RE: Hardware

[leif.asbrink@xxxxxxxxxxxxxxxx>]

Hi Jim and All,

>  I have been on a number of web pages, mostly Leif's, and am getting
> to the point that the information is jumbled and hard to find 
> again what I had
> read already. I think I understand the need for a pure LO and 
> good mixer (AM and FM noise 
> rejection), so I will probably be getting an RX2500. The question 
> will be going from 
> 144 Mhz to 2.5 Mhz. I do not fully understand why so many conversion steps
> are recommended. I can see the difficultly for spurs if 
> frequencies are not carefully chosen.
That is not the reason. If you want 144MHz, low end only, a good 
solution would be to use a couple of X-tal oscillators to convert 
directly to 2.5 MHz.
If you start with 15.722MHz and add two triplers to get 141.5 you
will easily be able to pull the crystal 50 or 100kHz on 144MHz.
The IC202 uses this method to pull 200kHz, but at some stage there 
will be a loss of Q causing too much sideband noise.
If you are happy with coverage from 144.000 to 144.400 this will give
very good performance at low cost with easily made hardware.

> Certainly it is useful to be able to QSY using the 
> crystals that are included
> in Leifs design. Since so many conversions will be expensive and 
> difficult (for me)
> to build using easy to find components, I am very tempted to use 
> fewer conversions
> from 144 to 2.5 Mhz. The obvious simple solution is to use single 
> conversion from
> 144 to 2.5 Mhz. I think I can provide very good RF selectivity, 
> so image rejection
> is not a problem. If this proves to be a poor choice due to 
> intermod, then perhaps
> an intermediate conversion into the 25 - 30 Mhz range would help. 
> Intuition says that
> the fewer frequencies involved, the less chance of spurs.
This is all perfectly ok. If you do not need the HF bands this is
what you should go for.

The hardware I make is a set of building blocks that can be used 
in several ways. The IF frequencies are selected to fit standard
equipment. You may for example use the IC202, FT221 or similar
to feed 10.7MHz into the RX10700. Or you may extract the LO from
any such rig and use your own (much better) mixers.
This way you can get the HF bands from any HF rig having a 10.7
MHz IF but performance will be limited by the performance of the HF
rig. Likewise, many modern HF rigs use 70MHz for the first IF.
You may use them with RX70 -> RX10700 -> RX2500 if the HF front
end is good enough for your RF environment.

You are absolutely right in that the many conversions makes my
receiver system rather complicated. I am not worried about the
complexity because the individual boxes are so well isolated
from each other and each unit performs only a single and well
defined task. 


> I have not run any intermod
> calculations yet, but I am not limited to the 10 meter band for an HF IF.
> If I use the 25 - 30 MHz range as an intermediate IF, I would 
> probably try to use my
> CB PLL that has a very wide operating range. This could easily 
> provide IFs in the
> 27.5 to 30 Mhz range. A crystal oscillator would make the first 
> conversion. A lucky
> find in my crystal box along with the wide band CB PLL and a 
> lucky result from a spur study,
> and I am off to the races. If I go with single conversion, I can 
> probably synthesis the 
> needed LO by mixing a crystal oscillator and the CB PLL. Again, 
> another spur study needed.
> My guess is that by converting from 144 to 70 Mhz, 70 Mhz to 10.7 
> Mhz, then finally 10.7 to
> 2.5 Mhz, this produces a set of numbers that make spurs nearly 
> impossible to detect. 
Like always, spurs come up when the mixers go near saturation. 
There will always be things like 18*RF (192.6 MHz) - 15*LO (198.0MHz)
to generate 2.6MHz. This spur sweeps rapidly across 2.5MHz as the
RF frequency is changed and it grows with a very high order of
RF power. This means of course also that it falls off with a very 
high order of the RF power and these spurs typically disappear
into the noise floor at signal levels around 0dBm for the RX10700.
The unwanted signal is then something like 160dB above the noise
floor in 1Hz bandwidth so the spur will be at least 20dB below
the noise sidebands of the offending transmitter if it is a good
one with a sideband noise level at -140dBc/Hz.

With simple mixers like dual gate MOSFETS these spurs would be a
serious problem. When going directly from 144 MHz to 2.5 MHz
even a dual gate MOSFET will produce good performance since
the LO to RF tatio is so close to 1.00.

> Is this  the only reason for it?
No.
The 70 to 10.7 to 2.5 is not selected for being particularly good.
It is for the compatibility with other equipment I have selected
"standard" frequencies. 

> As I say, I can control RF selectivity at 2m. I use good 5 pole helical
> filters for RF. They are available for the LO as well.
OK. But image rejection will probably not be quite as good as I 
like to have it;) I try to make all false responses well below
-140dBc/Hz for a single strong interference signal.

For two simultaneous signals IP3 will set the limit.
Two signals, both at -10dBm, will generate IM3 at a level corresponding
to somewhere about -80dBm. This is 70dB above the noise in 1Hz
bandwidth or about 15dB above the normal noise floor in 3kHz bandwidth
with a noise floor lifted 20dB by the 144MHz preamplifier. 
The IM3 free dynamic range will be about 120dB for the 70MHz IF
input while spurs will start to occur at about 140dB. Reciprocal
mixing comes at 145dB (170dBc/Hz) if no RF amplifier is used.
With an RF amplifier saturation will set the limit somewhere around
+20dBm into the 70MHz IF input. 

> At $12 a pop for the good mixers, I would like to minimize them.
Sure. But to compete with the RX70 chain you need to go for
VAY-1 which cost something like $60....

> I may have missed this too on the web sites, but how much gain 
> ahead of the RS2500 is needed? 
You have two choices. Run it with minimum gain in the Delta44
for a noise figure of about 16dB at 2.5MHz or maximum gain
with a noise figure of about 10dB at 2.5MHz.
With maximum Delta44 gain you loose some dynamic range with
respect to blocking and reciprocal mixing but IM3 will be
slightly better since you need less input to the RX2500. 

If you use no gain at all after the mixer you should definitely
run the Delta44 at max gain. The mixer will then be followed by
a noise figure of 10dB.

> Is an LNA at 2.5 Mhz helpful here to keep the mixers 
> after my front end out of trouble? 
Yes. You should then run the Delta44 at minimum gain.
To not loose more than 1dB by an elevated noise floor,
considering the input noise temp of RX2500, 16dBNF => 
12000K the output temp of your 2.5MHz LNA should be
no more than 4000K so with a NF of 2dB the gain should
be no more than 8dB with a 1dB saturation somewhere
around +18dBm.

> I suppose that selectivity between mixers is helpful too...?
?????
If you go from 144 to 2.5 ?

> 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.
Do you have a cross yagi antenna? For a single polarised antenna
475 MHz may well be enough. With an X-yagi you can start by
using 48kHz sampling. You will then get alias spurs somewhere
around 100dB down but the 40 to 45 kHz useful bandwidth is
quite a lot for 144MHz EME.

73

Leif  /  SM5BSZ



LINRADDARNIL