<GitHub-m-labs>
[artiq] sbourdeauducq commented on issue #967: Though, the maximum range of IDELAY varies between 1.28ns and 7.68ns with PVT (not a typo, welcome to Ultrascale). So, anything below ~780Mbps has to deal with the IDELAY range being maxed out, otherwise we will stay in the above-mentioned bug hell. Maybe the OSERDES can be in 4:1 and the ISERDES in 1:8 and you switch to the other set of outputs when the IDELAY max
<hartytp>
sb0: fundamentally there is nothing wrong with that approach, and it can be the best solution
<hartytp>
depends on what your problem is, in terms of BW, current you need to measure, target noise etc
<hartytp>
the flux-gate sensors are nice because you don't have to put the current you're trying to measure onto your PCB, which can be non-trivial for currents above 100A
<hartytp>
they are also often the easiest way of getting low noise at higher currents
<hartytp>
e.g. for a 10W resistor and 100A you're limited to 0.1V max sense voltage
<hartytp>
in practice, one ends up being a factor of a few less than that for a variety of reasons (not wanting to run the resistor exactly at its max power, leaving some headroom in the design in case one finds on needs a bit more current later on, not being able to buy a resistor value that's perfectly matched to the target current)
<hartytp>
so, you're generally only measuring 10s of mV in practice
<hartytp>
in a 10Hz BW the LEM sensors have 0.05ppm RMS noise (an we found that they generally exceed their specification)
<hartytp>
even for 1kHz it's only 1ppm
<hartytp>
so, say you have a 50mV signal, 1ppm is equivalent to 50nV
<hartytp>
IIRC the best OpAmps have a 1/f noise floor (10Hz) around 100nV
<hartytp>
but, let's ignore that
<hartytp>
30nV/sqrt(1kHz) =1nV/rtHz
<hartytp>
which is about the noise floor of the best OpAmp you can buy
<hartytp>
add some Johnson/current noise and your circuit is a few times that
<hartytp>
now, if your current is lower, the resistors can be great. Or, if you are price sensitive and don't absolutely need those last few factors of two then again they can be the best solution
<hartytp>
to get the best noise out of the flux-gate sensors we generally put a few turns of wire around them
<hartytp>
they are basically zero-crossing sensors so the noise floor for the different sensors is proportional to their max current
<hartytp>
IIRC the biggest sensor is the 400A sensor, so we try to run as close to 400A-turns as possible. IIRC from doing the maths carefully , for any current above about 50A the flux-gate sensor tend to out perform a resistor