[aprssig] 9600 baud capable radios

[Henk de Groot]

Ralph Milnes schreef:
> Henk seems to be saying most radios billed as 9600-capable are NOT
> 9600-capable

They used to be, maybe things have changed. As I showed it can be fixed by 
design so it is not a law that what ever was will be so in the future. So 
I don't know if this is still true for today's radios. Just adding a 9600 
in/out to a traditional PLL radio design doesn't work and this is what 
used to happen.

> George says the V7A and D700 are OK. Henk do you agree with that?  

Sorry, I never looked at their designs. It's pritty sad but in the past 
radios came shipped with their schematics. Now I would not neven know how 
my TH-D7E is put together...

Usability also depends on who is the receiver. I just send a private reply 
to Bob to discuss this topic, I will repeat it below since I guess it is 
interesting for others too.

> George says V7A and D700 do need a good signal path,  and Gerhard points out

A strong signal always helps, especially when the detection margin is low. 
  So the need for a good signal is not an indication that the radio is 
suitable, its when the signal becomes weaker that the differences show. So 
even with a not so good 9k6 radio you may get away with it when the signal 
is strong enough.

By the way, the G3RUH modem has a Rom with profiles to predistort the 
signal to compensate for the distortion introduced by the radio. Selection 
of the right profile can give a good quality link. Experts say however 
that this way you tune the radio to work with one specific other station 
and that this is no guarantee at all that you can work with any station. 
When you are at home and the other station is a node or BBS than this is 
okay, but for APRS every body needs to be able to decode your signal.

> Henk, do you have any examples of the #2-type  radios you mention below?

I mentioned my modified car phone, but I guess that's not what you have in 
mind. I think the TS2000 is such a radio, when you have to cover a lot of 
bands then a simple design doesn't work anymore, so this is one of the 
ways to build it and get good 9k6 performance as a bonus. Since the D700 
and D7 are multiband radios they also may have a similar design. One has 
to examine the schematics to know. I'm not sure about the TS2000 but I 
seem to remember this was the case. Better verify before you buy..

Thanks John for the pointer to the V7A measurements. The TX curve looks 
very nice, I just wonder how low it goes before the PLL spoils it, unless 
of course they took care of that by design. The curves in the picture stop 
at .25 kHz, that is still 250 Hz and you need to be able to get as low as 
10 Hz. Any radio does at least 67 Hz to be able to use CTCSS. It is a 
pitty the reception is not equally flat, but if you read below then maybe 
a soundcard receiver could fix that!

Ok, below is the reply I promised, it was send to Bob earlier this 
evening. I hope Bob doesn't mind a I quote his part too to define the 
context.

Kind regards,

Henk.

<quote>
Hello Bob,

Robert Bruninga schreef:
 > I thought in the receiver, there was LF phase noise because the PLL 
still is a closed loop that will hunt back and forth as it
 > maintains lock and these excursions will add to the
 > data coming off the discrimintator?

Oh, yes, thanks! That's another factor but this highly depends on the 
quality of the VCO. If it wanders only a little then the frequency 
variations will be small, if it is cheap then you get more jitter and it 
becomes harder for the FSK modem to lock to the zero crossing and detect 
the bits.

This problem also exists with the other synthesiser designs, a pure 
crystal tranceiver works best. But the major problem is on the TX side 
when the PLL starts hunting to counteract the "drift" caused by the low 
frequency components in the 9600 baud signal.

 > comment?  (I just accepted that as fact, but may be remembering
 > the wrong fact...)...

No, this is correct, it *is* an issue with PLL designs, a designer has to 
solve this after he solved the first problem: how to modulate without the 
PLL distorting the signal. A good quality VCO is what is needed, something 
a lot of japaneese sets do not have (they just fully rely on the PLL to 
keep the frequency stable). You can hear this in the audio too, it a kind 
of "rumble" in the background.

Also here G3RUH modulation certainly has its drawbacks, a lot of it can be 
overcome with another type of modulation, however at the expense of 
bandwith and complexity. There is no free lunch.

In a digital cell phone system the detector in the receiver is much 
smarter than a simple bit slicer to determine if the symbol is a "1" or a 
"0". The receiver tries to predict what way the signal was distorted 
because knowing that the receiver can just compensate (eqaualize). Not 
only simple amplitude distortion but also the accompanying group delay 
differences rip the signal apart.

Usualy those advanced receivers are used in the basestations. The idea is, 
the basestation send a very clean signal so the receiver in the cell phone 
can be kept simple, the phone has a simple transmitter which does not send 
a very high quality signal and the basestation has an advanced receiver to 
decode the signal. This way the cell phone can be a very cheap mass product.

The soundmodem of Thomas Sailer, HB9JNX/AE4WA, also uses some of these 
advanced reception techniques, as written in his paper for TAPR in 2000 
(http://www.baycom.org/~tom/ham/dcc2000/soundmodem.pdf). He writes:

<quote>
Because the frequency response of the transceiver introduces intersymbol 
interference, the receiver has to be designed to cope with it. I have used 
a constraint length 3 viterbi equalizer to combat the ISI. Constraint 
length 3 captures most of the channel energy, and is about the maximum 
that can be implemented with reasonable CPU consumption. For every symbol, 
8(power)3 = 512 trellis branches have to be processed.
</quote>

The viterbi equalizer is also what is used in GSM basestation receivers to 
counteract distortion. So the soundcard modem is pritty advanced compared 
to the G3RUH modem design, of course at the expense of a lot of processing 
power... GSM receivers use a better equalizer but this is done with DSP's 
especially designed for this job. Also in GSM they use PSK with 4 bits in 
every symbol (QPSK), so it is more advanced than the soundmodem, but I 
think the soundmodem is the most advanced for HAM use.

In APRS this would also be ideal: high quality digis to allow your mobile 
to be plain, straight forward and simple.

Kind regards,

Henk.
</quote>