Re: Off Topic-NUC Computer


Dale Ghent
 

On Jul 14, 2021, at 16:21, Christopher Erickson <christopher.k.erickson@...> wrote:

For a long time I have been a firm believer in putting a NUC, power dist. and a 50u multimode fiber transceiver (if a permanent installation) at the balance point of the OTA. Run 13.7V power, fiber and RS-232 down from the NUC. RS-232 to the mount. FTDI USB adapter at the NUC. Fiber transceiver next to the NUC. All short, high-grade USB cables. HDMI spoofing adapter plugged into NUC. Maybe a high-grade USB3 hub, if needed. Don't mix USB2 and 3 devices on same hub, if possible. Main cam connected directly to NUC.

1. NUC means I can use ASCOM, APCC, Horizons, SharpCap, PHD2, MaximDL, Nebulosity, PixInsight, etc. Maximum ASCOM devices and drivers supported.

2. Minimal, tough cables coming down from OTA.

3. Minimal lengths and movement on fragile, temperamental USB cables.

4. RS-232 cable going down to the mount for the absolute most stable and reliable connection possible.

5. 50u multimode fiber communications in and to the observatory for maximum protection from ESD, GPR, RFI, EMP, etc.
I'm with you on 1-4 but this last one is a bit over the top, I have to say.

I'm having a chuckle picturing it. If an EMP event were to happen that was of a proximity and magnitude to affect your network comms (and, mysteriously, only your network comms and not anything else), then you would also be running for the closest iodine stash, or be too distracted with other concerns to care about the integrity of your NUC's network connection.

I've worked in the data center environment all my life, with every kind of network and data interconnect up to 100Gb ethernet and leased lines, and stuff is not as fragile as you make it out to be. We're not even talking about mil-spec stuff. We even eschew the use of fibre SFPs on short intra-rack wiring because why have something that can break on its own when DACs have no moving parts and are more dumb, and are thus far less likely to go bad on their own. These places are also RFI boom boxes and every case of incrementing bit error rates, at least in my recollection, have been due to failing cable or port. Network and storage cabling failures are almost always mechanical in nature, like if someone shuts a cabinet door on a cable that somehow came loose and flopped outside of the cabinet a few unnoticed inches, which happens a lot.

Now if you ran fiber because the NUC and the remote network gear are on two different circuits (such as between buildings) then yeah, fiber makes sense in that case for ground isolation. But even still the optical connection is probably going to connect two switches and you'd do the home run from switch to NUC with twisted pair.

Join main@ap-gto.groups.io to automatically receive all group messages.