[hpsdr] ALEX - Call for comments

[Lyle Johnson]

Hello Graham!

  > A block diagram of Alex and other information is on the Wiki.
> The current plan is to use an I2C I/O expander chip to control
> the RF switches, not the full CPLD bus interface as shown.

Hmm, is it possible to also include SPI as an alternative selection 
method?  Looking ahead to Sasquatch, and not sure how the mix of SPI and 
I2C peripheral pins will work out, so I'd like to have both options if 
feasible.  SPI is very easy to implement in a CPLD or FPGA, and requires 
few resources.


> 1.) Is ten filter sections enough?  If more is needed, will
> there be any issues going to a longer card, such as the 160 mm
> length card proposed for Penelope?

Penelope may even be 185 mm :-/

> 2.) Does anyone care if I put SMT chip components on the
> back side of the board, and perhaps a shield?

I suggest that the back side of the board be used for routing control 
signals to the Hittite switches, and the front side of the board only, 
or mostly, carry the RF paths.  It is easy to get stray coupling to the 
control lines and really limit the filter performance.

> 3.) Any preference for SMA verus BNC connectors?  I propose
> to use BNC for "external" connections such as antenna(s), and
> SMA for "internal" or card to card connections.

Sounds good.  Also, "TMP" connectors maybe suitable for between-board 
signals, and sch connectors and cables, while less common, are quite 
inexpensive.

> 4.) Should we allow for multiple antenna inputs? - we could
> select each one on a Tx/Rx and band basis using a GUI on the PC.

Hmm, I wonder if we should keep the board to mostly an internal 
component as opposed to a complete RF front end.  OTOH, if it is like 
the RFE board that Flex made, we'd probably want a strong preamp in the 
receive direction and a bypass in the Tx direction.

> 5.) Using Mercury as a base line, we have an MDS of -140dBm in
> 500Hz without Alex in circuit. Alex will degrade this by say
> 2.5dB on 6m due to the loss through the filter and RF
> switches.  Is an MDS of -137.5dBm acceptable on 6m or should we
> consider adding a low noise preamp after the 6m filter?

If a preamp is included, it should work with all the filter sections, if 
possible, and be bypassable.  It also needs to be very strong so it 
doesn't degrade the IMD3 and BDR3 (or the DDC-based Rx equivalent).  If 
the filters are used to drive a QSD rather than Mercury, having a 
controlled, known source impedance may be helpful (thinking of the RFE 
again).

Thanks for taking this on, Graham, and welcome aboard!

73,

Lyle KK7P



 1173149894.0