Re: [Discuss-gnuradio] frontend4702 schematic and layout

[Matt Ettus]

Quoting cfk <address@hidden>:

> I'm really looking for guidance here as I havent used the Microtune device
> before. The reason I put the 5 resistors and 2 capacitors was to match your
> schematic on the Wiki. In looking through the Microtune data sheet, I am
> guessing it is a 300ohm output impedance based on their test circuit in
> section 6. Thats also where I got the 680ohm resistors from. Certainly we
> can not stuff (DNI, or DoNotInstall) some of the resistors/capacitors on
> ifout1 & ifout2 coming from the 4702 tuner until we figure out its proper
> care and feeding. We can also put zero ohm resistors or change to other
> values for the same reason.

I see.  I can't find anywhere in the datasheet where it says what the output
impedance on the 4702 is.  That circuit you have shown will work fine as long
as we change those resistor values (which, as you point out, we don't have to
finalize now), so I would keep the topology as-is.  The values shown have 22dB
of loss and were for their test circuit.

> Perhaps someone else might have an opinion on the newsgroup regarding a
> better guess at what the first cut of this daughterboard should have between
> the ifout1/ifout2 of the Microtune 4702 and the input to the USRP.

I'd be interested in hearing anyone else's experience, since Eric has sent out a
bunch of them now.


> > I'd also repeat my pleas for allowing 2 of the 4702's to be hooked up.
>
> I shall be done. I just didnt get it before. Sorry for my denseness. With
> two 4702's then we get I/Q for free, right?

Not really.  Basically, some up/downconverters use IQ in order to do image
rejection easier.  Since we would be sampling both I and Q, the image reject is
done mathematically.  In the case of the 4702, they downconvert internally, and
they might be using filters or IQ methods to do image reject.  I'm not sure
which, but it doesn't matter because they only give us one output with the
images already removed.  So we would sample this "real" signal with one AD
converter.

Now since each daughterboard has access to 2 analog channels, we have a spare. 
By having a second 4702 we can listen to a second antenna on the same or a
different frequency, or (with a splitter on the antenna input) to a different
frequency on the same antenna.  This makes for a lot of interesting
applications, like "smart" antennas, direction finding, listening to multiple
signals which are widely separated in frequency, etc.

> For the sake of argument, lets assume the daughter board is in the upper
> left slot. The one where the interface connector is J668. On the layout, the
> 4702 is a through-hole part and shouldn't be over the top of the USRP
> interface connector. What I am currently thinking is to rotate the 4702's so
> that they come out the left side of the board. I would need to put one above
> the USRP interface connector and one below the USRP interface connector.

If it fits, I think this is the ideal setup.

> Each 4702 is a littel over half an inch or about 12.8mm. It is a really
> tight fit to put two above the interface connector. The rest of the parts
> can basically fit anywhere, with the proviso that the digital stuff needs to
> be as far away from the analog input to the USRP as possible.

Can you fit one below?  The lower-right mounting hole seems to get in the way.

> If I do this, it would mean making the daughterboard a little wider then
> 55mm to allow the mounting tabs on the ends of the 4702 to have some board
> to fold around. I actually measure the 4702 at 61mm and change. There would
> then be two of them as you had described. One would feed the A input and the
> other would feed the B input.

Great.  When making the board a little wider than 55mm, be careful you don't run
into the USB connector, J401, when the board is mounted.

> I think that a single size Rx board would hold two 4702's that way. Perhaps
> you can let me know if this causes you any consternation by my being a
> little fast and loose with the size of the daughterboard and making it 65mm
> hi by 61.75mm wide instead of 65mm hi by 55mm wide.

I think my calculations on board sizes were wrong.  Let me fix that here, and
I'll update the wiki page.  I hope this won't cause you any problems.  Also,
although I prefer metric myself, the board layout guy did everything in mils,
so that's how I'll do it here.

The "basic" size is 2000 mils by 2000 mils.  There are 4 holes in this basic
square.  3 of them are 150 mils away from the corner both horizontally and
vertically.  The 4th hole (the one closest to pins 1 and 2 on the interface
connector) is 300 mils away from the corner in both the horizontally and
vertically.

If you grow this basic size horizontally (along the major axis of the interface
connector) on the side nearest to pin 64 on the interface connector, your
daughterboard will overhang the edge of the USRP.  The amount you can grow this
way is not really limited except by how far you are willing to overhang.

To grow horizontally on the side nearest pin 1 of the connector (towards the
center of the USRP) is somewhat more limited.  You can grow it by as much as
600 mils, but you must make a cutout to fit the power and usb connectors.  The
notch needs to extend 300 mils from the edge closest to the edge of the USRP.

To grow vertically on the edge closest to the broad side of the interface
connector, you have 250 mils.

To grow vertically on the edge furthest to the broad side of the interface
connector, you are essentially unlimited, but will be overhanging the edge of
the USRP.

I hope that clears things up :)

Matt