Anritsu MS8604A as generic SA & repair

[Ice-Tea]

Just wanted to throw this out there. Amongst the stuff I have inbound to check and sell, I have a few Anritsu MS8604A's. As a "Digital Mobile Radio Transmitter Tester" it's probably close to useless (maybe a few things like occupied BW still have some use) but what caught my attention was the fact it has a "generic" SA inside. A 8.5GHz one. So, browsing through the spec sheet (there isn't all that much info out there, this is what I got:

Good

- 100Hz to 8.5GHz. 8.5GHz is nice, obviously, but 100Hz makes it usable for audio as well. Which a lot of machines don't support
- VBW down to 1Hz
- 40dBm input

Not great

- ANL < 112dBm doesn't seem stellar?

Wut?

(frequency) accuracy +/-2.5% I don't get this? It has a pretty decent 10MHz reference, so 2.5%? Of what?
(RBW) accuracy +/- 20% what now?

So, your thoughts? Yay or nay as generic SA?

[Mr Nutts]

Have you seen this?

https://www.eevblog.com/forum/testgear/anritsu-ms8609a-13-2ghz-spectrum-analyzer-short-review-(work-in-progress)/msg1014488/#msg1014488

Anritsu is a Japanese manufacturer of T&M gear. It was founded in 1895 as Sekisan-sha so it has been around for a very long time. Anritsu pretty much focusses on gear for RF and optical communications, and its main target is the telecoms industry (cell phone service providers).

Anritsu uses a common nomenclature for its test gear, which consists of a prefix (Mx) followed by a four or five digit product number. The 'M' in the prefix stands for Measurement Division, the second letter determines the product category (i.e. 'T'='Test Set', 'S'='Spectrum Analyzers', 'G'='Generators', 'L'='Level meters', 'P'='Photonics').

For someone looking to look at RF spectra, there are two options, MT and MS.

The MT Series contains integrated cell phone test sets of which many of the older ones (i.e. MT8801A, MT8820A) are used by hobbyists as inexpensive Spectrum Analyzer and RF generator. As with many cell phone test sets, the RF performance is pretty limited, which is why the Spectrum Analyzer in these devices is called a 'Spectrum Monitor'.

Anritsu's 'real' lab-grade Spectrum Analyzers carry the 'MS' prefix. 'MS' is also used for derivative products that are based on the standalone SA models, like the MS860xA Transmitter Test Sets which are covered in this short review.

While Anritsu offers test sets under both the 'MT' and 'MS' prefix, as the 'MS' is based on 'true' lab-grade SAs its performance is notably better than the 'MT' Series test sets, by a pretty large margin.

[Ice-Tea]

I haven't! Thanks for that. Searched for 8604 but not 8609. Doesn't seem to be the same family, though. The 8604 is a fair bit older and seems to be monochrome.

But if his review is anything to go by, seems like a full fledged SA inside. So yay! 

[Mr Nutts]

I can ask the author next time I'll contact him with questions but I think his point was that all "MS" are real spectrum analyzers.

Your instrument is probably just a predecessor of the one in the review 

[Ice-Tea]

 I miss him too  say hi from me.

Anyone else with some experience or input?

I'll add more info when I get to checking her out..

[nctnico]

I think this is a mostly an analog swept YIG oscillator machine (hence the 2.5% frequency accuracy and minimum RBW of 10Hz). Not useless but definitely less versatile compared to a modern SA with an minimum RBW of 1Hz and USB storage.

[Ice-Tea]

So... that's not something I misunderstood? You're looking for a 100MHz signal and it might as well say it's 102.5MHz?

That's pretty nasty.. Still don't get it, though.. All the 2G cell stuff is in essence a bit of SW sauce over an SA so I don't get how you'd get to pass/fail criteria on anything with that kind of uncertainty..

[nctnico]

So... that's not something I misunderstood? You're looking for a 100MHz signal and it might as well say it's 102.5MHz?

That's pretty nasty.. Still don't get it, though.. All the 2G cell stuff is in essence a bit of SW sauce over an SA so I don't get how you'd get to pass/fail criteria on anything with that kind of uncertainty..

AFAIK it is 2.5% of the span. So if you are looking at a 100MHz spike with a span of 10MHz then the error is 250kHz.

[Ice-Tea]

Ah, that makes sense. And makes it quite OK. If you reduce the span (which you would if you don't want to wait forever to refresh anyways) the error decreases.

[Ice-Tea]

Last time I was this happy to unpack a unit I just turned it on.

Got smarter though 

[Ice-Tea]

That.. was easy!

[Ice-Tea]

Or not. It turned on, worked perfectly, went to the loo, came back and the screen was gone.

Frack.

[Ice-Tea]

Several hours of dissasembly, cleaning connectors, inventing new curse words I come across this fellow:



Provides the HV for the EL display. Looks a bit crusty on the bottom left side so I zoom in on it and... hello!



Yep. Shoddy solder and a crack. Display is up and running again, even though there's now a bad line in it. May have something to do with the way the flex from the row and collumn drivers are bended?? That's not pretty. But that's a job for tomorrow!

[Ice-Tea]

So, just to make sure, I took the 'good' flux (toxic looking, feeling and smelling) to town on that slab of bad looking solderjoints.

No dice.

Heated up the hot air gun, got some mild flux on column drivers corresponding to the bad line.

No dice.

So, I drag soldered over the suspected culprits... (and cleaned it up off coarse)

No dice.

At this point, I was kinda done. So I started to put humpty dumpty back together again. As I have bad experiences with putting a unit completely together, I do interim checks to see if everything still works. As soon as I put back the EMC enclosure and the earth connections to the two EL driver boards...

it worked. I'm sure I'll think this is funny somewhere in the future, but not right now.

[tautech]

Is C27 leaking ? The surrounding conform coating looks dodgy like you sometimes see around leaking caps.
Hook it out and test it.

[nctnico]

Is C27 leaking ? The surrounding conform coating looks dodgy like you sometimes see around leaking caps.
Hook it out and test it.

Well spotted. Leaking electrolyte is probably also the reason why the solder joints look so bad. If C27 is leaking I'd replace all the capacitors on the board. Preferably with good quality low ESR capacitors.

[tautech]

Is C27 leaking ? The surrounding conform coating looks dodgy like you sometimes see around leaking caps.
Hook it out and test it.

Well spotted. Leaking electrolyte is probably also the reason why the solder joints look so bad. If C27 is leaking I'd replace all the capacitors on the board. Preferably with good quality low ESR capacitors.

The ‘flag’ is dull and pitted SMD solder fillets.....whenever spotted and localised.....look harder !

[Ice-Tea]

Oh! TBH I never suspected a leaking cap. But it makes a lot of sense as I already wondered how they could have passed inspection looking like that (even when taking aging into account).

Put her back together & on eBay but I'll replace c27 at least.

Thanks guys!

[nonlinear]

These units (as well as their brothers MS2701A) suffer from capacitors leaks.

Places are LO1 (YIG drive PCB) , LO2 (DRO PLL PCB) and sometimes (worst case) - synthesizer PCBs.