Smithsonian / ngeht-vdiftools

VDIF-related tools initially developed to facilitate testing of the ngDBE
Other
0 stars 0 forks source link

check code for EHT vdif #1

Open wumpus opened 8 months ago

wumpus commented 8 months ago

I made some changes to baseband to get it working with EHT's idiosyncratic Mark 6 recorder software, which inserts frames of the wrong length to replace missing ones. I never sent these changes upstream, and you might want to double-check if someone else fixed this already. It's most important if you want to ingest existing EHT data, so perhaps you won't be bothered by this bug.

https://github.com/wumpus/baseband

lindyblackburn commented 8 months ago

@wumpus thanks for the heads up on this. I was wondering if you recall the specific details?

fill frames of the wrong length would violate the vdif spec, and complicate any process of seeking into the frame file, deriving the sample rate, etc. I'd be pretty surprised if the Mark6 dplane has a hard coded fill frame size and cannot be set either manually or automatically to what is used by the r2dbe (8192+32 bytes).

wumpus commented 8 months ago

The "fill frame" is a hard-coded Mark5a frame. I think all of the reading code assumes all frames are the same size. That might even be required in the standard?

lindyblackburn commented 8 months ago

Got it. Yes this would violate the VDIF rules for a single vdif thread where, "All Data Frames must have the same Data Frame Header length, Data Array length, #channels, #bit/sample and Station ID." Predictable frame length could be used for all sorts of things.