Must-have RF "Plumbing" for the lab

[0culus]

This year I'm planning to embark on some ham related projects that I haven't quite decided yet. But in the meantime I am looking to acquire the tools I need.

For those of you with more experience, what are some "must have" items for making connections that no RF lab ought to be without? There's dizzying arrays of attentuators, pads, splitters, combiners, and more. I have a few attenuators already.  Also, eventually, I want to get into microwave, though I likely will want to stay at or below 18GHz to at least help keep cost under control.

Even further, are there things that are worth building yourself for the experience?

Thanks!

[T3sl4co1l]

Definitely a set of cables, tees, adapters and so forth for whatever your equipment mostly has.  BNC is probably your first choice, followed by SMA.  Difference being, BNC is very common on test equipment and lots of hardware, handles reasonable power, but tops out in the low-GHz; while SMA is good to higher frequencies, maybe more vulnerable (smaller connector, more wear on the thread and nut?), and less power handling.

Adapters and fittings are readily available either way.  Attenuators, splitters, directional couplers...  Incidentally, watch out for 75 vs 50 ohm BNC, always triple check what you're getting.  They're mostly compatible but not always, and obviously your SWR will be worse with dissimilar devices.

When you get to microwaves, probably better to buy the hardware for that on an as-needed basis ($$$!).  Wouldn't want to buy up a dozen microwave-capable cables and connectors, just to step on and kink and break them doing pedestrian tasks!

Regarding self-made things, depends.  I keep a stock of PC mount BNC connectors handy for projects.  Easy enough to solder one onto a proto board (edge launch style) and get something almost as good as a proper metallic cylinder/box module.  Whatever you want, you can make: filters, splitters, couplers, mixers...  If nothing else, do a few to get a feel for how difficult it is to make a well-balanced coupler or mixer or whatever!

Tim

[dmills]

And **REALLY** watch out for 75R vs 50R N types!
They do not mix, and a 50R Male into a 75R female will make for a poor fit for any future 75R males.

You can quickly wind up with a LOT of money in RF connectors, and inter series adaptors, so one trick is to mostly pick one connector (I suggest SMA as being reasonable at low power) and adapt everything to it, slightly more loss but it turns an N^2 problem into a 2N problem.

My top tip, some small tin boxes with a couple of whatever your favoured connectors are and some feedthru caps fitted are surprisingly useful for screening prototypes, well worth keeping a few around.

When buying adaptors and the like, go for the used silver plated variety rather then the new overly shiny nickel ones, used Amphenol, MA-Com, Huber-Schurner, Radial are generally better in all ways then new ali-express.

[T3sl4co1l]

Speaking of nickel plating, it's awful for PC mount parts, but nearly unavoidable.   Takes forever to tin.  Tin-plated connectors are almost nonexistent, at reasonable prices anyway.

That's a definite advantage to SMA, gold-plated connectors are cheap and plentiful, also in edge-launch types.  No solderability problem there!

Anyway, nickel plating is undesirable for low noise or low distortion RF applications, because its surface oxide tends to be nonlinear.  In a mated connector, there's probably direct metal-on-metal contact somewhere, but the nonzero voltage drop across those contacts means there's some voltage drop across the rest of the mating area, including nickel oxide contacts, which draw nonlinear currents.  These currents are small relative to the dominant (metallic contact) current, but that still leaves some.  AFAIK, this is usually in the -80dB (THD/IMD) sort of range, not something you'll notice casually, but annoying in a sensitive measurement.  Silver plating does tarnish, but the tarnish is "wiped" easily (scraped off the connectors are mated), which helps.  Best of course is gold, which doesn't tarnish at all.

Tim

[dmills]

The real way to wind up dumping money into the rf connector and cable money pit of course is to buy a VNA, at which point you can see the mismatch and poor return loss....

They also come with weird precision metrology connectors like APC7 and precision 2.92mm which tend to be both delicate and more money then god (And do NOT get me started on the cal kits).

Seriously you think the 6.5 digit voltmeters of the volt nuts are expensive? I give you the Gore "Phase stable Test port extension cable", costs more then audiophile speaker wire, but is actually (sort of) justified.

Regards, Dan.

[hagster]

Just buy what you need as and when you need it. There is no point in stocking up om connectors and cables that you may never use.

If you do buy connectors and crimp tools its worth looking for WiFi targeted stuff as it is generally much cheaper. Only buy yhe expensive stuff if you find you need it or you are connecting to something expensive that may be damaged by cheap stuff.

 Btw, The very cheapest way to get into RF is with an RTL dongle IMO

[tggzzz]

A good place to get good quality cables (without having to sell parts of your body!) Is at hamfests. Obviously you can't guarantee finding them when you need them, so they have to be bought in advance.

[TheSteve]

I'm a big fan of buying used cables/connectors from ebay. Obviously you can get great stuff and also junk. If you want to build up a stockpile for a great price it takes time and careful searching. The last Gore phaseflex 3.5mm cables I bought were $35 each.
I find I use N to BNC and N to SMA adapters the most.
I would try avoid all of the chinese stuff as much as possible. End of the day though we all make do with what we have and crazy high end is rarely required. If you end up with some very nice test gear though(VNA/SA etc) you do want to exercise great care with the connectors.

[cdev]

For a beginner there is a new cheap way to jump start a journey into RF.

Its amazing how much mileage can be gotten out of an RTLSDR dongle, as far as seat of the pants 'measurement' is concerned. Its not measurement, all numbers coming from an RTLSDR are always relative and impacted by the weird way it behaves in the presence of other signals, however they work extremely well for a device which costs as little as $10.  Also, for a traveler or for entertainment in a pinch, they cant be beat, a tiny gadget that can do more than any other tiny gadget its size, that I know of, except maybe an SBC like a Raspberry Pi. They work great in hotels or any high rise building.

With a little work you can use them to receive signals under 24 MHz too..Its quite straightforward what needs to be done and there are lots of peoples experiences out there to draw upon to learn how to do things. You can also buy a $25 dongle the 'rtlsdr blog v3" dongle that has a bunch of modifications that extend its capabilities already made, a very nice implementation of a lot of suggestions over a period of years. If you are just going to buy one I would splurge and get that one. It can be improved on additionally a great deal from there with filters. A good opportunity to learn by building them.

 If you divide the HF band (the part you use direct sampling to receive) into two or three chunks (or more is even better) you can improve the performance more than if you treat it as one band. You can receive HF surprisingly well with an RTLSDR considering their cost. You can use them to adjust filters, and scan large bands of signals. A noise source is I understand helpful, but I still dont own one, need to do that. Then it should be possible to use it to adjust them visually - very useful and hard to beat for the price.

I am able to receive around 2.8 to 3 MHz of spectrum at once reliably without errors. (thats potentially a lot of signals) More than 2.8 MHz sampling rate errors start getting worse fast and it seems that sensitivity declines.

Compared to a modern communications receiver they are not really comparable but for a great many things they are 'good enough' and often you're getting capabilities that a few years ago would have been unattainable for less than a small fortune.

[chris_leyson]

Must haves are 50 BNC to BNC and SMA to SMA cables, a couple of N to BNC, N to SMA and BNC to SMA adaptors. I prefer ready made cables as they last a lot longer and you don't need expensive crimp tools. For couplers, splitters, combiners and mixers etc you can pick up used or surplus MiniCircuits stuff on ebay or buy them new, they're reasonably priced considering the amount of effort you would have to go to to make something home brewed.
50 terminators, you want a few of those as well and a couple of through-line terminators for scope inputs.

[0culus]

Thanks for all the suggestions everyone! This forum is awesome! 

[LapTop006]

For adapters and decent cables Digikey is also a great source.

"Grab bags" of cables & adaptors can be very hit & miss (particularly the random old collections sold on eBay), I've had some that easily justified their price, and several not so much.

[rastro]

I'm a big fan of buying used cables/connectors from ebay. Obviously you can get great stuff and also junk. If you want to build up a stockpile for a great price it takes time and careful searching. The last Gore phaseflex 3.5mm cables I bought were $35 each.
I find I use N to BNC and N to SMA adapters the most.
I would try avoid all of the chinese stuff as much as possible. End of the day though we all make do with what we have and crazy high end is rarely required. If you end up with some very nice test gear though(VNA/SA etc) you do want to exercise great care with the connectors.

I agree.  Try and stick with quality used connectors and cables. 
- Since you are not above 18GHz you'll probably be using N-type and SMA. 
- N-type is generally more robust and most 18GHz HP/etc signal generators will have N-type output.
- - I would look to get good used N cables first.
- - HP has good fixed attenuators in this connector type; you can also find step actuators from HP/etc
I would look for SMA adaptors for SMB or SMC connectors you may occasionally run across.
- - Be careful of using cheap ebay/etc SMA's with good quality SMA as you might ruin them.
- - I've gauged numerous (ebay)SMA-male connectors and it's a crap shoot on if the pin extends past the required specification.

- You may also want to consider 7mm/APC connectors which are rated to 18GHz.
- - You can find good used ones for under $50.00
- - One side is a unisex connector (can act as either M/F) and the other can be N,SMA,3.5mm, TNC ...) 
- - So you can mix and match between adapter types.

rastro

[chris_leyson]

+1 I would look for SMA adaptors for SMB or SMC connectors you may occasionally run across.

I had forgotten about SMB and SMC as I never use them but SMA adaptors are handy because you never know what you might find in pieces of equipment. If you've got any 75 connectors keep them separate and somehow mark them so it's obvious it's 75 .

N adaptors for sig gens, spectrum analyzers and VNA's, you need those if the equipment gets a lot of use because replacing an N socket especially on a VNA is time consuming and/or expensive. The adaptor becomes the easy to replace sacrificial connector when it's worn out.

[bd139]

Minimalist:

1. Signal generator with levelled output and calibrated frequency output
2. Step attenuator
3. Power meter
4. Patches and adapters for BNC/SO239/SMA.

You can do one hell of a lot of stuff with just that. I’ve built all of these myself now apart from step attenuator which is in progress but they are only good to the top end of HF at generation end and VHF at metering end.

[mc172]

Some sound advice already, but avoid BNC for anything above a few hundred MHz. The 50 Ohm ones are not 50 Ohms and the 75 Ohm ones are not 75 Ohms. Same story with PL259/SO239.

If you are seriously looking at going up to 18 GHz, don't mess around, go N and SMA. These are proper RF connectors designed for the purpose, not some bodged together connector with the upper frequency limit extended for the n-th time in the "standard".

I wouldn't suggest that you need ultra-expensive test leads. I make my own with cheap crimp type (not those crappy compression type) N connectors and cable from eBay and have no problems performance wise. But some cheap crimp tools, cheap crimp dies, and you can be going to 3 GHz without much more effort if you have got the gear.
Then will probably come a calibration kit. Easy to DIY.

One thing about some (well, a lot of) eBay N connectors is that the crimp ferrules are often not made with the correct wall thickness, so they never work properly with the supposedly correct crimp die. You have two options, source the correct ferrules (still cheaper than buying Amphenol) or wrap them in copper tape or something before you crimp them. 

[LaserSteve]

One of these:

http://www.hsmicrowave.com/CMG.html

Steve

[TheUnnamedNewbie]

I personally get everything SMA, and just adapt BNC and N to SMA. SMA is far more compact and I like that. For low-frequency (up to like a GHz or so) I think the cheap ebay stuff is okay. I have a few higher-end ones for anything higher, including some phaseflex.

When you see what good, high performance cable goes for on ebay sometimes (the second hand ones, not new non-reputable ones) it really makes sense to just keep your eyes open for them. I have a few ebay auto-searches running that I get updates from, and they always give some nice stuff.

And I have a hypothesis: People who say 'cheap connectors are just as good as the expensive ones' or 'you can easily make your own cal kit' or finally 'torque wrenches are unnecessary' probably don't own/have access to a VNA with good cal-kit. I'm not saying you will need it for your hobby or HAM needs or anything - esp not if you only go up to a few GHz - but once you compare the performance (or lack thereof) of the cheap stuff, you understand why all the professional labs get the good stuff, and you learn that you can't really measure anything with more than 1-2 dB precision, because your cables will just be all over the place.

[mc172]

Torque wrenches are completely necessary, if you need to measure right down near the noise or with high dynamic range, but let's be realistic. This is quite unlikely for someone to be  if they are new to RF. You don't need to be using the torque wrench every time you connect something if you're measuring insertion loss of a metre long cable or putting a signal generator on a spectrum analyser etc. and if you are doing that, you are wasting your employers and your own time. They make connectors with no hex feature and instead a knurled nut for this reason.

I hear a lot of RF guys talking about always having to follow these mythical special procedures or else nothing works like it's some kind of voodoo ritual that can only be known by some old wise guru. It's not, it's just common sense.

1 dB precision is easily achievable with DIY cables, you calibrate them out for insertion loss. If your return loss isn't good enough to enable that, the cheapness of the connector isn't the factor, it's the design of the connector and the joints. So cut it up and have a look inside, calculate the impedances and see why. Then avoid that particular type. I have no experience at 18 GHz, I run out of equipment at 3 GHz but I have no problems uo to 3 GHz with DIY cables.

The expensive cables are often expensive not because of RF performance but the amount of special sauce put in to make them nice and flexible.

Expensive connectors contain nothing that the cheap ones don't other than correct geometry, and even then I find that the cheap ones are rip offs of expensive ones and thus even they are close enough to perform well. As long as you're buying connectors with PTFE dielectric material you can't really go wrong. More often than not, the size of the solder blob on the pin has a bigger effect than the connector brand. This I have found very commonly with N connectors I have seen that other people have made that I have ended up with.

I've got an HP E5062A, the proper HP N type cal kit and the E-cal set. My DIY cal kit load, made using cheap eBay connectors and random bits of metal I had laying around, achieves about 30-35 dB return loss, when calibrated with either of my proper cal kits. Why do you need better than that for some DIY as your first foray into RF? You might even learn a few things by building your own.

[tggzzz]

Torque wrenches are completely necessary, if you need to measure right down near the noise or with high dynamic range, but let's be realistic. This is quite unlikely for someone to be  if they are new to RF. You don't need to be using the torque wrench every time you connect something if you're measuring insertion loss of a metre long cable or putting a signal generator on a spectrum analyser etc. and if you are doing that, you are wasting your employers and your own time. They make connectors with no hex feature and instead a knurled nut for this reason.

Given the expense, it is worth considering a DIY solution: a spanner with a ring at one end, plus a spring balance and ruler, and a little bit of arithmetic

1 dB precision is easily achievable with DIY cables, you calibrate them out for insertion loss.

Indeed, but repeatability and flatness are more of an issue.

[bd139]

Avoiding having to care about some measurements is the most powerful tool

[TheUnnamedNewbie]

To reply to the OP: Some simple attenuators can be great to reduce the impact of mismatch.
A DC block can also be very useful when you want to be sure you don't put DC up an instrument or other component.

Torque wrenches are completely necessary, if you need to measure right down near the noise or with high dynamic range

Torque wrenches are mainly needed for repeatability, and to tighten far enough so it never loosens on its own without risking damage to the thread. In fact, most of those 'mythical special procedure' things us RF guys do has little to do with loss in a single measurement, and everything to do with repeatability.

1 dB precision is easily achievable with DIY cables, you calibrate them out for insertion loss. If your return loss isn't good enough to enable that, the cheapness of the connector isn't the factor, it's the design of the connector and the joints. So cut it up and have a look inside, calculate the impedances and see why. Then avoid that particular type. I have no experience at 18 GHz, I run out of equipment at 3 GHz but I have no problems uo to 3 GHz with DIY cables.

It is not necessary the 'cheapness' of the connector - it has to do with the tolerances and surface finish - but often, cost is indicative of quality. This is why precision standards have much finer center contacts, or even spring contacts within a solid circular center contact. (see picture) You can only calibrate out the cable, if your cable and connection is repeatable enough. Just do a calibration on a VNA (with even very good cables), and then measure an attenuator, and just move the cables around a bit. Even on the VNA at work, with cables that cost more than my car, the S11 will be dancing all over the place, which can be a big deal if you are trying to do a measurement of input/output impedance so you can design a matching network, or are running load-pulls.

The expensive cables are often expensive not because of RF performance but the amount of special sauce put in to make them nice and flexible.

Yet the most expensive set of cables I own, a set of Gore Phaseflex cables, which would cost you about 7k$ new, has a solid core.

To my understanding (I am by no means an expert on cable manufacturing) the main differentiator between different cable grades is
 A) the concentricity of the center conductor, which has an impact on bending performance (as a less perfectly centered center conductor will have a larger impedance drift when the cable is bent)
 B) the periodicity of the outer foil (if it is foil...). As this gets less periodic with less overlap, it starts to impact phase, even more so under bending. I have found that certain chinese brands of coax, esp the thinner ones, start to show notching well below their rated frequency (this is due to the periodicity of the grooves made by the rollers during manufacturing. It causes a small, but highly repetitive discontinuity, which will show up as a very narrow notch filter at higher frequencies. Can produce very odd effects when measuring a QAM signal that has that notch in the band).

Expensive connectors contain nothing that the cheap ones don't other than correct geometry, and even then I find that the cheap ones are rip offs of expensive ones and thus even they are close enough to perform well. As long as you're buying connectors with PTFE dielectric material you can't really go wrong. More often than not, the size of the solder blob on the pin has a bigger effect than the connector brand. This I have found very commonly with N connectors I have seen that other people have made that I have ended up with.

I think most cheap connectors aren't PTFE filled, but PE or Nylon. Cheaper, easier to work with, but gives you non-ideal impedances (but again, at the length of an SMA connector, it really doesn't matter under a few GHz). Center conductors are often misaligned - not enough to damage your cheaper cable, but I wouldn't risk connecting them to metrology/air-core connectors unless I first verified with a gauge set. I also tend to find that the center conductors on the male SMA connectors are way to long, which risks damaging the female one. Diameter is also too high, which will bend the center pins of the female connector out too far, resulting in poor performance (impedance mismatch) in the best case, and a broken female connector in the worst case.

[mc172]

I don't disagree about using torque wrenches, I use them all the time.
Their main purpose is to guarantee sufficient preload to the mating flange for low series impedance and also as you said to preload the thread so that it doesn't come undone.

This is my point, the OP is starting out, and probably not going to 18 GHz for a while yet. What piece of equipment are they going to have that has cal kit type SMA connectors on it? They're not.

Your point about your Gore cables, are about flexibility. Maintaining a level of performance under flexed conditions.

The comment about PE and nylon I don't agree with, you can have a perfectly matched connector using HDPE or even glass if you want as a dielectric, the geometry has to be correct. All of the £2-ish N connectors I buy from eBay are PTFE.

[TheUnnamedNewbie]

Your point about your Gore cables, are about flexibility. Maintaining a level of performance under flexed conditions.

Ah, a misunderstanding on my part then. My bad!

The comment about PE and nylon I don't agree with, you can have a perfectly matched connector using HDPE or even glass if you want as a dielectric, the geometry has to be correct.

Yes, I should have been more clear here. I meant to say 'given the same geometry'. The thing with most of the cheap connectors I run into (fro, non-reputable sources such as random aliexpress or ebay sellers) is that they are often just clones of some reputable connector (often a higher performance one). Those original connectors often came with PTFE, but then had PE or Nylon placed instead, since it is easier to work with. PE in itself will have only lmismatch (it's dielectric properties are not that different, PTFE has an \$\epsilon_r\$ of about 2.05, HDPE is about \$\epsilon_r\$ of 2.35, if memory serves me well. The problem with PE is that it melts at low temperature (sub-150 C). Nylon has far more loss and mismatch.

I have run into more than a few non-PTFE cores.

[0culus]

After reading all the comments, I'm liking the idea of standardizing around SMA and using the odd adapter to make it work with the (cheap) stuff I already have. I like the small size, rather than type N, which makes for heavy connections which like falling off the edge of the table.