[Skimmertalk] red pitaya noisy

[W3OA]

Hi Dave -

This looks like great progress in using the Red Pitaya.  Thank you.

But I'm concerned because of the following FAQ on 
http://redpitaya.com/faq-page/

Do you think it would be advisable to provide some external protection 
circuit on the antenna input?

73 - Dick, W3OA




On 7/31/2016 11:28 AM, David Robbins wrote:
>
> Previously for those not included in my discussion with pete:  “Got it 
> running now with the hermes interface dll, but I am plugged in to the 
> router.    Running it on the splitter with a qs1r on the other port, 
> both running cw skimmer server 1.50.  the rp receive seems much less 
> sensitive.  On 20m the qs1r has 340 decoders running, the rp only has 
> 45 and is generating a lot less spots.  The ones that both of them 
> spot about the same time the snr is 2-6db lower on the rp.   More 
> later, going to cut some grass or something while the wx is nice.”
>
> Then when I dug into it some more:  “While I would like someone with 
> professional instrumentation to repeat this I think you can see a 
> clear difference in the narrow frequency displays of hdsdr below.  The 
> one on the left is the qs1r, the right is the rp.  It looks like the 
> rp is about 5db noisier.  I zoomed in on the signal peaks to compare 
> them closer and the amplitudes are almost identical as close as I can 
> tell, so the signal strength calibration is about the same.
>
> cid:image001.png at 01D1E020.4FD7C0D0
>
> There are of course many things that could contribute to this.
>
> 1. Power supplies are different, I am using the supplied rp supply for 
> it but an after market supply for the qs1r.
>
> 2. Front end or adc noise
>
> 3. The rp is not in an enclosure like the qs1r maybe some local noise 
> pickup
>
> 4. The sample rate conversion could contribute noise
>
> p.s. that was a xg2 at 1uv on 40m going into a dxengineering splitter 
> feeding each of the receivers.  The rp is using the hermes interface 
> dll to cwsl_tee to hdsdr and skimmer server.  the qs1r is the same 
> software setup except using the qs1r interface dll.
>
> ”
>
> Today:
>
> Hmmmm, nice rainy morning so played some more with the rp.  I think I 
> found the problem.  on the input lines there are two sets of jumpers… 
> I didn’t quickly find a final schematic, apparently they are keeping 
> those private, but I did find a ‘development’ schematic, excerpt shown 
> here:
>
> So jp1 sets the input to one of two attenuators, and jp2 sets the 
> output.  I thought I would try the HV setting, which attenuated my 1uv 
> to below the noise level.  But I noticed when I pulled the jp2 jumper 
> off the output noise went well below the display level I had set on 
> the hdsdr.  So after playing around injecting my signal to pin 5 of 
> jp2 through capacitors and jumpers and generally making a lot more 
> noise, I just put one of their jumpers from pin 2 to pin 5 which 
> bypasses the two input attenuators and voila, now the two cw skimmers 
> are generating almost identical spots.  In fact some of them from the 
> rp are 1db higher snr, and a few are a bit lower, but that is bound to 
> happen just from the skimmer software timing of when it generates the 
> snr value, one may sample before or after the other and get a noise 
> spike or not.
>
> So my guess is that somehow the input attenuator is coupling noise 
> from the board ground into the input signal going to the amp/filter 
> block.  But it would take someone who knows more about the low level 
> design of the amp to figure it out.  I guess I would have expected to 
> see the inputs isolated from the digital board ground and going into a 
> differential amp, but I can’t tell that from the schematic I found. 
>   it could also just be that the attenuation of the input before the 
> amp dropped the signal closer to the amp noise floor and all I’m doing 
> is boosting it a bit, but whatever the mechanism it seems to have 
> helped a lot.
>
> David Robbins K1TTT
> e-mail: mailto:k1ttt at arrl.net
> web: http://wiki.k1ttt.net <http://wiki.k1ttt.net/>
> AR-Cluster node: 145.69MHz or telnet://k1ttt.net:7373
>