[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [linrad] RE: Hardware

Hi Jim and all,

> My IC202 is grossly unstable. I think it would be a poor choice to use a 
> VCXO that drifts and provides frequency uncertainty in an otherwise
> top notch receiver.
If you need the phase noise performance it would be an 
extremely good idea. The frequency uncertainty is no problem
at all, just press the key of your stable tx and you will
immediately have Linrad calibrated:)

Frequency drift is another thing, that is (in part) why I 
suggested 50kHz tuning range rather than 200kHz. 

> My plan would probably sacrific a bit of phase noise
> performance for frequency agility. I can produce a signal by multipliying
> a temperature compensated channel element from a Motorola Micor operating
> at 12.945 Mhz. It would be multiplied by a modified Motorola 
> Micor VHF exciter by
> a factor of 9 to 116.505 Mhz. The 25 Mhz PLL signal is mixed with 
> the 116.505 Mhz frequency
> to produce the needed 141.5 Mhz LO.
Try it. If the LO sideband noise is a problem you will soon
notice;) Just listen to the IC202. If you see noise on the sides
above something like -135dBc/Hz your LO is the limiting factor. 

> The intermod study for this set of frequencies suggest 
> that some intermod from the 146 - 147 Mhz range may be 
> possible, but most  combinations include
> 2.5 Mhz as an input, suggesting that a strong signal must exist 
> inside my passband.
You do not need to include 2.5 MHz, the same product can
be described as n*RF m*LO also.

> I will probably not have this during normal EME.
It depends essentially on the mixer. If the mixer is far from
saturation there should be no problem.

> By the way, that should be 141.55 Mhz
> L.O.  if the middle of the Linrad window is to be at 144.050 Mhz? 
> That would cover the
> random range nicely with 96KHZ bw.
Yes - but remember the audiocards are AC-coupled. You will get
a very deep notch exactly at 144.050 with a width of about
20Hz. It is nice to be able to shift the frequency a little.

> Each additional box adds cost

> and frequency drift

If you do not lock the oscillators to a frequency standard
the 141.5MHz will drift by typically 5Hz per degree. There will
also be a short time random FM modulation with a bandwidth
of something like 0.1Hz

The errors are proportional to the frequency. With the
RX144 -> RX70 -> RX10700 the LO frequencies are 74, 59.3 and 13.2
If the temperature was changing equally the termal drift would be
proportional to 146.5 MHz rather than to 141.5 MHz so the difference 
is very small and caused by my use LO above RF in one case.
The short term FM modulation will be much better because it
will be the sum of three uncorrellated parts so it will be
proportional to 95.7MHz, only 67% of the modulation expected
from a single oscillator at 141.5MHz.

If the oscillators are locked to a GPS or similar, it does not matter
which solution you prefer.

> in the best case and additional IMD and noise in the worse case
Hmmm, with a lower LO frequency you may find it easier to make
a really good mixer. The RX70,RX10700,RX2500 combination is designed
for an input signal level of 0dBm or less. IP3 is at about +30dBm
with a noise figure of something like 18dB. There are some details
not finally decided yet so there may be small changes.

If you go from 144 to 70MHz or from 144 to 2.5MHz by use of a TUF-1H
performance will be identical and entirely determined by the SRA-1H.

If you go from 144 to 2.5 with a VAY-1 mixer ($60 or so) you are right.
Then you may perhaps gain 3dB in IP3.

> but they can enhance image rejection and Spur rejection. 
> I like KISS in this case.
It is a matter of taste. To me the well defined 50 ohm boxes
represent a straightforward solution. I can verify what is
happening step by step by injecting test signals in-band and 
out-of-band at the various IF frequencies to make sure spurs
are under control. If you make the 141.5 LO as the sum of
several frequencies (the n-th overtone of one plus another)
it may be difficult to know what is going on at high RF levels.

> Is there a 2.5 Mhz LNA circuit I can copy?
The 2.5 MHz amplifier at the output of RX10700 is one
example. Similar to the input amplifier of the RX2500.

> I am not sure that
> you said how much front end gain is needed. I guess the TUF1H
> adds 6db of noise figure, so if an LNA at 2.5 Mhz adds 2db
> of noise figure then 8db of noise figure has to be overcome
> at 144 Mhz. This 2.5 Mhz LNA is probably something I can
> add later.
Not quite like this.
If the LNA has a noise figure of 2dB, you will have to
make the system noise figure 5dB if 50% of the noise
is allowed to come from this amplifier. You then degrade
the Delta44 noise floor by 3dB!
If you allow 25% of the noise to come from the 2.5MHz LNA
and 75% from Delta44 (and RX2500) the system noise figure 
will be about 8dB at the LNA input. At the mixer input
you then have probably NF=14dB at 144MHz. With a 3dB
attenuator to make the RF port reasonably well matched
at all frequencies you have NF=17dB at 144MHz. 

Without the 2.5 MHz LNA the NF would be 25dB if the
Delta44 is run at low gain and 19dB if it is run at 
high gain but then you have a degraded noise floor.

I suggest you start without the LNA. You may want to 
bringg the IF noise figure down to say 2.5dB while
loosing 10dB of inband dynamic range if the dominating
problem is IM3 from stations outside the visible passband.
If on the other hand the dominating problem is fellow 
weak signal enthusiasts within the passband you will 
want a low gain LNA as I suggest above.

> I wonder, since I am not changing the PLL to a new frequency to switch
> from TX to RX, as it does in a CB, if I should experiment with 
> loop filtering
> time constants to reduce PLL phase noise at a later time? I guess Linrad
> can be used as the tool to optimize (minimize) phase noise?
Yes. And your good old IC202 will be a good test signal.
Check the latest QEX for a hint how to improve it if you
really reach as far as to the flat noise floor it emits.

> > > Per Conrad, it looks like my Pentium 475 Mhz computer will not be
> > > fast enough for proper operation at 96 KHZ bandwidth. I will
> > > continue my search for a capable Linrad PC.
Every day by which you postpone a computer purchase will give you the
same computer at a lower cost. A good idea is to start with what
you already have while building and verifying the RF hardware.
When all is running the computer that really fits the task has
become significantly cheaper:)

> Considering the relatively large cost of the Delta 44 card 
> and the RX2500, it does not seem that expensive to have a 
> proper PC to do the full 96 KHZ bandwidth. 

> I do not wish to make an early investment obsolete.
> Better to plan for the long run and not waste money 
> in the short run. 
Sure, but the PC purchase should be the last one. If the
way you generate the LO signal is too noisy it might take
some time to do it some other way. Any old PC will be 
good enough to verify the LO quality because you can 
sample slowly while checking the hardware.


Leif / SM5BSZ