I/Q modulation alignment tools?

[Elroy Jetson]

Is anyone here aware of any AFFORDABLE service aids which will allow the technician to measure I/Q
balance on any given digital radio, which is NOT platform specific? I have multi brands of P25 CAI
portable and mobile radios and apparently one or two of them are slightly out on their I/Q balance
and amplitude parameters, but I have no way to quantify their parameters on the bench.

Ideally I'd like to find something like a piece of software that can take the demod output of a
service monitor, and read it via the input to the sound card on a PC, and display the I/Q
values in a constellation or bar graph format.

I'm guessing something like that exists, but I haven't yet found it.


Elroy

[Wowbagger]

"Ideally I'd like to find something like a piece of software that can take the demod output of a
service monitor, and read it via the input to the sound card on a PC, and display the I/Q
values in a constellation or bar graph format."

That won't work - the demod out from a service monitor (by which I am assuming you mean the demodulated FM) doesn't have enough data to recover I and Q. FM data is arctan(I/Q) - you lose the magnitude data.

You would have either have a demodulator that gave you I and Q directly, or you would have to get an IF and then sample that and recover I and Q from that. In order for a sound card to handle the data you would have to have an IF of less than about 10kHz to be able to recover the data (assuming a channel bandwidth of 12.5 kHz for standard APCO-25, and a soundcard passband of 20kHz, allowing for the anti-aliasing filter roll-off, you would want the highest frequency of interest to be less than about 16kHz, so that puts your IF at 10kHz).

You also need to have a VERY precise, known good receiver, or else you don't know if the errors are in the transmitter or in the receiver.

Then your software has to correctly sync to the symbol time, slice it, generate an ideal waveform based upon the data, compare the ideal and the actual waveform, do a minimization across {phase rotation, carrier frequency error, I/Q imbalance error, magnitude error} and generate the results.

Sure the signal processing side of things isn't too bad IF you can get reliable data, but the sort of lash-up you are describing....

Yes, I am biased as hell. I know what it takes to make the measurements you are talking about and to do it right, because I have done so in the 2975. I'd be really surprised if you could lash something up that would give you meaningful results. Hell, General Dynamics STILL doesn't do it on their gear (or at least not worth a damn).

[Elroy Jetson]

That's the kind of information I'm looking for.


I really think that these days, EVERY service monitor should have a good general purpose I/Q monitor in it. With digital radios becoming more
commonplace every day, it's only reasonable.


I recently submitted a detailed evaluation of the G-D R8000 to G-D via one of their reps, who visits my shop regularly, and let me use an R8000,
and they really appreciated my input and some changes are going to be made as a result of it. Some useful features are buried too far down into the
menus when they should be readily available, so how about just putting them up on those UNDER-utilized soft keys?

And, they have an I/Q monitor but only for MotoTrbo mode. I suggested that it shouldn't be a major feat to make it usable in other modes, such
as and especially P25 CAI.

And, the operation of the R8000 should not require you to be a veteran of the R2600 to take right to it. Right now someone who's used R2600s
a lot would have an advantage in using the R8000, but someone who's been using IFR/Aeroflex monitors or never had an R2600 generation monitor
won't necessarily adapt to it as easily. The new monitor should not be based on the assumption that you know how to use the earlier one.

Right now I'm using an R2002B as my basic monitor. I'd like to get into something newer, but frankly I don't think the R8000 fits my needs very well
in its current incarnation. So I'm using several other pieces of gear to deliver high precision test results, with the R2002B being used only as a quick
tester. If a radio comes out to be a bit questionable, it gets to meet my Rohde & Schwarz SMY 02 signal generator, my HP 435B power meter w/30 dB power pad, and my HP 8901B modulation analyzer with demod out displayed on my Tek 2465B. Spectrum analysis is done via my Tek 492 and lots of attenuation.
It's by no means as convenient as using the service monitor alone, but it's highly accurate and all in traceable calibration. The 2002B is not, and has some
"issues" with the attenuator.

Elroy

[Wowbagger]

I really think that these days, EVERY service monitor should have a good general purpose I/Q monitor in it. With digital radios becoming more
commonplace every day, it's only reasonable.

There really isn't a "general purpose I/Q monitor" - that's like saying "I want a general purpose cure-everything pill." In order to measure I/Q you have to be able to regenerate the ideal signal (as I stated above) - in order to do that you have to be able to do the following modulation specific things:

1) Determine symbol time. This means you not only have to know what the baud rate is (and this is one of the few times "baud rate" applies!), but you have to be able to determine exactly WHEN symbol time is, so you need a symbol sync detector.
2) You have to be able to slice the analog data into symbols. You cannot just say "this is 8psk" - for example APCO25 CQPSK is 8PSK, but with a mandatory 45 degree shift on successive symbols, so you only get 4 symbol states rather than 8. The device has to know that.
3) You need the appropriate modulation values for ideal symbols.
4) You need the modulation filters for the protocol to regenerate the ideal signal.

Sure, a vector signal analyzer (e.g. an Agilent VSA at US$200K) can measure a large number of different protocols - but that is because the makers of that device have spent the time to create the needed definitions for the protocols. Even if Aeroflex gave you all the hooks needed to program a 2975 or 3900 for any given modulation, you would have to be as competent a signal processing engineer as my group members to make it work for any given modulation scheme.

About the closest that you are going to see is what is out there now: devices with enough hardware flexibility such that different protocols are a software update (read: $$$ to make the hardware, and more $$$ to pay for the software options - guys like me aren't cheap).