Opinions on new gear for embedded and RF work

[rhb]

I am looking to upgrade my test bench.  Items of interest are:

Scope with 4 channels at 200-300 Mhz with LA and SPI, etc triggering

Spectrum analyzer to 3 GHz

SCPI controllable  power supply

My current short list is:

GW Instek MSO-3204EA

GW Instek GSP-9300B

GW Instek GPE-4323

Current pricing on this totals about $6k.  I haven't read all the manuals yet, but the ones I've read look pretty decent.   If I can find Tek-Keysight-R&S for the scope and SA at good prices I'm prepared to spend more.  I've seen mention of people getting very attractive pricing on top line stuff.  But so far, every time I price the FW options the price goes through the ceiling and roof.

Used I've got, but it needs attention and doesn't quite do what I want anyway. It's not as old as I am, but a bit closer than I'd like.  I don't want to buy repair jobs.

 I've had a DS1102E for several years.  Buying it was better than fixing my Tek 465 (it's been in storage for over 10 years), but the Rigol has crashed a number of times since I upgraded from the original FW in hopes of getting faster USB transfers of data.  Unfortunately, that did not happen.  It still writes to a USB disk at 167 KB/s, but now it crashes.

GW Instek looks to be a reasonable compromise between price and quality.  Having written tens of thousands of lines of software that was deployed for close to 20 years without any bugs being reported by users, I have a real issue with the software engineering of the T&M vendors. Even the top tier companies give me a queasy feeling.  I'm sure that Lockheed Martin will get a prompt response, but I'm just an old guy in Arkansas.  What Arlo Guthrie called "The Last Guy".

[Berni]

This covers quite a broad range of equipment.

For the scope the 300MHz requirement rules out a lot of the cheap scopes. I myself am a bit of a Keysight fan when it comes to scopes. My day to day scope is a MSO6034A and it was a rock solid reliable scope so far. Occasionally you can find a good bargain on one of these on ebay, if you are lucky you can even find a 1GHz model of it. Feature wise its very close to the newer MSOX3000 but with a higher resolution display, more memory, all the ports right on the back from LAN,USB,VGA,GPIB etc. Also it boots in under 10 seconds compared to the annoying long boot times of the X series scopes.(Tho is does have less wfm/s and is buklier)

As far as spectrum analyzers go there is a lot more to them than just bandwith. Things can get expensive quick if you want a very fast real time spectrum analyzer or have super low phase noise. But seeing that you are in the market for both a SA and scope maybe the Tektronix units that combine both in to one instrument is a good match for you. They are not cheap, but comparable to the price of a SA and scope added together.

[rhb]

300 MHz is "would be nice".

I thought that the Tek MDO3000 would be a good fit until I saw that the SA is an FFT fed by an 8 bit ADC rather than an actual SA. In addition there are a lot of negative comments about the UI design and responsiveness.

I'm not absolutely opposed to used, but age is a major factor and unlike 5 year old computers, top line test gear holds its value.

I want to get setup so that I can efficiently do low volume custom embedded systems as well as cover at least the 2.4 GHz ISM band. I don't need anything exotic.   5 GHz ISM would be great, but I don't see a justification for the 2x uptick in cost.

I bought a Siglent SSA3021X because of the price, but returned it.  I don't know if the issues were all FW or part FW and part HW, but I concluded I would rather spend more money for something with a useable manual set.  Overall I thought the hardware was quite nice, so I think it a real shame that management doesn't think that documentation and SW QA matter.

[Berni]

Well the Tek MDO scopes can still be quite a impressive spectrum analyzer since it makes it a real time spectrum analyzer with a lot of analysis bandwidth. This is useful for looking at complex modulation patterns or looking at RF in the time domain. But yeah the firmware on the these Teks is not the best.

On the other hand if you want to just use a spectrum analyzer to measure the output power of a RF transmitter or check that your circuit is not creating any unwanted EMI radiation. Then you are better off with a classical "swept receiver" style spectrum analyzer. It won't have the time resolution or be able to make sense of anything past simple AM or FM modulation, but these spectrum analyzers are cheaper and even a old >20 year old design can perform very well.

Buying slightly older used equipment can get you a much better deal. For example for within your budget i bought a Agilent DSO9024H scope (3.5GHz 20GS/s 1Gpts) and a very old HP 8569B (22GHz Spectrum analyzer). But these two are not the best instruments for day to day use. Both of them are huge and heavy, both consume 100s of W of power, run hot and make hell of a noise with the fans that work hard to keep it cool. Performance comes at a cost other than in $$$ too.

The scope i actually want to have is the old Agilent MSO7000 series. It has the same rock solid firmware that was developed over 20 years of hard work, but also has the nice big screen like you find on the more modern Keysight scopes. Tho i was never able to spot one for a good bargain so i stick to my trusty old MSO6000, it never let me down.

[cdev]

Caveat here that I only owned a spectrum analyzer once and it was an extremely old one.

If you want to be able to figure out what's going on with complex modulation schemes, a Gnuradio-capable SDR (or several) might be a good addition, and might even be able to do double duty as a sort of poor mans SA, unless you need a verifiable, calibrated device.

But many spectrum analyzer tasks aren't possible in realtime. You cant look at the entire frequency range it covers at once, without scripting.

So getting the big picture requires scripting it with external tools. Which is now quite easy and reveals a lot of info if you are willing to follow a workflow for extracting what you need from it, with a minimum of external components. (attenuators, cables, probes to track signals down) which you likely already have.

What I am saying is, its a good addition that doesn't cost much compared to some pro equipment and might be much more helpful in some situations. But others, less so.

For example, extremely broadband RFI might likely be much easier to find quickly with a portable spectrum analyzer.

You can see it on an SDR but figuring out its fundamental frequency and power vs. time characteristics requires running a sweep and recording the findings to a CSV file, and then visualizing it in a graphing program.

Summary: SDRs combined with data vis tools can give you a lot of info, and they are cheap, but its not 'plug and play'.

[rhb]

I have an Airplay RSP2 and a LimeSDR for which I'm designing an enclosure.    I need to design some notch filters to cut out a local AM and FM station so I can use the SDRs in those bands,

Ultimately I want to connectorize and package a bunch of RF blocks so I can experiment with RF designs without having to build an entire radio each time.  I'd like a VNA with an S parameter test set, but finding a recent one that I can afford looks challenging.

[cdev]

I've thought about this a lot too.

Solderless breadboards are great for making it easier, as far as their RF blocks non-rf control signals.

The best I have been able to mnage so far is standardizing on SMA and BNC connectors, etc.

When I was a kid I had an electronics building block kit that used a thin steel sheet and magnets.. Something like that might make for a good platform, as its ground plane.. magnets under the modular devices. But the magnets I have don't conduct, and in fact, they seem likely to be fairly ferrite-like. so need a conductive (capacitive) tab under them.. coated with a thin layer of tape or vinyl, like mini mag mount antennas have.

Ultimately I want to connectorize and package a bunch of RF blocks so I can experiment with RF designs without having to build an entire radio each time.  I'd like a VNA with an S parameter test set, but finding a recent one that I can afford looks challenging.

[rhb]

Solderless breadboards are horrible for RF.  The parasitics just cause no end of trouble. 

Dead bug is the way to go for RF breadboarding.  But in any case it all has to be shielded.

[cdev]

Thats what I was saying..

Solderless breadboards have made non-rf prototyping much easier. They make non-rf signal prototyping much less messy, and generally easier. Rf blocks could benefit from something that made connecting modular pieces easier.

Solderless breadboards are horrible for RF.  The parasitics just cause no end of trouble. 

Dead bug is the way to go for RF breadboarding.  But in any case it all has to be shielded.

[Berni]

Real time spectrum analyzers are not limited to looking at narrow spans. Its just that things like SDRs are designed to be radios rather than test equipment so they are not built to handle wide spans.

There are benifits to real time SAs for broadband even. Using wide spans and narrow RBW on a classical SA will slow it down to a snails pace or having to wait seconds or even minutes for a full sweep (Filter settling and such). In contrast a real time SA will march trough the wide span in chunks of 10s of MHz. Proper ones can tune to a new frequency band incredibly quickly so very little time is wasted between steps. For narrow RBW and wide spans they will still slow down to being far from real time but they will complete the sweep much quicker. Tho these sort of SAs tend to carry a hefty price tag, Tek seams to make the cheapest ones from all the big brands.

I personally don't work much with RF, just dabble with it on occasion. But dead bug on some grounded copper clad is pretty good for high frequency things. But for things above 1Ghz its probably better to stick to RF metal brick modules connected by SMAs.

Oh and FFT on is not always terrible, just that in most scopes it is. If the scope has a good AFE+ADC and it can do FFTs with over 1M points along with being able to automatically set capture parameters according to a designed span/rbw then it starts to feel like a pretty capable spectrum analyzer. Tho they will all still have lots of spurious tones in there so for looking at tiny signals you have to make sure what you are looking at is actually the signal and not a harmonic of some clock inside the scope.

[tggzzz]

Solderless breadboards are horrible for RF.  The parasitics just cause no end of trouble. 

Standard early 80s logic (e.g. LSTTL) required VHF techniques.

Modern jellybean logic requires UHF techniques (e.g. LVC has <1ns transition times)

Interesting modern logic requires microwave techniques (down to <50ps transition times).

Dead bug is the way to go for RF breadboarding.  But in any case it all has to be shielded.

Ditto for logic circuits.