SDR technology as it often had a limited data window width and less than great dynamic range. Tektronix were proud of a 64Khz window when I checked out one of their new oscilloscopes. SDR is likely the future though and the Signal Hound is definitely a respectable attempt at a USB based SDR spectrum analyser. Capabilities are getting better all the time as hardware and software improve.

Will a USB based spectrum analyser ever replace the bench top conventional designs ? No idea but it may well be the future of such test equipment. Imagine having a single smallish box that contains all your test equipment and that has great capabilities that may be upgraded through just a software update. Multimeter + LCR meter + ESR meter + DSO + LA + SA + ? ..... sounds nice if it can be done at an affordable price 

Aurora

[Mr Simpleton]

Unable to find any pics from a dismembered SA44B, but the SA12B...
http://forum.vhfdx.ru/izmeritelnaya-apparatura/analizator-spektra-signal-hound-usb-sa44b/?action=dlattach;attach=45656;image
See if I can find the guts to pluck mine open

[davorin]

Thanks for the print screen :-)

Yes I saw somewhere in the datasheet that it has two frequency ranges with a switch at 150MHz...
Also remembering that the input is DC coupled and has a limit of +/- 0.2V....but the European distributor sells a kit for Euro 40 also including a DC blocker...

But also according to the datasheet...theoretically I could also do a spectrum plot in the audio range....

Have to put now some more stuff on ebay to get my Signalhound SA44b sooner ;-)

[davorin]

Okay...regarding the few replies that the Signalhound SA44B can't replace a real HP spectrum analyzer....

Which HP analyzer would you recommend to have a look out on Ebay then? And which wouldn't kill my budget/wallet? (o;

[Mr Simpleton]

B4 we can give a definite answer... You need to clarify your needs!
What freqency range?
Portable or bench?
Size matters?
Saving data on computer?
Warranty vs fixing your self when breaks down?

There have been numerous discussions on this very topic lately on the forum... just search for spectrum analyzer.

[Wuerstchenhund]

Which HP analyzer would you recommend to have a look out on eBay then? And which wouldn't kill my budget/wallet? (o;

Doesn't have to be HP. Another option might be a Rohde & Schwarz CRTU-RU (a Wireless tester containing a Spectrum Analyzer, RF generator, RF Analyzer with 11(?)digit frequency counter, and Power Meter).

There recently was a huge surplus auction in Finland where lots of CRTU-RUs from some big cell phone manufacturer were sold for peanuts, and they now show up on eBay in Europe (and prices will probably drop further). Some forum members also bought several of these things and might sell you one for a reasonable price.

The SA in the CRTU is pretty good, with a bandwidth of 10MHz to 2.7GHz. Not as good as a high end SA like the HP 8566 Series of course, but the CRTU holds up quite well against the current crop of entry level SAs, and in some cases even against my lab grade R&S FSP7. It also has the advantage that the CRTUs from that auction are only around 7-8 years old and not ~30yrs as most HP SAs, and CRTUs don't suck as much power and produce as much noise as the HP boat anchors.

Depending on what you need, this could be a viable option.

[Mr Simpleton]

Just had a quick look at e-bay, but cannot really say the marked is flooded with cheap CRTU's... prices runs from $2k for non-working units up to $20k
Guess those Nokia units still have to show up and kill the market

[Wuerstchenhund]

Just had a quick look at e-bay, but cannot really say the marked is flooded with cheap CRTU's... prices runs from $2k for non-working units up to $20k
Guess those Nokia units still have to show up and kill the market

I guess you checked ebay US (which in term of CRTUs doesn't have much and what's on offer is mostly way overpriced) but there are some on ebay in Europe:

www.ebay.co.uk/itm/Rohde-Schwarz-CRTU-RU-1138-4000-83-Universal-Protocol-Tester-RefRUS-/171760548301

www.ebay.co.uk/itm/Rohde-Schwarz-CRTU-RU-1138-4000-82-Universal-Protocol-Tester-Ref1693-/171760535938

www.ebay.co.uk/itm/Rohde-Schwarz-CRTU-RU-1138-4000-82-Universal-Protocol-Tester-Ref1695-/171760521674

As far as I know Nokia wasn't the only one off-loading CRTUs, companies like Broadcom have as well as the CRTU doesn't support 4G (LTE) and is now successively replaced with CMWs. Which means prices will fall even more.

[G0HZU]

If you just want to use an analyser for casual EMC testing then the SignalHound is going to be worth considering.

It is a very unconventional analyser (in terms of block diagram and how it tries to get a decent image rejection) so it will be compromised in some areas of its performance and also in some parts of its frequency range. I've not used one myself though... I'm sure if I had one here to review I'd find quite a few glaring limitations in its performance.

But I think as long as it is used with a degree of sympathy wrt these limitations it should be a lot better than the typical USB dongle type analysers and it will be capable of more than respectable performance in some areas even when compared to some classic old lab analysers.

However, I suspect that the 'spectrum analyser' in the R&S CRTU and CMU200 models will be pretty ropey if you want a decent spurious free dynamic range. R&S seem to offer very little info on the hardware used in the receiver section and I suspect this is because it would make it obvious how limited they are in terms of performance. The spec sheets offer a few clues that the RF performance of the receiver/downconverter section will be pretty dire compared to a conventional lab analyser. Despite the fact that they are made by R&S I'm not so sure the analyser in these instruments is quite as good as it may appear to some people.

[Electro Fan]

Okay...regarding the few replies that the Signalhound SA44B can't replace a real HP spectrum analyzer....

Which HP analyzer would you recommend to have a look out on Ebay then? And which wouldn't kill my budget/wallet? (o;

Oklahoma is not exactly right next door to Switzerland (and I don't know anything about this unit or this seller), but a HP 8560E seems like a worthy candidate:
 
http://www.ebay.com/itm/48-HOUR-PRICE-DROP-HP-8560E-SPECTRUM-ANALYZER-OPT-07-MASS-MEMORY-MODULE-/111672894396?pt=LH_DefaultDomain_0&hash=item1a003923bc

[Mr Simpleton]

And the seller seems very keen on selling to within US...  Still price is very very good for a working 8560E!

[davorin]

Well...at least I know now the model names which would fit my purpose...

But for a start I still go with the SA44B...almost got my Paypal account filled up to order (o;

Not only that I work 3 days at home and 2 days in the office...so taking this device with me is a great plus...
But also the fact it is based on SDR and their API allows to read out Q/I directly or do custom sweeps is a benefit....

And sure I will order a DC blocker, attenuator and the like....not sure if their passive probe comes handy....not expensive for 2GHz bandwidth probe...

[Wuerstchenhund]

However, I suspect that the 'spectrum analyser' in the R&S CRTU and CMU200 models will be pretty ropey if you want a decent spurious free dynamic range. R&S seem to offer very little info on the hardware used in the receiver section and I suspect this is because it would make it obvious how limited they are in terms of performance. The spec sheets offer a few clues that the RF performance of the receiver/downconverter section will be pretty dire compared to a conventional lab analyser.

There's quite a lot of information on how these units work available in the Service Manual (I can see if I find a link).

Not sure why you put spectrum analyzer in quotas, as the CRTU doesn't work any different than most newer standalone SAs. The platform the CRTU and CMU200 is built on is the same platform that R&S had built their previous generation of mid-range and high-end standalone SAs. We're also not talking about cheap basic GSM testers here, a CRTU with a few options did cost in excess of $100k.

I'd say that specs like a SSB phase noise of -110dBc to -118dBc and a frequency stability of +5 * 10E-09 aren't too shabby when we are talking about cases where the alternative is to use some Realtek-based TV receiver as software SA. The CRTU also has a pretty flat response over the whole bandwidth (20MHz to 2.7GHz).

Despite the fact that they are made by R&S I'm not so sure the analyser in these instruments is quite as good as it may appear to some people.

Really, get over yourself. It' has already become obvious that anything that performs worse than a HP 856x is too inferior for you to just even look at and therefore must be crap. Don't worry, I get that.

Reality is however that not everyone needs a high end lab SA just for doing RF work, in a similar way as not every EE needs a 20+GHz scope. Again, the OP was considering using USB TV receivers. You really believe they would give a better RF performance than a lab-grade communications test platform?

BTW, my old CMU200 is at the moment sitting in an EMI lab to do simpler pre-compliance testing. Seems to be non-"ropey" enough for the task.

[Mr Simpleton]

If I understand correctly the CMU-family do have a 10 MHz sampling bandwith... how fast does it manage to step through say 1 GHz sweep?
Can you choose RBW ?

And how easy is it to replace/service the Win2000 image? Just a matter of cloning disks? Are there any calibration constants that may be lost and make life miserable?

It sure looks like a sweet box and at prices shown i.e. £500 it looks even better!

[jadew]

Don't bother with USB sticks, they all suck in one way or another. Even the new Tek RSA toy fails miserably with those spurs that get mixed in with the signal.

I have the HackRF and its sensitivity is worse than the RTL dongles and on top of that, it has lot more spurs.

My suggestion is to save up and get a real SA. If you decide to go for a signal analyzer instead, make sure it's one that can do a full sweep and doesn't take a million years to do it. Also make sure that it doesn't suffer from the same issues that the USB dongles have, since signal analyzers are nothing more than SDRs and their primary goals are very different than those of SAs.

[Wuerstchenhund]

If I understand correctly the CMU-family do have a 10 MHz sampling bandwith... how fast does it manage to step through say 1 GHz sweep?
Can you choose RBW ?

Both CMU200 and CRTU do a frequency span from 0 to 2.98Ghz bandwith (so are not limited to just in 10MHz as the Agilent E4406A Comms Tester). The RBW can be set automatically and manually between 10Hz and 1MHz in 1-2-5 steps.

The fastest sweep time for the full bandwidth (20MHz to 2.7GHz) is approx 100ms, but the sweep time of course depends on the RBW.

And how easy is it to replace/service the Win2000 image? Just a matter of cloning disks?

Yes, exactly. Both CMU200 and CRTU use 2.5" IDE hard disks (with depending on age of the device between 40GB and 80GB of capacity). Simple to clone.

Some words about the software:

The CMU200 comes with a MsDOS partion only, as the user interface software is based on MsDOS. The hard drive can be cloned easily to a 4GB CF card with plenty of spare sectors for wear leveling.

The CRTU comes with three partitions, one with Windows 2000, one with MsDOS, and one restore partition with Norton Ghost. The DOS partition is essentially the same as the one on the CMU200, and the DOS software is what one would use for using the CRTU as a normal test instrument, The Windows 2000 partition does not contain software that can be manually used for testing but an application suite for automated cell phone testing, which for general use is worthless. However, it also comes with Windows drivers for all the hardware and Microsoft Visual C++ 2003/2005, which means someone with the necessary time and skills could write his own Windows test applications to make the CRTU in whatever he wants it to be.

Are there any calibration constants that may be lost and make life miserable?

No. Unlike many old kit like the mentioned HP analyzers which store all calibration data in a single (often volatile) memory, or newer kit which stores cal data on the hard drive, R&S uses a different approach where each individual module contains a NVRAM with the module's calibration data. That means should your CMU200/CRTU hard drive dies you can get another one from a different unit and use it without having to worry about calibration. It also means that if you replace a module that you won't have to recalibrate the whole unit. Just plug in the replacement module, initiate a firmware update from the File Manager, and done.

It sure looks like a sweet box and at prices shown i.e. £500 it looks even better!

The CMU200 is still somewhat in demand and therefore still pretty expensive (but then it offers some options that are not available on the CRTU), but the CRTU is certainly a lot of value for the money, and even more so if you want to experiment with your own software.

I haven't had much time to play around with mine but I might do some review in the near future, and if you or anyone else has questions or wants to see some screenshots then let me know.