04:56 <sb0> rjo, you mentioned using a HeNe laser to calibrate wavemeters. but, a good (and expensive) stabilized HeNe has a ~2MHz frequency variation. I suppose this isn't enough for laser cooling, and one has to use something like a vapor cell stabilized laser for calibration instead?

04:58 <sb0> rjo, why does Joe have a fancy wavemeter with multiple inputs and MEMS mirrors? can't one simply combine the input beams? aren't the wavelengths far enough apart that they will produce very distinct peaks?

05:56 <sb0> rjo, irc logging looks broken so if you replied already idk if I'll see it

05:58 <GitHub103> [artiq] mntng opened issue #828: artiq.coredevice.comm_kernel.LoadError: cannot load kernel: symbol lookup error: floor https://github.com/m-labs/artiq/issues/828

08:36 <rjo> sb0: 2 mhz is more than enough. the line is 20 mhz wide.

08:37 <sb0> rjo, okay, but what is the precision of the detuning required to get doppler cooling?

08:38 <rjo> sb0: they are not that far apart. some are just GHz apart. apart from the fizeau algos having problems apparently, you loose all dynamic range from exposure time.

08:38 <rjo> sb0: anything from -2 Gamma to -1/4 Gamma will usually work.

08:38 <sb0> what are the problems with the traveling michelson interferometer devices? they sound cheaper and simpler. slow speed? noise? vibrations?

08:38 <rjo> see the papers.

08:39 <rjo> slow, imprecise, large, have moving parts, they rust, etc

08:39 <sb0> imprecise?

08:40 <rjo> just estimate the number of fringes you need to count.

09:14 <GitHub107> [sinara] gkasprow pushed 1 new commit to master: https://github.com/m-labs/sinara/commit/d2f1895625cd08757701e205b856d0c564c4beb7

09:14 <GitHub107> sinara/master d2f1895 Greg: initial PCB layout. Several bugs fixed in schematics

09:56 <GitHub181> [sinara] gkasprow pushed 1 new commit to master: https://github.com/m-labs/sinara/commit/45ecc040591789d6885647d2eb59f73e9baf09c1

09:56 <GitHub181> sinara/master 45ecc04 Greg: Implemented most of @hartytp issues

12:30 <whitequark> rjo, sb0: irc logging is fixed

12:31 <whitequark> rjo: there's a VNC console and such. but logging in to it required an email OTP code.

12:31 <whitequark> and email was hosted on the same VM. (not anymore.)

12:32 <whitequark> the VM provider added that OTP code without announcing. very annoying.

13:56 <sb0> whitequark, thanks

13:56 <sb0> whitequark, did you power up a sayma?

14:08 <cr1901_modern> whitequark: In smoltcp, I noticed that before sending a TCP packet down a wire, you copy it from a "high-level" repr into an array (which is good! Avoids the need for #[packed]). Is llvm smart enough to avoid the copy from "high-level" repr to "what's actually sent down the wire"?

14:20 <sb0> rjo, don't you get better resolution by doing a fourier transform instead of a dumb count?

14:36 <whitequark> sb0: no. got the wrong kind of power connector. (6-pin vs 4-pin)

14:37 <sb0> whitequark, there were pictures of sayma online.

14:37 <sb0> whitequark, you can cut the 6-pin one and rearrange the pins

14:39 <whitequark> or... just rearrange

14:39 <whitequark> do you have any small screwdriver?

14:39 <whitequark> I got the 6-pin ones because I was under impression that they're keyed the same way as 4-pin ones

14:39 <whitequark> but they aren't

14:39 <sb0> maybe on one of the tables near the rga

15:02 <whitequark> sb0: rjo: powered up a sayma

15:02 <whitequark> anything else i should do in the lab rn/

15:02 <whitequark> ?

18:29 <whitequark> rjo: ok, I see the requirement about processing all queued segments before sending ACKs

19:34 <whitequark> I can try to do something about it, sure

19:34 <whitequark> let's see if that helps