Re: [Discuss-gnuradio] 4702 usrp daughterboard

[Eric Blossom]

On Sat, Feb 07, 2004 at 08:26:52PM -0800, cfk wrote:
>     All right. Let me flesh out the idea a little bit more and ask a few
> questions for you and Matt. First a procedural thing. Although I am most
> partial to Linux (capitalization intentional), my schematic capture skills
> are best with Orcad and Orcad Layout. I hope that you have no problem with
> me doing the schematic in capture. I believe that I have Acrobat Distiller
> on my windows (non-capitalization intentional) machine and can transfer both
> capture design files and pdf files for the schematic and both layout
> database and gerber (and perhaps pdf) files for the layout. This only has to
> do with the fact that I can think & draw faster in capture/layout then
> xcircuit/pcb.

Fine by me.  FWIW, Matt's currently doing schematic capture with a
free tool (gEDA <http://www.geda.seul.org/>), but the board folks are
using PADS.  PDF, ascii and gerber seem to be the common denominator.

>     I was originally thinking of a companion board about the same size as
> the usrp, but the notion of a daughter board implies a slightly smaller
> board then the one to which it is attached. I think we need to ensure that
> the size, mounting configuration and connector placement of all the parts on
> this board are consistent with your and Matt's vision of how it should look.
> If we are clever, then we can have a form factor which makes a 1.2Ghz,
> 2.4Ghz, VLF or other form factor all the same.

That's the plan.  The 4702 module is a little big, but should be
doable if we hang it over the edge a bit.  The daughterboard cards are
about 50mm x 50mm.  We figured they'd be surface mount.

>     In thinking of the output circuit from the 4702, would you or Matt have
> an idea of the quadrature demodulator you might like to use.

I think you can get by without a demodulator.  Assume bandpass
sampling and put the 36 MHz IF into a single channel (I). You can
probably do this with a mini-circuits RF xformer.  Properly terminate
the Q input and you're done.  The A/D's are running at 64 MHz so
you'll fold down to 4 MHz which is not so great for the 8 MHz SAW, but
OK for 7 MHz.  Matt: comments?

> on Maxim and found a MAX2451 about an hour or so ago, but I have not used
> this particular chip before so I am a little unfamiliar with its care and
> feeding. Perhaps you might have a better one in mind. Additionally, I
> suspect we may want to buffer with a dual or quad opamp the output of the
> quadrature demodulator chip as it leaves this board on the way to the
> AD9862.

The buffer may not be needed.  We've got 20dB of programmable gain on
the AD9862 and the total trace length will be pretty short.

>     I am thinking of a four layer board with power/ground in the middle and
> signals on top and bottom. Not because it will be tight (it wont), but I
> believe we get a better noise floor with a power/ground sandwhich. Its a
> little better if the ground/power are on the outside by a dB or so, but then
> you cant change things on the circuit board hardly at all.

OK.

>     On the 24LC024/025, which do you want? The 024 or the 025? Do you care
> if its a DIP or do you wish the TSSOP package. I see this is also an IIC
> device, so I presume tying their SCL/SDA to the other chips in one bus with
> the WP* brought to a pin of yours or Matts choice on the interface connector
> is the appropriate circuit for this one. I have used IIC eeproms before,
> although I have been partial to Philips, the Microchip ones look very
> similar. VCC=5V for the 024, I presume.

24LC024 or 24LC025, doesn't matter.  We won't be using the WP pin.
I suggest TSSOP package.  The daughterboard bus sends up the I2C A0
and A1 pins depending on what slot the board's plugged in.  A2 should
be high.  No need to program it ahead of time.  The part can be
programmed through the host software as part of check out.
The LC parts run on 3.3V.  All digital i/o is 3.3V.

>     You'll have to help me out on the IEC connector a bit. From the data
> sheet it appears it is an "IEC 169-2", which looks like a european version
> of a cable TV inlet connector. I'll have to get my $6 to Matt and get two of
> these down here so I can scope that one out and make sure we are not trying
> to put square pegs into round holes for our US engineers.

You're right.  It's the Euro cable TV connector.

Send me your snail mail address and I'll get a couple of modules out
to you.

>     My original thought on the LNA was that it would be in one of those
> seperate little boxes with a male connector to mate with the tuner on one
> side and a female BNC on the other. Those little guys are about 1 inch X 1
> inch X 0.5 inch thick with ground a terminal on the case and power as a
> feedthru. I think mini-circuits, Triquint and others have that sort of
> thing. Perhaps you might have a part/model number in mind. I'll also look
> over our LNA's on Monday and see what we are using in the lab. The thought
> went that we provide a 2 pin MTA-100 header with power/ground for the LNA
> box on the 4702 input since the 4702 is a big item sitting off the board a
> couple of inches with its input snout off the edge of the board.

Nothing in mind.  Let us know what you come up with.

>     I think I can get a fur piece along the path if I can get an assembly or
> top layer gerber of the usrp, along with a sketch of how you and Matt would
> like the form factor to look. In the meantime, I'll gather my thoughts
> together and start the schematic capture. I will plan on iterating the
> schematic and layout a couple of times. It is a lot easier to move the
> little pixels around on the screen before the board is fabbed. A design
> review via e-mail would also let me have confidence that this is what we
> want. I shant draw that out, I'll put it together, publish it, iterate it
> once, accept comments a second time and them we cast the result in
> fiberglas.
> 
>     Maybe Brian can jump in here with his Maxim knowledge and help us out
> with the schematic a bit as it fleshes out over the next week or so.
> 
> Charles, WA6LWB
> 

Thanks for diving in!

Eric