P25 LSM (Linear Simulcast Modulation) Difficulties

[KA1RBI]

[cross-posted from op25-dev]

Locally we have a Motorola system that runs in the 800 MHz band and is believed to be running with LSM. I have been having extreme difficulty in achieving correct symbol recovery from any of the channels in this system, even though the signal strength is very strong here (there is another more distant P25 system, weaker on the receiver S meter, that comes in 100% perfectly).

I've read that LSM uses CQPSK modulation; that it uses a pulse-shaping filter that is slightly different than the "standard"; that C4FM is used for the uplinks, LSM is used on the downlinks only; and that there is some controversy as to whether it should qualify as fully "APCO P25 compliant...

It is also interesting to note that the Radio Shack people have introduced a firmware upgrade to the DSP software for their scanner radios which supposedly make these radios work with LSM now.

When looking at received signals on the datascope, there are periods where everything looks fine, with four distinct horizontal bands, but there are frequent "hits" in which the entire display is filled with random noise.

There is a Motorola PDF file (can upload if anyone is interested) that talks about LSM but is not very technical.

Also I've heard that in order to properly receive LSM (and CQPSK) requires an I/Q receiver. This would imply that the Radio Rausch decoder (which is excellent for what it does work with) would not be usable. This is because the Radio Rausch decoder accepts demodulated baseband (float) input rather than I/Q (complex) input...

Also have tried using the Gnuradio mpsk_receiver block on it but it won't even phase-sync with the signal. The mpsk_receiver block is going to need extra code someday, so that it understands PI/4, but even so it should show some semblance of trying to lock to it, which this isn't.

Any ideas or tech tips would be *much* appreciated!

Thanks & 73

Max

 

[WayneH]

I would try contacting Wowbagger on the Batlab's forum. I would think he could provide some insight in to this. Maybe you'll get lucky on here though I haven't run across anyone who really knows LSM inside and out.

 

[qball]

I can confirm that the Pro96, 396, and 996 (all with the latest firmware) don't do a great job at decoding LSM. I use Pro96com with a Pro96 and it can't keep up with the control channel. I am 2 miles line-of-sight from a 300' site.

As for using the radios listed above; a 15 element yagi or some sort of corner reflector to null all but the closest site is the only thing that I believe may improve the rx capability. I have tried attenuators, paperclips for antennas, and semi-directional antennas all to no avail (haven't forked over the cash to buy a yagi). The only thing that works almost always is to leave the desired coverage area (leave the county) to find that the scanners decode pretty well.

As long as there is a great deal of simulcast site overlap, you may have to hang up searching for that perfect audio. An XTS5000 may be the only saving grace, and that may prove not to work if there is too much overlap from really distant sites. I have seen a tall site 20 miles away absolutely destroy outbound audio quality because of timing/signal issues. The timing couldn't be resolved so the antenna gain from the tall site was cut in half to reduce the overlap....and it mostly corrected the problem for the public safety radios. However, the scanners poorly decode the trunked system anywhere.

I know this doesn't really answer your questions, but it is my experience in attempting to decode LSM. I would like to hear how you resolve your issue if you are able to find a solution(s). Good luck!

q

 

[KA1RBI]

Well there is one new piece to add to the puzzle. From this location (outside the county in question) I'm seeing very severe fluctuations on an oscilloscope hooked to the AGC line when tuned to the control channel. This says that there are wild swings in received signal strength (perhaps as much as 20-40 dB).

It would be interesting to know if all of this variation is due to random adding and cancelling of the waves at the receiving antenna, or whether the transmitters are maintaining constant output power or not...

Thanks for the replies

Max

 

[N_Jay]

Well there is one new piece to add to the puzzle. From this location (outside the county in question) I'm seeing very severe fluctuations on an oscilloscope hooked to the AGC line when tuned to the control channel. This says that there are wild swings in received signal strength (perhaps as much as 20-40 dB).

It would be interesting to know if all of this variation is due to random adding and cancelling of the waves at the receiving antenna, or whether the transmitters are maintaining constant output power or not...

Thanks for the replies

Max

There is no way a properly operating system has 20 to 40 dB carrier strength changes.

How fast are the swings?
Are they constant?

 

[KA1RBI]

How fast are the swings?
Are they constant?

The scope was an analog model, ideally would like to get a digital sample of the AGC voltage over a reasonable time period, for analysis. Very roughly I'd say the swings were occurring at a sub-audio rate, perhaps 50-100 Hz or so, or less. Definitely low frequency. The waveform does appear to have some repetitive-ness, but it's not a clear recurring pattern. Also, there is an AGC time constant that could be smearing or low-pass filtering the display that I see....

It should be noted that there are other digital systems I can tune in which the AGC is rock-solid, the "problem" only affects this one particular P25 system, and is consistent.

Thx & 73

Max

 

[N_Jay]

The scope was an analog model, ideally would like to get a digital sample of the AGC voltage over a reasonable time period, for analysis. Very roughly I'd say the swings were occurring at a sub-audio rate, perhaps 50-100 Hz or so, or less. Definitely low frequency. The waveform does appear to have some repetitive-ness, but it's not a clear recurring pattern. Also, there is an AGC time constant that could be smearing or low-pass filtering the display that I see....

It should be noted that there are other digital systems I can tune in which the AGC is rock-solid, the "problem" only affects this one particular P25 system, and is consistent.

Thx & 73

Max

Is that the only simulcast system you are monitoring?

Is it an older system?

Could it be "Wide-Pulse" and not LSM?

 

[PHOENIX_SCANNER]

I live with a large simulcast system, and with a handheld, reception depends on being very close to one tower, relative to the others. If you are somewhat between towers, its rough.

At home matters are much better. I have a dedicated yagi / scanner combination just for this system. It is pointed so it attenuates all but my primary tower as much as possible, per the field pattern of the antenna, and it works pretty well. To really attenuate the 2nd and 3rd closest towers, I end up pointing it about 11 degrees off of my primary tower oddly enough. Pointed RIGHT AT the primary tower, one of the other towers starts having an impact, as that angle on the field pattern starts allowing it to be less attenuated it starts doing bad things to reception. Keep in mind, being 11 degrees off meant that I went from a gain of 11db to 8db, but the attenuation of the number 2 tower went from -20db to -50db, so the ratio was MUCH better this way. The field pattern of a high gain yagi is a pretty funny looking thing, with sharp dips in places. So really, in this case you are using a yagi in a way that was never really intended, as an ATTENUATOR, more than as a device to pick up a weak signal from one location, as it was intended to be used. This took many hours of tinkering to figure out, but in the end it makes sense.

My point is, pointing a yagi right at your nearest tower may well not be where it needs to be pointed. A good ratio of signal strength is the key, and that comes from plotting out nearby towers and aiming the antenna based on its field pattern to achieve the best ratio. Properly pointed, it works like a charm. Not 100%, but with the new PSR-600 / Pro-197 with updated dsp, its probably about 95% reception, and that is about as good or better than the Police portables get, judging by the number of times they have to ask for a repeat because of a break-up or someone "going digital".

Now, as these systems become commonplace, I hope that scanner manufacturers start including the guts to make these things work like police radios do. However, I suspect that will mean a BIG jump in price. I have heard a Phoenix PD portable is almost two grand, but since the scanner doesn't have to transmit, maybe it can be done for less than half of that. People like myself would come up with the money if it worked well.

 

[N_Jay]

Even control stations usually use a yagi or corner reflector antenna.

It is always hard to receive a simulcast system if you are outside the designed area of coverage.

You may want to pay $1000 for a "good" receiver, but there are not many "You's" out there.

 

[slicerwizard]

To really attenuate the 2nd and 3rd closest towers, I end up pointing it about 11 degrees off of my primary tower oddly enough.

Nothing odd about it. That's what you're supposed to do when you have competing signals.

Same deal with AM DX at night - rotate the loop antenna to null the unwanted signal.

 

[KA1RBI]

Is that the only simulcast system you are monitoring?

Is it an older system?

Could it be "Wide-Pulse" and not LSM?

So far as I know it's the only simulcast P25 system that I can hear from here. It's brand-new (Tompkins County, NY) and was made by Motorola; I couldn't locate the radioreference thread where this system was discussed at length, but as I recall it was LSM, yes.

Max

 

[wlmr]

It's probable that you already know what I'm going to share - but here goes anyway!
(I'm also open to corrections if any of the info below doesn't seem right, the training I attended was a while back already and I've been known to suffer from CRS at times.)

Could be that the reason your scope doesn't want to show a pretty display is the fact that you are likely receiving multiple signals from the different sites that make up the simulcast site.

Remember, simulcast as far as P25 goes means the exact same signal being broadcast simultaneously from all the tower locations that make up the simulcast site. The radios that work with the system are designed to deal with multiple slightly out of sync arrival times from the different transmitters but it seems reasonable to expect your display would show a mess, especially compared to a single tower source signal.

If you can get where you only receive one of the transmitters I suspect that your display would be cleaner. That's what one of the earlier posts by PHOENIX_SCANNER is alluding to, using a directional antenna to null out or at least severely attenuate all but one of the sources.

As far as the signal strength from the site varying in amplitude, my understanding is that neither the transmit power output for the P25 control channel or the voice channel(s) is a smooth level. Multipath from several simulcast transmitters can only mess that up further. (The % meter on a Pro96 scanner sitting here by me bounces around badly on simulcast sites & is steadier on single sites.)

When system tests are done for either power output levels or received signal strength testing, the transmitters are told to generate a special test pattern (try googling "P25 V.52 test pattern"). This pattern is more consistent so the transmitter's resultant power output is smoother & makes measuring things much easier. As an added bonus, the receiving equipment can decode the test pattern and test for bit errors. Too bad scanners aren't designed to indicate that's what's going on when you hit a signal that's sorta P25ish but can't be decoded as a control or voice channel.

I've droned on long enough, hopefully I've helped.

 

[KA1RBI]

As far as the signal strength from the site varying in amplitude, my understanding is that neither the transmit power output for the P25 control channel or the voice channel(s) is a smooth level. Multipath from several simulcast transmitters can only mess that up further.
I've droned on long enough, hopefully I've helped.

OK, that answers one of the most crucial questions I had, which was whether any given transmitter in the system transmitted a constant or varying power envelope. Many, many thanks for this info and for all the posts on this thread - 73

Max

 

[MattSR]

Hi Max, have you had any progress with the I/Q based decoder? I know Steve has done a few things recently on OP25..

 

[KA1RBI]

Hi Max, have you had any progress with the I/Q based decoder? I know Steve has done a few things recently on OP25..

Greetings

It has indeed been a lazy summer (at least here in the Northern Hemisphere ;^), but there are a few tidbits to report... (Sorry if any/all of this comes as old news)...

1. Stevie informs me that the new GNURadio 3.2 contains a bugfix that is apparently needed in order to use the new CQPSK demodulator (which is separate from and independent of the Radio Rausch demodulator). The project is seeking for further feedback/testing/comments on this demodulator. It's currrently in the python/ branch of SVN (files cqpsk.py, cqpsk-mod.py, cqpsk-demod.py).

2. I have a couple of tutorial/project pages which are at:
A 455 KHz IF Downconverter for Digital Radio Reception
Constellation and Eye Pattern Diagrams

The /iq/ page consists of a project to build a downconverter to interface the 455 KHz output of a receiver to a sound card line input; the I and Q channels are fed to the Left and Right channel inputs. It's Receive-Only for now, but it will produce signal-capture files that are compatible with those generated by the USRP (albeit at a much lower sample rate).

The /diagrams/ page has the starting point of a general-purpose data-scope/viewer...

3. No further work has been done to try to crack the LSM problem. At that time the only demodulator available was the Radio Rausch (which used signals already converted to demodulated FM baseband). The newer CQPSK demodulator may (or may not) work better in this environment. It might be interesting to note that the CQPSK demodulator requires I/Q signals as input; therefore, it won't work with the simpler (discriminator tap) method.

Best Regards

Max

 

[MattSR]

Awesome - Is there any code in OP25 that supports your 455KHz downconverter? What im getting at - is there any point in me attempting to build one and use it as a demodulator? I already have a disc tapped scanner but I never got it working as an source to Franks decoder.

I'm committed to buying a USRP once transmit code appears in the trunk

 

[KA1RBI]

Awesome - Is there any code in OP25 that supports your 455KHz downconverter? What im getting at - is there any point in me attempting to build one and use it as a demodulator? I already have a disc tapped scanner but I never got it working as an source to Franks decoder.

No code's been released to accompany the IF downconverter. It needs to be added!

As to the question "what good is it?" - I built mine as a way to have access to:
1. samples in complex format (i.e., data compatibility with the USRP), and
2. the RF signal prior to passing through the radio's FM demodulator stage

We currently have two demodulators (Frank's plus the CQPSK block). Perhaps it should be noted that this refers to demodulating the C4FM/CQPSK and not to FM demodulation. One essential difference between these two demodulators is that Frank's Radio Rausch demodulator requires a (float) input that is already FM-demodulated, whereas the CQPSK demodulator requires a complex (I/Q) input. Among other things this means that the CQPSK demodulator block requires either a USRP or the 455 KHz downconverter whereas the Radio Rausch demodulator will work with the USRP, the 455 downconverter, or the discriminator tap.

I don't think we've had enough experience or tuning yet of the CQPSK alternative in order to know if it has significant advantages over Frank's...

Speaking of the disciminator tap, I have a new ALSA script that hopefully should get uploaded within the next few days that should work much better. In the meantime if you can - send me an audio capture of your disc. tap - something like this :
arecord -f dat -d 5 cap.wav

I'm committed to buying a USRP once transmit code appears in the trunk

Depends on what you mean by "transmit". There's a CQPSK transmit block already...

73

Max

 

[MattSR]

Hi Max, Yep - when I said "is there any point" I was kinda meaning "is there code to support it"

I understand the benefits of I/Q decoding - your webpage has helped me greatly in looking at the constellations and getting an idea of whats going on. Its great stuff and you are to be congratulated for it

Interestingly P25 repeaters don't transmit CQPSK at all - its only ever used for the uplink from mobile to base station, so ideally a C4FM transmitter would be used.

I'll have a play around and grab a sample and let you know of the yahoogroup once its done.

Cheers,
Matt

 

[KA1RBI]

Interestingly P25 repeaters don't transmit CQPSK at all - its only ever used for the uplink from mobile to base station, so ideally a C4FM transmitter would be used.

Is this true ? I thought P25 repeater systems (that is, those using LSM) worked as follows: [?]
- for mobile-to-repeater, the mobile transmits C4FM
- for repeater-to-mobiles, the base repeater transmits CQPSK
- for the trunk CC, the base station also transmits CQPSK

Max

 

[MattSR]

I will dig through the documentation and find whats correct...

Thanks for the update to the trunk BTW, assuming work doesn't get me to do any crazy 1am changes ill have a play with it tonght.

Regards,
Matt