Design Review for a No-Compromises SatNOGS station

SatNOGS Hardware
1

Hi all,

We are designing and building a SatNOGS station that will be deployed in Alaska in a couple months. Some details are temporarily confidential like exactly where it is going, but there are some unique challenges to this location. Namely, we must mount our antenna to a mast that has an existing 500W 400MHz transmitter and a 150W 437MHz transmitter, and we must be very close (1-2m away). The reason we have to be this close is to point North towards the pole (for best visibility of polar sats), we need to be up high to clear the roofline. It’s also a very high priority to cover 400MHz from this station so we can’t just filter out the 500W transmitter.

Here’s a cropped view of the existing hardware, we’ll probably be around 2 or 3:

image
image928Ă—1280 130 KB

The TX yagis are on a 2-axis rotator and although we’re definitely close, we are outside of the direct line of fire. We will definitely be concerned about backlobes and sidelobes though. The duty cycle of the transmitters should be very low - they are transmitting to a CubeSat in LEO. Our RX antenna will be statically mounted (~45deg elevation) and pointed North for best visibility of polar satellites as previously stated.

The biggest risk in this setup is the received power seen at our antenna/LNA. I am not an expert on this type of calculation, so as a first approximation I used ChatGPT and some online calculators and came to an estimated max input seen at our RX antenna of 42dBm. If someone here thinks they can make a better approximation I would greatly appreciate it!

This station has to be very reliable - if something breaks, we’re not gonna get a chance to fix it for years potentially. We do have a budget allocated for this setup and currently have some room so if there’s a higher quality / more robust part you’d like to see here then please let me know. I would like to place a better computer there and ideally have some form of out-of-band access (PiKVM?).

Here’s a BOM for the most critical parts:
Pasternack PE51YA1015 UHF Yagi, 400-470MHz, made for harsh conditions and has a relatively wide beamwidth of ~50-70deg

Pasternack Lightning Protector

Qorvo TGL2210-SM Limiter, first line of defense for TX

MiniCircuits 400-510MHz BPF Filter out-of-band interference (expected to have lots of it)

MiniCircuits ZX60-P103LN+ LNA Wideband LNA with high max-input power

MiniCircuits Bias-T ZFBT-6GW-FT+ Bias-T to power LNA, one inside and one outside

MiniCircuits ZFLM-252-1WL-S+ Limiter, bring down levels to a survivable point for USRP

MiniCircuits 3dB attenuator Need to drop down the output level just a lil bit more in the worst case to protect USRP

USRP B200mini, max input level 0dBm

All parts will need to be ordered next week in order to assemble, test, and ship everything.

image
image1308Ă—1532 166 KB

Thanks,
Kevin

3 Likes
2

Regarding the estimation of the received power at our yagi, @azisi pointed me to this:

“A experiment with 2 x-quad antennas are done. The antennas are placed at distance of 1.7m. The PA transmits 37dBm before a cable of 5m and the antenna. The Rx antenna receives after a 2m cable -8dBm. The experiment took place in open air field and the measurement tool is a calibrated SA.”

3

Kevin,

I have a colleague that runs a SATNOGS station at a remote airfield about 3 hours from town. He found that he needed a way to reset the system if it locked up……which they do every so often. He used an internet connected mains switch to allow him to dial in and reset the power.

I was having the odd lockup on my two SATNOGS stations that a simple power reset fixed, so I adopted a similar strategy. I simply added a timer switch to the AC power, and it does a reset once a day. That seems to keep things going reliably.

Naturally you need to make sure that everything boots to full operation at power up with no human intervention.

If access is so difficult you may like to consider either of these options.

John – VK4JBE

4

Ah, I just reread your first post and see it isn’t a pi you are looking at so some of the below won’t be directly applicable to your situation.

I haven’t implemented it myself yet but you might consider using the raspberry’s watchdog feature.

One thing to consider with a nightly hard power cycle is you may end up corrupting the SD card.
If you are going to implement something like that then I would personally consider the following approach, if you want to power cycle at 23:00 then.
22:55 shutdown the pi via a cron entry. (Ensures the pi is shutdown cleanly thus reducing the chances of a corrupt SD card)
23:00 trigger the external power cycle, this will hard reboot the pi and it should come up cleanly again.

Again things I have not implemented (on a raspberry pi) yet.
Consider RAID1 for storage something like here.

Would you also consider instead of the 3dB attenuator before the USRP putting in a splitter and bringing the signal out to a RTLSDR and a second pi or similar, this was if something happened to the primary system you may have the option to switch to a secondary system (maybe at a lower performance) until you can get on site to restore the original system.
Of course this wouldn’t protect you from any failures upstream from this point.

You may also be able to implement things such as having a serial link between the two boards that might provide you with some method of remotely restoring.

As I said I haven’t implemented any of these solutions on either a Pi or a droid (I have done similar in higher end enterprise equipment but not all things there would carry over)

Mark

1 Like
5

Not a bad idea. Even hooking it up to the same host would be a benefit.
Something like the Z99SC-62-S+ perhaps.

6

What you haven’t stated is how often those 400 and 437 MHz transmitters are operating.

At that kind of distance, without significant filtering (and we’re probably talking cavity notch filters for this kind of situation), all of your receive amplifiers will be pushed into distortion whenever the transmitters are active. Your limiters will protect your hardware from damage sure, but do not expect your receive system to produce useful results while the transmitters are running.

Just trying to set some expectations on what is possible here without filtering.

1 Like
7

Yes, that is what we expect. We’re not building cavity notch filters and the observations just have to live with the sporadic TX’ing.

8

Mark – EI4FNB,

I like your ideas about protecting the SD card from being corrupted during powerdown.

The cron option is very simple to do before I power down the system.

The RPi watchdog has the advantage of rebooting the system as soon as it fails as opposed to waiting until the predetermied shutdown time (which could be up to 24 hours). The only thing that may catch you out is that I think the SATNoGS software may be still running but giving corrupted results. I mostly see that as failed passes. Often the recording size does not match pass length but I am never really sure why. Perhaps you should really try and fix that rather than just reset my way out of it.

I might implement your cron option though, as that is really simple.

Regards,
John - VK4JBE

9

Hi all,

Thanks everyone for your input so far!

Yes, we do need a way to remotely reset, and I don’t want to put my faith in a SD card being reliable for a long-term installation like this. I’m looking at two options right now:

  • Protectli VP4670
    • I have extensive firsthand experience, they are very reliable.
    • Fanless / passively cooled, all metal case is helpful for shielding
    • SATA/NVMe
    • Internal 16GB eMMC for backup boot device
    • 12V input
    • Externally accessible console, but only via a Micro-USB (questionable)
  • Supermicro IOT Superserver, like E302-12D-8C
    • MUCH higher quality hardware
    • IPMI for remote access/control, no need for the PiKVM anymore
    • ECC RAM
    • SATA/NVMe
    • 12V input
    • Twice the price of the Protectli… doable but makes the budget very tight.

Unfortunately, neither option makes it easy to do a RAID1 of storage devices. Another “nice to have” would be a large mechanical hard drive, like 10TB+ for archival of IQ recordings, STRF bins, etc. I’d rather do that over SATA than USB.

As for reset/recovery, If we don’t use a device with IPMI, I’m planning to put a PiKVM in the box. This will let us remotely reinstall the OS (via an emulated USB device), talk to the console port, trigger a power reset, or interact with the system with video/mouse/keyboard. I also believe that the organization hosting this device has remote controlled outlets for a last-resort recovery method, if not then we might try to add our own remote-controlled relay box, like this DINRelay IV (also a device I have firsthand experience with).

A couple other notes - we’re going to fit this entire indoor system into a 2U or 3U rackmount enclosure, and will be using a very high quality AC-DC power supply and filtered power entry modules.

@ei4fnb Your splitter idea is excellent. Even with the above setup, there’s still a risk something could go wrong and it would be helpful to have a (cheaper) system that just runs a basic SatNOGS station. Perhaps Pi4 and AirSpy R2? That way it could still benefit from the Leo Bodnar GPSDO.

@vk5qi You might have missed the edit but the duty cycle on the 400/437 transmitters is expected to be very low. They are backup TT&C uplinks for a couple LEO CubeSats, so we should expect them to operate a few times a day for ~10 min in the worst case. You’re also absolutely right that we should not expect anything useful during a TX - as long as we survive, that’s good enough for me.

1 Like
10

For high latitudes (e.g. polar circle), the apex of sun-synchronous orbits at, say, 500km, is still only at 10 deg elevation towards the North. A better approach would be to point North West or North East at say, 45 deg. Pointing away from zenith is always a better approach as you sample more volume and the satellites have lower angular velocities.

The attached plot shows the passes for Connecta T1.1 (~520km @ 97.6 deg) for 24h for a location at the polar circle.

Screenshot_2023-07-29_00-04-39
Screenshot_2023-07-29_00-04-39871Ă—699 41.1 KB

1 Like
11

Thanks for the analysis @cgbsat!

After talking it over with @SA2KNG, I’ve made some more decisions. Although the Supermicro machine I previously mentioned has IPMI and probably slightly better reliability than average, the Protectli VP4670’s cost allows for more flexibility elsewhere in the design and we will still have ample ability to remotely control/reset/reinstall it.

I’ve also added the DINRelay to the BOM since it will give us a degree of control over the individual components, versus the brute force approach of hard-resetting the entire box with the rack PDU.

Another consideration here is a different PGA-103+ based LNA that has a builtin bias-T. I rather like the appearance of this one by SV1AFN, and it has the added benefit of relays. We’ll know which satellites are going to be TX’d to from the yagis right above us, so we could cut power to the bias-T and open the relays when these satellites pass overhead to have an extra layer of assurance that our RF equipment won’t be damaged. I’m not sure if that’s worth the added risk of relays in a very cold environment, thoughts?

Here’s what the design currently looks like:

alaskaSatNOGSStation
alaskaSatNOGSStation4050Ă—6510 677 KB

2 Likes
12

I think such a bypass relay should in this case be modified to DC short while powered off, not passing the initial limited signal to the SDR’s. This also means that the quorvo limiter doesn’t have to dissipate the incoming power either.

13

@Kevin have you check this, https://digitalcommons.usu.edu/cgi/viewcontent.cgi?article=3368&context=smallsat. I think that @danwhite has suggested to use something like this, Transmitting FM, AM, SSB, SSTV and FSQ with just a Raspberry Pi as a self-test of the station.

1 Like
14

Just a quick update - we’re probably going to move this station to a different location that has better access, doesn’t have the 400/437MHz transmitters, and has other benefits. For the moment this project is on-hold while we figure out these plans.

2 Likes