RF/IF board
The orginal design seems to be fairly solid in the RF area. I looked at a few things I might like to change. Mostly, though re-packaging was my initial goal for this board.

Improvements
Bob Larkin pointed out that the gain is all ahead of the filters so overloading could be a problem and selectivity and noise figure isn't as good as it could be. On the flip side, Bob Larkin's original premise that this probably wouldn't be a problem holds true. Especially since my intended use is primarily as an IF for weak signal work. With RF attenuator, overloading shouldn't be a problem. Moving the gain to the IF would add complexity of more T/R switching.

Another consideration is using newer MMICs like Mini-Circuits ERA series. Changing the MAR-3 to a ERA-2 would lower the noise figure slightly and raise gain by 2.7db, raise the 1db compression point 2.5db and increase power consumption by 5ma.

More improvement would be had by using a a low noise MESFET like NE25139 with an ERA-2. This would lower the noise figure 2db overall antenna to 2nd mixer input. A similar 2.5db increase in the 1db compression point into the first mixer and 2.7db gain increase. Power connsumption is reduced 1mA.

The two MMICs in the transmitt PA could be replaced by ERA-3 and ERA-2. This would raise the output power from 20mw to 200mw. Alternatively, ERA-3 followed by and ERA-5 would yield about 800mw output. Both of these would allow 2 meter mobile/portable operation with local repeaters or driving higher power brick amplifiers without going the double brick route. Of course this would require some tweeking of the final PA stage and power supply. Board layout could be the same for all three configurations if taken into account at design time.

Changes
I will change the coils from Toko to handwound on Amidon tuned cores. These are somewhat smaller and help for a more compact design. I looked at using 7mm Toko coils. but the Q is not as good. Also prices for the TK series recently took a sizable jump in price in the Digi-Key catalog. The Amidon cores are little over half the cost.

I will eliminate the first T/R switch. I doubt there is much use in a 20mw qrp mode. This saves a few parts, especially some handwound coils. If the PA is beefed up, low power operation might reasonable, so including the T/R switch again makes sense.

I will layout the IF strip down one half of the board and the receiver RF amps and transmitter PA on the other two quarters of the board. This would allow adding a vertical shield down the middle of the board between the IF and RF sections. Layout will take into account stacking and interconnection of the boards. Also some consideration will be taken for the HF RF/IF board (or more likely pair of boards).