wide range dual freq synth .. with simple cascaded frequency doublers

[colin]

Hi, Im looking at generating RF and LO signals 450khz apart at various
frequencies from 100mhz -> ~2ghz, wich proved too tricky doing it with vcos
at the top end, even at 100mhz it seems I need oudles of
seperation/screening/isolation and decoupling etc so Im thinking of using a
dual DDS 400msps, and suitable frequency multiplication.

With cascaded frequency doublers to get the widest range without having to
do so much band switching, question is wich is best frequency doubler, a
gilbert cell is comonly used but i'd need 5 per channel, idealy it should
have a gain of >=1 so it can be cascaded easily, the best solution I have
come up with so far is to use a long tailed pair and a diode from each
collector providing rectification going to the next stage.

I could use varactors on each stage and drive them from an agc to give a
tracking filter.

I doubt if this is new but is there any neater solutions ?

Colin =^.^=

 

[Mac]

Hi, Im looking at generating RF and LO signals 450khz apart at various
frequencies from 100mhz -> ~2ghz, wich proved too tricky doing it with vcos
at the top end, even at 100mhz it seems I need oudles of
seperation/screening/isolation and decoupling etc so Im thinking of using a
dual DDS 400msps, and suitable frequency multiplication.

With cascaded frequency doublers to get the widest range without having to
do so much band switching, question is wich is best frequency doubler, a
gilbert cell is comonly used but i'd need 5 per channel, idealy it should
have a gain of >=1 so it can be cascaded easily, the best solution I have
come up with so far is to use a long tailed pair and a diode from each
collector providing rectification going to the next stage.

I could use varactors on each stage and drive them from an agc to give a
tracking filter.

I doubt if this is new but is there any neater solutions ?

Colin =^.^=

I don't know how cheap you are trying to be, but you might try using a
frequency doubler from mini-circuits, for example.

Something to consider is that you have less than 200 MHz of bandwidth
coming out of the DDS, and you want to get to almost 2 GHz of bandwidth,
so you are going to need to multiply by at least 10.

In practice, I think you will have to go more than 10x on the frequency.
Actually, if you stick only to doublers, you will have to go to 16x, I
guess. ;-)

--Mac

 

[colin]

I don't know how cheap you are trying to be, but you might try using a
frequency doubler from mini-circuits, for example.

Something to consider is that you have less than 200 MHz of bandwidth
coming out of the DDS, and you want to get to almost 2 GHz of bandwidth,
so you are going to need to multiply by at least 10.

In practice, I think you will have to go more than 10x on the frequency.
Actually, if you stick only to doublers, you will have to go to 16x, I
guess. ;-)

Like I said I think il need 5 doublers if im going to avoid spurs from the
dds wich is a 500msps dual device so should be very good at 60mhz (x32), I
also need 2 multiplier chains wich is why im looking for small cheap
circuits !

I think the mini circuit ones are lossy so I would need an amplifier
inbetween each one.

At the moment im trying to optimise in the simulator a doubler using a
bfe520 dual transistor differential pair, the first doubler works quite well
completly rejecting the input frequency, the second stage doesnt work so
well maybe its the harmonics or dc offset, it needs ac coupling but this
makes it awkward to simulate needing a long run time to settle to the right
dc value. interestingly I found the harmonics can be reduced quite a lot by
using 2 diodes in series.

The output from the DDS is balanced so should be able to double the
frequency here with just 2 diodes.

thanks
Colin =^.^=