Setting Up an RF Engineering Lab/Workshop

[German_EE]

We have seen various test and measurement labs on the Internet from Dave's collection to the ultimate RF lab at 'The Signal Path'. So, if you were setting up a lab/workshop dedicated to RF engineering what test equipment would you have on the bench?

My wish list is as follows:

Four Channel Oscilloscope
Spectrum Analyzer
VNA                                          An MDO Scope could perhaps replace these three?
Bench Multimeter
RF Signal Generator
AF Signal Generator / AWG
Deviation Meter
Audio Millivoltmeter
RF Noise Source
RF Power Meter
10-Digit Frequency Counter
0-120dB Attenuator
RLC Meter
10 MHz GPS Locked Reference
8-ohm 10W Dummy Load
50 ohm 200W Dummy Load
Stereo SMT Microscope
I2C / SPI Bus Exerciser (Reading it back can be done on the scope)

[KJDS]

For accurately measuring filters, small signal amplifiers in their linear range, switches and the like, then I'll use a VNA. It's the most used tool when sorting out RF circuitry, with the exception of non linear and high power stuff. If you're doing any serious design work then nothing else will do.

To measure third order intercept point of an amplifier then it's two signal generators and a spec ana. Most modern sig gens have decent modulation capability built in.

It's rare that I use an oscilloscope unless I have unwanted oscillations or want to measure something that isn't RF.

For high power amplifiers then a load pull set is very useful, but unless there's a lot of HPA work going on then you can get by without one. The overhead in setting one up means it gives no benefit.

It's handy to have a power meter and a bench DMM about, but it's not often they are needed.

If you're doing anything where noise figure is to be measured you'll need a noise source.

It's also useful to have a variety of couplers, splitters, attenuators and more adaptors and cables than you think you'll need. A decent soldering station and if you're doing anything small then a microscope. Books of Rs and Cs, boxes of Coilcraft inductors.

If you're doing PLL or synthesizer design, then a frequency counter and possibly a phase noise test set, or at least a spec ana with a phase noise capability is required.

It's very easy to spend a lot of money on decent RF kit.

The one essential you missed off your list was power supplies.

[G0HZU]

The most important thing for RF design work (for me) is having access to decent RF CAD tools. Especially for work well into the GHz region.

Next is a decent working environment (always maintain reasonable free bench space).

What test equipment is required depends on the nature of the RF engineering of course but for mainstream stuff a decent VNA and cal kit is a nice thing to have and so is a high end old school spectrum analyser and a few good quality signal generators and a decent oldschool 100MHz analogue scope.

After that the most important things are good quality RF cables, attenuators, adaptors and to build up a suite of test modules. eg amplifiers, filters, couplers to help verify stuff designed at system level on the CAD tools.

A decent RF power meter would be next on the list and also a reasonable true rms AF->RF voltmeter. Then a noise source.

I really don't think I'd need a MDO scope or an RLC meter or the GPS reference or a 10 digit counter for general RF design work so these would be surplus/luxury items IMO

[nctnico]

I second the suggestion for having all kinds of adapters between N, BNC and SMA Furthermore:
- directional coupler
- splitter
- 50 Ohm terminators
- Good RG223 SMA cables (count on spending around $30 for a good cable of 1m on Ebay)
- Cheap SMA cables you don't mind cutting up
- Several attenuators

The most basic equipment are a spectrum analyser and an RF generator. I wouldn't go for USB devices but the get real deal not older than from the 90's.

Regarding CAD tools: Sonnet Lite (free version) has been good to me although the free version has it's limits. It does take a while to learn how to use it but it is cool to simulate a board and verify it in reality.

[G0HZU]

The other issue is calibration. I don't bother with sending stuff for formal calibration for various reasons but I do cross check all of my equipment using precision attenuators, various complex loads, the VNA, a noise source and the various 0dBm, -10dBm and -20dBm signal references I have here. They can cross check each other quite well.

Between all this is it possible to keep the gear in acceptable calibration and I value this capability over the value of having a crude annual cal/healthcheck at a UK calibration house.

So it's worth trying to buy some precision (step) attenuators etc to give you the the ability to cross check your test gear and also check/record the results over time. Some of my precision microwave 1dB step attenuators have very low VSWR and the attenuation steps are within 0.05dB for most settings across a wide frequency range and a wide attenuation range and this is good enough for me .

Here's a recent crosscheck of one of my old HP/Agilent sig gens and a couple of RF power meters. I control/log the test gear with a PC to produce an automated sweep to check power vs frequency at various power levels.

You can see that the old HP431/478A power meter head has some mismatch uncertainty issues up above 2GHz compared to the modern Anritsu power meter but cross checks like this can be done in a few minutes over GPIB with a simple PC program. I also check the return loss of the power meter heads with a VNA before this sweep.

[G0HZU]

Here's a low frequency check of one of my attenuators using two DMMs and a similar check using the anritsu power meter at 50MHz. This kind of crosschecks the accuracy of the attenuator and also the linearity of the power meter over its optimum range.

I also check the attenuator at DC and the results comeouot the same as at 10kHz but I haven't listed these results.

The results cross check very well and are much better than I need and I get more confidence from tests like this than I do from a cal sticker on the front panel



Level Fluke 45 Keithley 2015THD at 10kHz
0 0.00 0.00
-10 -10.01 -10.007 dB
-20 -20.07 -20.073 dB
-30 -30.00 -30.010 dB
-40 -39.93 -39.958 dB

-1 -0.99 -0.989 dB
-2 -1,98 -1.98 dB
-3 -2.99 -2.994 dB
-4 -3.98 -3.989 dB
-5 -5.00 -5.002 dB
-6 -6.02 -6.03 dB
-7 -6.99 -6.994 dB
-8 -7.95 -7.951 dB
-9 -9.00 -9.000 dB
-10 -10.02 -10.022 dB


Here's the attenuator results at 50MHz on the Anritsu power meter. Again, I always get the same results within about 0.01dB here.

Level
0 dBm 0.00
-10 -10.00 dB
-20 -20.04 dB
-1 -1.00 dB
-2 -1.99 dB
-3 -3.00 dB
-4 -4.00 dB
-5 -5.01 dB
-6 -6.03 dB
-7 -7.00 dB
-8 -7.96 dB
-9 -9.00 dB
-10 -10.03 dB

You can see that the precision step attenuator is very accurate and a very useful confidence building tool! It serves to help cross check several items of test gear!

[Electro Fan]

Does anyone have any examples (links) showing a "I2C / SPI Bus Exerciser (Reading it back can be done on the scope)"? - the last item in the OP's list above

[G0HZU]

Here's another (much wider) range 1dB step attenuator I use to check the health of my spectrum analyser log amp stages over a wide range

The images show the results for tests at DC, LF, 50MHz and 1GHz and they help to cross check a DMM, a power meter and a HP VNA.

I can then go on to use this data with some confidence to check the log amp accuracy of my old school spectrum analysers.

So this really shows how useful this type of attenuator is on any RF workbench

[Howardlong]

Pretty sure you might want an analogue communications test set.

I have both a Marconi 2955A and an Anritsu MT8802A, the Marconi is for backup but in a lot of ways I prefer it to the Anritsu which is a more modern machine: the Marconi boots up a lot faster and is less noisy. However the Anritsu has the spectrum Analyser option and operates to 3GHz (the Marconi only up to about 1GHz). The Marconi's sig gen is pretty crusty, very jittery, but good enough for AM and FM signals.

Comms test sets like the Anritsu mentioned and the Marconi often come with all sorts of cellular test suites, and the good news is that you can get a really decent bit of kit at a reasonable price because the cellular features are no longer of any use. However the analogue testing parts are just as useful as ever.

Be careful to make sure that whatever comms test set you get that it has those all-important analogue tests, some are limited to certain cellular functions only.

[Howardlong]

...and possibly add a Rubidium source, particularly if you're going to be looking at much over 100MHz. I prefer a Rubidium source to GPS as you don't have the hassle of finding an antenna spot (I am in the middle of a city without any roof access). Not sure how GPS technology is these days, but you used to have to wait for some time to get a decent lock on a GPS and then leave it on 24/7.