Page 1 of 1

Is it possible for 2 Maxtracs to repeat Astro?

Posted: Fri Feb 07, 2003 6:16 pm
by The Pager Geek
Just curious what people think.... Do you think it's possible to tie 2 Maxtracs together to pass Astro Data?


Posted: Fri Feb 07, 2003 6:31 pm
by Bob
My vote is yes. Perhaps not for the faint of heart though; probably easy for someone who knows his way around a maxtrac and surface-mount equipment.

Posted: Sat Feb 08, 2003 4:41 am
by wavetar
I'm gonna vote no. It's been shown not to pass it in a RICK type configuration, nor with Will's module. I've seen it said they cannot handle 2-level FSK for SecureNet either, so I really have my doubts about 4-level C4FM. Plus I don't think it can pass frequencies outside of the 300-3000Hz range with enough deviation to be successful.


Posted: Sat Feb 08, 2003 9:07 am
Minus some technical details it would be very possible if you know where to look.....

Just like they teach in business classes......have to think outside the box....

Posted: Sat Feb 08, 2003 11:05 am
by Pj
I am going to say with MAJOR (not work time/effort in the long run, but good science project) it may be able to work, but if that's the case, we would be see MSF5000's with IMBE add on modules. Too many little details to change.

Posted: Sat Feb 08, 2003 1:41 pm
I guess it was a good thing to not hold one's breath awaiting the 2 ASTRO Spectras back-to-back repeating IMBE voice! No wonder those Quantars cost as much as the family car!

Posted: Wed Feb 12, 2003 3:32 am
by The Pager Geek
You're right Astromodat.. I wouldn't hold your breath for 2 Astro Spectras to repeat astro. It looks like you need a pretty good understanding of the DSP processing in the radio to make it possible. BUT, I'm sure it's possible.

HOWEVER, I was able to make 2 MAXTRACS be "transparent." It's a CSQ configuration, SO.... whatever you throw at it, it repeats... including PL or DPL. You transmit 455.000 with a 186.2pl, it retransmits 450.000 w/ 186.2. If you Tx with a pl of 100.0, it retransmits 100.0.

One cool feature is: IT PASSES ASTRO!!!! (Yes, tested and proven by other members here.) Secure or unsecure astro passes thru w/ no problem. All astro information passes thru with no problem also. Very clear.

Please don't ask for the plans, I'm having them tested by someone with a Astro Capable service monitor to test "bit error rate." Once he returns with his findings, I'll post them. What can I say, I'm a perfectionist.

The modifications to the radio take 15 minutes.. SO.... the answer to the above POLL Question is "Yes, it can be done and it WAS easy." (You will need a service monitor to do it though, "non-astro" is fine.)

Good luck on everybody's quest to figure it out now....

(I can't stand people who say "It can't be done, I've never tried it, but it's not possible.")


Posted: Wed Feb 12, 2003 6:30 am
by wavetar
Hey Astromodat...pinch me, I must be "pipedreaming".

Way to go Tim. I'm happy to be proven wrong in this case.

I just received my basic & detailed service manuals for the Astro Spectra on Monday. On first ain't kidding. I commend the engineers on overdesigning just about everything. There's some parallel stuff going on as well as serial, so it's gonna take me a little while to test my ideas. If it won't work that way, I'll take a look at making it transparent to the C4FM instead, like you did with the Maxtracs (I'm assuming).


Posted: Wed Feb 12, 2003 4:51 pm
by Jay
Pager geek-

Do share when you can....Folks are waiting to see what the story is.


Posted: Wed Feb 12, 2003 5:57 pm
by Mr. Nextel
:-? OK what if you take 2 Motorola radios w/16 pin acc. and make them into a repeater with a R.I.C.K and then have it repeat with no PL or DPL tone so it would be CSQ only. Then program up the Astro radios with the same TX and RE of the repeater would that work? I have always wanted to know but I don't have a VHF repeater that I can use to test my idea. Has anyone tried??? 8)

Posted: Wed Feb 12, 2003 6:04 pm
by wavetar
Mr. Nextel wrote: :-? OK what if you take 2 Motorola radios w/16 pin acc. and make them into a repeater with a R.I.C.K and then have it repeat with no PL or DPL tone so it would be CSQ only. Then program up the Astro radios with the same TX and RE of the repeater would that work? I have always wanted to know but I don't have a VHF repeater that I can use to test my idea. Has anyone tried??? 8)

Many times, with no success. Astro digital won't go through any "regular" (read-unmodified) analog repeater. That's why it's big news.

Posted: Wed Feb 12, 2003 6:05 pm
by alex
No, it doesn't work that way. If it were that easy, no one would buy IMBE CAI enabled repeaters now would they?

There is A LOT that needs to be done to the audio process (and it's above my head) to ensure that the full range of audio required goest through the radios with little or no filtering.

Your going to need a service monitor for this stuff to work.


Posted: Wed Feb 12, 2003 6:06 pm
by Mr. Nextel
:-? I see!!! Well then I want to know. This is BIG news!!!

Posted: Thu Feb 13, 2003 1:04 pm
by The Pager Geek
I see a bunch of ways the DSP communicates in the astro spectra. It's going to be a real battle trying to figure out how to seperate them and then re-enter them into another spectra.

I figure you could feed the audio from the astro spectra and utilize it's COR for keying the other radio. Since a maxtrac is a possible TX radio, have the Spectra set for "Mixed Mode" receive. PL or DPL plus DSQ. Then you have true mixed mode repeater.

Or just make it a closed astro repeater by forcing it to look for only certain Network ID's.

That's my next project.... now I just need another Astro Spectra!!! ;)

more to come.......


Posted: Thu Feb 13, 2003 4:44 pm
by wavetar
I have spent a couple of days going through the Astro Spectra basic & detailed manuals, as well as taking many readings inside the Spectra Vocon & Command boards.

Bottom line...the Astro Spectra clocks the receive data at 2.4MHz, and the transmit data at 1.2MHz. Thus my idea of re-routing the rx serial data directly to another Spectra's tx serial data stream will not work. This is the final serial data going directly to/from the ASFIC (actual IMBE D/A converter) to the DSP (actual data processor). The only other spot to look at on the raw data side of things was the parallel shift register bus between the two IC's, which is well beyond my capabilities.

On the plus side, I am now very familiar with the ASFIC D/A & DSP vocoding processes!

I also spent a good deal of time checking out the various points of both tx & rx audio once converted to analog. I'm quite positive I can make a repeater work by double vocoding (completing the D/A process in the receive radio and forwarding it on to the transmitting radio to be converted once again to digital). The question with that is how well it'll work. As of right now, I only have one Astro Spectra, so that'll have to wait 'til a second one crosses my path.
I haven't as of yet explored the possibilities of tapping the C4FM before the ASFIC gets it, and passing that straight onto the transmit exciter circuitry. That'll take another day or so I'm sure.

I'll post with anything else I find relevant in the near future.


Posted: Thu Feb 13, 2003 6:03 pm
by xmo
"...I'm quite positive I can make a repeater work by double vocoding (completing the D/A process in the receive radio and forwarding it on to the transmitting radio to be converted once again to digital). The question with that is how well it'll work..."

An excellent question - just how well will it work? The question of double vocoding also applies to interoperability between systems and other 'patch' situations. For example - if you have a VSELP system and your neighbor has IMBE, the only way to have interoperability between the systems would be a double vocoding interconnection. In Motorola's console literature, they warn that such patches 'may exhibit degraded audio'.

At work we are currently faced with establishing interoperability with neighboring systems that employ different technology. I am setting up two Quantars with DIU's which will be connected to spare BIMs in our Gold Elite Embassy CEB. Then we will use a variety of Astro Spectras, XTS3000 and Astro Sabers to evaluate a variety of double vocoding scenarios.

I'll post results.

Posted: Fri Feb 14, 2003 3:53 am
by wavetar
Thanks XMO, I look forward to your findings. I guess the same thing would apply if you were to patch UHF IMBE to 800MHz IMBE at the Centracom console. I'm not sure why it would make a big difference....all audio gets converted to TDMA data within the CEB anyway....hopefully just the Motorola engineers covering their butts, and it'll work out just fine in most applications. Good luck,


Posted: Fri Feb 14, 2003 7:51 am
by xmo
"...I guess the same thing would apply if you were to patch UHF IMBE to 800MHz IMBE at the Centracom console..."


That is exactly the sort of situation we are trying to understand, whether UHF to 800 or high band or even a cross tie between a SZ6 system and a SZ3 system.

Just as making a repeater out of two Astro Spectras with a RICK, a console patch will involve double vocoding the voice. While it is true that the console will also digitize the voice, we need to remember that the digitizing process and vocoding are two separate things.

Digitizing (like that done in a console) is done at a high sample rate. The same data is then converted back to analog. This results in a very faithful representation of the original voice.

Vocoding is actually a form of data reduction or compression in order to fit the voice into the smaller channel bandwidth available in two-way radio. The sampling in a Centracom is done at 8K (8192 samples per sec) with 8 bit resolution giving 64KBPS as a data rate (the same as a standard telco T1 DS0).

IMBE vocoding fits the voice into only 4800 BPS. This process results in audio that is quite intelligible but is only an approximate reconstruction of the original signal.

That approximation will faithfully survive further high data rate digitization processes but will suffer degradation if it is compressed a second time.

The question is how noticable is that degradation. Presumeably this will take the form of audible 'artifacts' due to quantization error - it will be an interesting test.

Posted: Fri Feb 14, 2003 12:38 pm
I may be splitting some fine hairs here, but...

Vocoding and digitizing are NOT two different activities. Typically, vocoders are CODECS (Code/Decoders) that are OPTIMIZED to digitize (e.g., encode and decode) the human voice. Optimization is necessary to take maximum advantage of the limited bandwidth available. Although the Telco has much more wider bandwidth available than a typical 2-Way land mobile system, even the Phone Company does not have an unlimited/infinite amount of bandwidth.

The Telco's DS0 64 kb/s PCM building block also uses human vocoders, which are located in their D4 channel banks at their Central Offices (COs), or in the switches' digital line card, depending upon the particular serving arrangement. The US Telco system also employs digital quantization (also called digital compandering) on the human voice within their vocoders. This technique supports quantization by allocating different amounts of sample steps (e.g., differential sampling at varying frequencies of the human voice) to be employed at diffferent fequencies for sampling across the 300 to 3,000 Hz audio passband range.

Keep in mind that there are 256 steps per sample. However, the size of each of these 256 steps can be adjusted to be larger, or smaller, at different frequency passbands. For example, if one desires to encode the low end with more relative integrity than the high end, then these lower passbands can have more than their "fair share" of steps allocated to their particular passband. More steps means that each step is smaller. This allows for a more accurate/precise encoding of the inputed analog waveform (e.g., human voice). Alternatively, at higher ranges, perhaps less audio integrity is needed for good voice reproduction. Accordingly, fewer steps may be allocated to a specified upper range passband. Fewer steps means bigger steps, which results is less accurate sampling of this frequency component of the human voice. This is called the quantization factor of the sampling algorythm, or so-called "digital compandering."

In the USA, the Mu-Law quantization scheme gives more relative amounts of sampled "slices," or steps, at the low end of the range vs. the high end of the 300 - 3,000 Hz range. On the other hand, the Europeans use so-called A-Law quantization algorythm in their Telco systems. This European approach is different, and somewhat the opposite of, our Mu-Law technique. Within the 64 kb/s, the European approach allocates more sample steps at the high end of the spectrum. (Remember that Europeans must always be back asswards from us folks in the USA. That's why their helicopter rotor blades turn in the opposite direction as US helicopters).

In all sincerety, there are tons of elegant white papers that discuss the Europeans' voice ranges, etc. that attempt to justify these different quantization schemes. After you read the first paragraph, these papers turn into a pure Math exercise. Good way to put oneself asleep if you ever get insomnia!

My point is that Telcos also use vocoders! They digitize human voice, and 2-Way systems digitize human voice. Yes, the Telco vocoders are much more robust than VSELP and/or IMBE vocoders, but that is because the Telcos have a huge amount of bandwidth, as opposed to our "skinny" 2-Way systems. The difference between the Telco vocoders and typical 2-Way land mobile and/or cellular vocoders is that the Telco landine 64 kb/s PCM system has nice copper wires to your home (read that "tons of bandwidth!"), and fiber optic cables pretty much everywhere else. The Telcos can "afford" all of this extra bandwidth that is unavailable in a narrow band, 2-Way land mobile system.

Digitizing is a very generic term. The term "vocoders" simply refers to CODECS that are optimized for encoding/decoding human voice. However, they can use VERY different sampling rates, word (sample) sizes, and thirdly various quantizing schemes. Yes, the Telco sytem is the very best, but that's because they have way more available bandwidth as compared to narrow band RF 2-Way systems.

One last point on quantization schemes. This gets very complex! For instasnce, the cellular companies utilize optimized human voice vocoders, and associated DSP, that "looks for" wind noise, air conditioning noises, etc. and other pre-modelled sounds. These sounds are "discriminated" against by quantization algorythms that allocate extremely few sample steps here so as to concentrate more on the human voice encoding, and secondly, in helping to filter out these annoying sounds. Makes for a non-realistic audio integrity (for non-human voice sounds), yet it results in a much better human voice quality. It's these types of schemes that differentiate the typical 2-Way radio vocoders from the Telco's vocoders. The Telco vocoder does an excellent job on the human voice (and it's definately optimized for same!), but it ALSO can properly encode other sounds, as well, such as DTMF. This latter distinction is probably what xmo is referring to.