[hpsdr] Cyclops Front end questions

[John Miles]

> OK - I have a weekend to start on this
> Some questions for the gurus:
> 
> (1) Input circuitry
> I note the SSA at 
> http://www.scottyspectrumanalyzer.com/msaslim.html#Mixer_1_using
> 
> just uses DC connection to the input mixer
> Should we have attenuators here?
> What about AC vs DC coupling/limiting/matching pad etc?

IMHO anything like this should ideally be kept off the main board (see below).

> (2) 1st Mixer
> We seem to have settled on the ADE-11X +7dBm device - any better ideas?
> What about associated circuitry?

I would use at least a level-10 mixer (see below).

> (3) 1st LO
> Several candidates for the VCO 
> 
> Minicircuits ROS - 2150VW as has been used in SSA
> +5V supply +4dBm  out Phase noise at 10KHz -96 dBc/Hz Range 975 - 2150 MHz
> Cost $29.95 in 5
> Stocked Minicircuits ?anywhere else
> 
> Crystek CVCO55CW - 1000 - 2000 
> +10V supply +5.5dBm  out Phase noise at 10KHz -100 dBc/Hz Range 
> 1000 - 2000 MHz
> Cost $30.00
> Stocked Mouser - not Digikey
> 
> Synergy Microwave "YIG replacement" DCYS100200-12
> +12V supply +4dBm  out Phase noise at 10KHz -105 dBc/Hz Range 
> 1000 - 2000 MHz
> Cost ? - anyone know?
> Stocked ?

Lacking a small-quantity distributor for the Synergy VCO, I'd favor the Crystek part.  It is important to use the standard CK605 footprint so you're not locked into one VCO vendor or another.
 
> I guess we will have to use a buffer to bring the outputs up to +7dBm
> ?Which

I'd use a MiniCircuits GALI-4 or similar MMIC for this.  The goal should be to hit the mixer with at least +10 to +13 dBm, so you can use a level-10 mixer.  The dynamic-range difference over a level-7 mixer is probably worth it.

> > (4) PLL 
> All the previous circuits have used the National LMX2326
> Is there a better alternative?

Yes, the ADF4112 is a much better choice for an octave-band synthesizer.  It is almost 100% backwards-compatible with the LMX2326, and it will let you vary the charge-pump gain to accommodate the large difference in N factors from one end of the band to the other.

> I note that the SSA uses a IC regulator with minimal filtering 
> after it to supply 
> the reference for the loop filter opamp - these have a fair bit 
> of low frequency 
> noise - we used a zener/pass transistor for our supplies in the 
> microwave synth a 
> few years ago. 
> What sort of noise floor can we expect from the loop filter?

With an LT1677 opamp, the out-of-band noise/spur performance can be quite good.  It represents a decent compromise between noise and input bias current (see graph at http://www.ke5fx.com/ssa/lo1.html ).  

Whatever you use, make sure it conforms to the standard 8-pin opamp footprint, so you aren't married to it.  Opamps are improved all the time, and while I like the LT1677 a lot, it's probably been eclipsed by newer parts.  

Also, regardless of your loop filter topology or choice of opamp, be sure to leave an RC pole between the opamp and VCO, as this really helps with the broadband floor.  

Supply-wise, there are lower-noise regulator chips than the 78xx parts, but I haven't adequately investigated the effect of power-bus noise and can't say what's necessary versus what's overkill.  The PLL will be able to reject supply noise to some extent.

> We would aim to have 2MHz steps so the reference shouldnt be an issue.
> The loop filter needs a higher voltage supply than Atlas 
> currently provides - could 
> we put say 24V on the Atlas bus? A switch mode regulator wouldnt 
> be great on the 
> Cyclops board

I don't think a small, low-current boost converter would be the end of the world, if laid out and filtered properly.  But it's best left off the main RF PCB.

> (5) Roofing filter
> I am proposing a Toko 4 pole 6DFC-1030C-10 dielectric filter as 
> in the Cyclops wiki.
> Would this be sufficient?
> 
> (6) What about all the losses through this setup - do we need some gain?
> will be in the order of 13-15dB so far with the same again in the 
> second mixer/SAW 
> filter.

You probably do not want any gain (beyond a minimal post-mixer amp) until after the 2nd mixer.  This is in keeping with commercial design practice for SAs, and can help with IMD performance.  Remember that you don't care about front-end noise figure in an SA; 30 to 40 dB is par for commercial designs.

Without knowing many details about the design you have in mind, I'd suggest a direct connection from the first mixer IF port to the input terminal, followed by a 6- to 10-dB attenuator at the RF port (which is really the IF output), followed by your roofing filter, then a low-gain, high-dynamic range amplifier to make up for the post-mixer losses (GALI-4 is workable here), then the second mixer.  I've had decent results with that basic receiver architecture ( e.g., http://www.ke5fx.com/equinox/equinox.html ).

Any desired input attenuators, preamps, preselectors, etc. are best left for a separate PCB, as many applications won't need those stages, or will want to handle them separately.

> (7) Where are all the signals for the tracking Generator derived?
> I am still a bit vague on this.

That's a complicated matter if you are planning on implementing a DSP back end.  I would install jacks on the board to make LO1 and LO2 available, and leave the rest of the tracking-generator details for later consideration.

-- john, KE5FX
 


 1210391233.0