Hi,
On 19.09.23 09:24, ouzan...@hotmail.com wrote:
Dear Marcus,
Thank you for your reply. I think your RF engineers have an aim to have the pads on the
PCB for bypassing the switches. In other case, why do you have these pads if they
haven't tested if they are working ?
When you build a complex device, you leave options to make the board work differently, for
example for investigating faults in manufacturing later on. Having that trace there costs
exactly nothing. Its existence doesn't mean anyone ever intended it to be used by customers!
It really doesn't matter – we have no knowledge of the characteristics of that path,
because it's not something qualified for usage. So, I can't give you measurements we don't
have.
Is it possible if you could ask your RF engineers which component values we need to
locate on these pads ?
No, sorry. That board was designed some ten years ago, most of the folks that were around
back in the day have moved on to other departments, and what do you expect to come back as
answer?
They couldn't qualify with their board, using their components and their soldering
techniques and qualities what you would achieve on your board. So, this is really a moot
question to ask!
But: I mean, these are just dummy connection parts, so you'd start with a 0Ω and then look
at Smith charts for long enough until you compensated any mismatch introduced by the small
pair of impedance discontinuities with a Z_0 trace of defined length in between; it will
depend on the frequency.
50-54 dB TX/RX isolation is very low as a dynamic range of many applications.
Well, that's not really true; very few integrated transceivers achieve more (for the same
reasons that the USRP doesn't achieve more: you have a limited number of layers,
affordable board substrate, and as Marcus says, integrated transceivers within the same chip).
You're barking up the wrong tree if you need more: that's really not something that
devices based on chips that do both the transmit and receive side in the same IC package
can typically achieve at reasonable cost.
There's plenty of examples of devices with isolations no better than that still working in
full duplex (for example, I know of a Munich-designed set of Wifi chipsets that do
self-interference cancellation; you can be relatively sure that a 20€ wifi card designed
by an OEM has no better TX/RX isolation that that). It's in the end a question of quality
of your algorithms, and whether your expectations match the hardware you bought.
Sure, if you need to build a high-PRF aircraft radar, this might be too little isolation
without external amplification; but maybe the B2xx just isn't the device for that kind of
application :)
For our application, this is our main problem. That is why we want to try to bypass and
see if the isolation of the card increases.
Really: nobody's stopping you from trying out, but in all honesty, unlikely that you gain
anything; that's why I informed you of that. The fact alone that you asked about the
capacitors is honestly also not a great sign when it comes to your team's RF hardware
testing capabilities. But you already use the word "try": so you will also incur the onus
of having to measure the output. Nothing we could deliver could change that, so: best of luck!
There is no data about the TX/RX isolation of AD9364 in its datasheet. There is some
information in the forums indicating that the TX1/RX1 isolation is measured as 65 dB.
… which is exactly what the other Marcus (Marcus Leech) and I noted. But, as said, the
AD9361 itself isn't the only thing with crosstalk in a complex system, and the switch
combination was chosen to fit the overall system, not just that component.
Best,
Marcus M
_______________________________________________
USRP-users mailing list -- usrp-users@lists.ettus.com
To unsubscribe send an email to usrp-users-le...@lists.ettus.com
