kuldeep_ has quit [Remote host closed the connection]
kuldeep_ has joined #m-labs
hedgeberg is now known as hedgeberg|away
<sb0>
the si5324 is another pain in the ass
<sb0>
takes forever to lock (up to 10 seconds), switching clocks by writing the register does not immediately clear the lock indicator
<sb0>
how is it even possible to know that the clock switch has completed and it's locked to the new clock? I cannot think of a way to do it that doesn't have races or other potential intermittent problems
<sb0>
other than waiting the full 10 seconds and checking the lock indicator...
<sb0>
sigh, proprietary chips
<cr1901_modern>
The older I get, the less patience I have for proprietary *anything*
<cr1901_modern>
a PLL taking 10 seconds to lock up sounds very odd to me, btw
<cr1901_modern>
is it more than one chip exhibiting that?
<sb0>
the loop bandwidth is very low, so there's an excuse for that
<cr1901_modern>
Okay, cool (and I assume you need the loop bandwidth that low for your application to prevent jitter)
<sb0>
i'll have to see if we can increase the loop bandwidth
<sb0>
the 5324 has some "fast lock" features that temporarily increases the loop bandwidth while it's locking, but this does not seem to work very well
<cr1901_modern>
"switching clocks by writing the register does not immediately clear the lock indicator" <-- prob is so slow to detect changes that it actually takes that long to lose lock :P?
<sb0>
something like that, either way it's annoying
<cr1901_modern>
Now I want to know how the Loss-of-Lock logic works. If I'm reading this app note correctly, when the fast-lock feature is used, the LOL alarm doesn't actually reflect the true lock time.
<cr1901_modern>
The "naive" way of measuring whether a loop has lock or not is to monitor the derivative of the phase difference. If the si5324 were doing that, the LOL flag should've been unset in ~1 second if going by figure 5
<cr1901_modern>
Oh. I see. The LOL time is basically a crapshoot, according to page 5.
<rjo>
sb0: ok. you can generally terminate it if i don't reply within five minutes.
rohitksingh has quit [Ping timeout: 240 seconds]
<GitHub>
[artiq] sbourdeauducq commented on issue #562: @cjbe Basic DRTIO with outputs only works at 1.25Gbps with commit 9501d37378f2998bc250edd71de55526d2ee624c. You can have a look if you want. https://github.com/m-labs/artiq/issues/562#issuecomment-280840408
kuldeep_ has quit [Remote host closed the connection]
kuldeep_ has joined #m-labs
rohitksingh has joined #m-labs
rohitksingh has quit [Read error: Connection reset by peer]