Low Cost VNA re-visited

[Roger Freeman]

One more vote for Copper Mountain Technologies. I have tested the R54 which only costs USD3000 and works up to 5.4 GHz, and it works nicely. It's only a 1-port though...
**broken link removed**
Aaron

I own a Copper Mountain Technologies (CMT) TR1300/1 (website shows cost = $US2750) and it is superb. You can read all about its performance specs on their website, but see below for a summary. I have tested it against a mainframe HP VNA and it meets the same performance criteria and displays the *same* stopband and passband trace (when I was measuring a 2.4GHz BPF).

One thing it does do (as per mainframes from the usual manufacturers) which some very cheap instruments cannot achieve, is possess a fast sweep time e.g. 120us per point rather than a few ms per point. This is a big deal if you're "live" tuning antennas or cavity filters.

The TR1300/1 is capable of:

Frequency range: 300 kHz to 1.3 GHz
Measured parameters: S11, S21
Sweep types: Linear frequency, log frequency, segment, power sweep
Dynamic range: 130 dB (10 Hz)
Measurement speed: 150 µs per point at 95 dB dynamic range
Output power adjustment range: -55 dBm to +3 dBm
Up to 16,001 measurement points per sweep

It's nonreversing so it won't automatically do S12 or S22 but that doesn't matter in most cases - or you can just turn the DUT round. I have spoken to Alex, their US director, and Ben, their main technical man, and they have both been very helpful with advice on equipment, measurements, feedback on software.

Talking of which - the software is very good. The UI (very suited to touchscreen and certainly works well with a mouse) is extremely usable and is fairly familiar to anyone who knows HP/Agilent/Keysight products. The functionality is all there - you can have several traces or channels, you can do de-embedding, time domain stuff (DTF), fixture simulations, Z/Y/S conversions; there are lots of cal sets available. It doesn't crash. It runs full screen. There *are* one or two "nuances" (not bugs) in the UI - and they correct/change them with quite frequent s/w updates. You can get a demo version of the software so you can try it before you get the hardware. According to **broken link removed** a native linux version is being developed.

The innards are built to be bombproof. The RF PCB design is first rate as is the manufacturing cleanliness. The device performance comes from a LOT of internal screening, with millions of vias, a big FPGA and some fast ADCs. It runs cold to the touch; it has a tough metal case; solid N types on the front.

This is a serious, laboratory/development/factory quality instrument. It is not a toy or amateur device. The support from CMT is very good. I think they will continue to gain a larger and larger market share.

I don't work for Copper Mountain or for Planar (parent designer in Russia) but I do recommend them without hesitation. Buying a Copper Mountain box and even adding the cost of a dedicated laptop (or Surface Pro) is very affordable compared with a traditional mainframe. And you can pick up a Copper Mountain VNA with one hand!

 

[Roger Freeman]

... The innards are built to be bombproof. The RF PCB design is first rate as is the manufacturing cleanliness. The device performance comes from a LOT of internal screening, with millions of vias, a big FPGA and some fast ADCs. It runs cold to the touch; it has a tough metal case; solid N types on the front ...

I've got some photographs of the insides: please also see this PDF which has some shots **broken link removed**

https://obrazki.elektroda.pl/5784763300_1460583643.jpg
https://obrazki.elektroda.pl/6018375600_1460583646.jpg
https://obrazki.elektroda.pl/6076684300_1460583647.jpg
https://obrazki.elektroda.pl/9711372900_1460583649.jpg

(I hope they've uploaded correctly)

 

[vfone]

I am a bit skeptical when is about performances of these kind of USB test equipment.
I used for a while an Aaronia USB Spectrum Analyzer - Spectran 9.4 GHz, and also their portable version with LCD display.
**broken link removed**
If you read what performance they claim in the datasheet you think that Keysight, Anritsu, and R&S, all are going tomorrow out of business.
But frankly speaking, I never seen a piece of crap bigger than this spectrum analyzer.
When the input is terminated with a 50 ohms load, the noise floor vs frequency range varies more than 60dB !!
They claim a minimum sample time of 1ms, but whatever RBW or number of sweeps you chose, the sweep time is huge.
If you use at the input a frequency sweep signal and chose Max Hold, you get repeatable gaps in the measurements. Perhaps they cannot correlate the sweep frequency with the sampling rate.
Always you have to use proper external attenuators because if the input level of your signals is greater than about -20dBm you get a jungle of signals on the display.

 

[hagster]

Thats a real time spectrum analyser. Basically a software radio with something like 40MHz bandwidth. To scan the whole band it needs to devide it into sections. The 14 bit ADC limits the dynamic range so you need to adjust the pre amp gain to see weak signals. These things are a totally different beast to a normal spectrum analyser. But they do offer some advanteges in some instances. If you set the bandwidth

- - - Updated - - -

[oops hit post too early]

...to less than the max stareing BW you can capture transient signals with 100% probability. You can also do real decode of complex modulation schemes.

The CMT usb VNAs on the other hand are real VNAs not software radios and are definately more than a match for high end Agilent and Keysight units.

 

[Roger Freeman]

I am a bit skeptical when is about performances of these kind of USB test equipment.
I used for a while an Aaronia USB Spectrum Analyzer

I too evaluated an Aaronia spectrum analyser (RSA9000). It had more in common with a comb generator than any form of signal measuring instrument. It is without doubt the worst-performance instrument I have ever encountered. The dynamic range s unusable and it never achieves the sweep rate claimed or demonstrated by their demo software. There is a complete "forest" of spurs at regular intervals, Aaronia's "solution" to which was to merely blank the display where the offending products were! I think they were every 100MHz and so, if you wanted to look at something like 890-915MHz, there would be a blank 1MHz region centred on 900 for example. Something like that, I don't remember the details. I am trying to forget.

However - Aaronia USB spectrum analysers are entirely different from CMT USB vector network analysers. There is no similarity other than the word analyser :lol: ... please don't consider Aaronia as being representative of USB instruments in general.

The CMT usb VNAs on the other hand are real VNAs not software radios and are definately more than a match for high end Agilent and Keysight units

hagster - You're right. Using a CMT unit is very similar to using a mainframe in terms of display update rates, sweep time, responsiveness of instrument controls. As I mentioned earlier, sweeping across a filter gave pretty well identical results compared with a mainframe when I did some comparison measurements (I think it was an E5071C ENA from Agilent). The S11 and S21 differences were insignificant, I mean to say. I did a quick look at some antenna ports too, and the "liveliness" of the display meant that tuning could be achieved in real-time. You can of course save S1P or S2P trace data etc. to a hard drive as per a typical mainframe.

The fact that you can pick up a CMT VNA between two fingers and move it around the factory or lab is a relief too ... no trolleys needed.

 

[D.A.(Tony)Stewart]

I am a bit skeptical when is about performances of these kind of USB test equipment.
I used for a while an Aaronia USB Spectrum Analyzer - Spectran 9.4 GHz, and also their portable version with LCD display.
**broken link removed**
If you read what performance they claim in the datasheet you think that Keysight, Anritsu, and R&S, all are going tomorrow out of business.
But frankly speaking, I never seen a piece of crap bigger than this spectrum analyzer.
When the input is terminated with a 50 ohms load, the noise floor vs frequency range varies more than 60dB !!
They claim a minimum sample time of 1ms, but whatever RBW or number of sweeps you chose, the sweep time is huge.
If you use at the input a frequency sweep signal and chose Max Hold, you get repeatable gaps in the measurements. Perhaps they cannot correlate the sweep frequency with the sampling rate.
Always you have to use proper external attenuators because if the input level of your signals is greater than about -20dBm you get a jungle of signals on the display.

I used the same unit last year and had some USB win8.1 driver issues but I found it work as I expected with an excellent noise floor and better with the pre-amp. Selectivity and graphical display on external PC or MAC was good. I could easily look at the frequency hopping of a cell phone and distinguish TV radio cell tower, GPS and phone signals.

THe problem I had was that the channel assignments and demodulation schemes were all standardized for EU spectrum and not North American spectrum, which made setup a PITA,

It does require lots of prior SA experience and knowing how to use an SA otherwise, unexpected results may happen. It would crash an old MAX with only 2GB of RAM at times but more reliable on WIN with 4GB.

THe periodic spikes were from the high speed ADC clock leakage which can be nullled out in software settings.


It is an SA, not a VNA, and certainly not the same as a 100k$ Anritsu VNA. The mix max compare features were useful.
I would put it's value at 20k$ after they fix some simple bugs.

 

[AndreyG]

R54 - single port - interesting. Does it display s-parameters in Smith Chart format?
Frequency range is somewhat odd: 85MHz to 5.4GHz. Low VHF, HF is out, 5.8 GHz is out.
$3k is not much. Can you show us plots?

- - - Updated - - -

I made myself a present by buying KC901S from China. US$1300 on EBay. This is odd device but can be quite useful for Antenna work:
S11 is in vector format, with Smith chart and phase measurements, S21 is scalar. For antenna measurements this is all that's required.
Calibration is noisy but may suffice. It has some programming capabilities not sure how useful. This is my main interest at present.
I just got it today and will be studying it trying to make it do what I need.
I probably post the results if it would be any interesting.

 

[hagster]

@AndreyG yes it does smith charts.

It also does.
Phase, Amplitude and geouo delay
Log, linear
Vswr
TDR(vswr and log etc)
Time gating(for whatever display format)
Z conversion
De embedding.
100,000 points


It just works like a high end VNA, but with just one port. You can use 2 R54s together to do scaler S12 measurements also. It can be controlled via SCPI (like the old GPIB) so it can interface into other software and comes with a number of examples from excel to vb.

I suggest you download the demo software and give it a go. The responsiveness is identical to the real device.

 

[Roger Freeman]

I probably post the results if it would be any interesting.

I'd certainly be interested to see screenshots and to hear how the instrument performs, and what it's like to use.

The problem with lower cost instruments *can* tend to be:
i - the sweep speed is usually quite slow - not easy to do real-time tuning of e.g. filters or antennas
ii - the dynamic range isn't high enough to see e.g. filter stop band performance alongside passband performance
iii - The directivity might not be good enough for accurate measurements - (D wants to be better than 40dB for a 20dB S11 measurement with about 1 dB of measurement uncertainty)

(However, those limitations are not necessarily always a problem).

 

[AndreyG]

KC901S report delayed. Some things I liked however some glitches appeared. Manufacturer rep is involved and I want to wait. I'd rather post good report later the bad now. Wait time ~2 months.

 

[Roger Freeman]

I'd rather post good report later the bad now. Wait time ~2 months.

That's a very fair and sensible approach. I look forward to hearing about it.

 

[Denkman]

For the record, the MegiQ VNA-0440 is around $3000, not 4000 as stated before.

This is a true 2-port analyzer to 4GHz aimed at professional wireless development. See https://www.megiq.com/products/vna-0440.

There is a comparison with the Rohde & Schwarz ZVL and the MiniVNA here: https://www.gsm-modem.de/M2M/wp-content/uploads/2014/04/IoT-M2M-Cookbook-Harald_excerpt.pdf