Your low Signal, low C/N and bursts of coninuity errors all direct to your frontend.
You should test the receiver with a *direct* connection to the antenna. Maybe
szap-s2 or driver module stv090x has a problem with your Diseqc switching.
Could you also check that you load the right driver module. Maybe something
went wrong when compiling and installing the TBS .../media/.. drivers?
# modinfo stv090x
shold show a first CrazyCat added module parameter
parm: ts_nosync:TS FIFO Minimum latence mode (default:on) (int)
I have seen a bit higher CN than 9.6 in the past (with my 80cm dish), maybe I need to recheck the pointing after "Sabine", but it is still good enough for Basic Service and that's all I want to receive at the moment. And it works perfectly with Win10 and exactly the same HW. Btw. in BDADataEx I have set the modcod to 8PSK3/5 for BS only. I'm not sure how this can be done with szap-s2.
This is the output of modinfo:
description: STV090x Multi-Std Broadcast frontend
author: Manu Abraham
vermagic: 5.4.0-0.bpo.2-amd64 SMP mod_unload modversions
parm: ts_nosync:TS FIFO Minimum latence mode (default:on) (int)
parm: verbose:Set Verbosity level (int)
description: TurboSight TBS 5925 DVB-S2 driver
author: Konstantin Dimitrov <kosio.dimitrov@...>
vermagic: 5.4.0-0.bpo.2-amd64 SMP mod_unload modversions
parm: debug:set debugging level (1=info 2=xfer (or-able)). (int)
parm: adapter_nr:DVB adapter numbers (array of short)
Device TurboSight TBS 5925 DVB-S/S2 (/dev/dvb/adapter0/frontend0) capabilities:
DVB API Version 5.11, Current v5 delivery system: DVBS2
Supported delivery systems:
Frequency range for the current standard:
From: 950 MHz
To: 2,15 GHz
Symbol rate ranges for the current standard:
From: 1,00 MBauds
To: 45,0 MBauds
SEC: set voltage to OFF
FREQUENCY = 1513000
INVERSION = AUTO
SYMBOL_RATE = 33000000
INNER_FEC = AUTO
MODULATION = QPSK
PILOT = AUTO
ROLLOFF = 35
POLARIZATION = OFF
STREAM_ID = 4294967295
DELIVERY_SYSTEM = DVBS2
We get closer to the point:
1) SNR 9.6 dB is low even for an 80cm dish. I'am quite sure you can do better.
2) Link Margins (LM) are (3-4 dB is recommended!)
BAS 8PSK 3/5 : LM = SNR or C/N - 5.9 dB
HVS-1 16APSK 2/3 : LM = SNR or C/N - 9.3 dB
3) THE REST IS HISTORY:
When we had to change to DVB-S2 EUMETSAT promised that ACM/VCM receivers would
solve all problems. People with smaller antennas would still receive BAS while
people with big antennas could receive BAS + HVS-1 (only T1 in use at that time).
EUMETSAT was WRONG. As far as I remember even the rather expensive SR1 could not
handle DVB-S2 EUMETCast properly at the beginning. The problem was (and still is!)
that as soon as HVS-1 falls to low link margins this affects BAS as well. The SR1
has the very same STV0900 Advanced kind of demodulator that is soldered in your
TBS-5925. This ST chip allows for MODCOD filtering under DVB-S2. As far as I
remember MODCOD filtering came with a later firmware update for the SR1. And it
seems that you can activate it under WINDOWS with CrazyCats BDADataEx on your
TBS-5925 (I never had a Windows EUMETCast setup here).
4) szap-s2 floats around the net in different versions. I use a CrazyCat version
which originally stemms from Igor M. Liplianin. It needs the source and some
googling to find out how to transmit all kind of stuff to the driver. My version
lets you also set stream IDs while the EUMETSAT distibuted version has no such
parameter. AFAIK the stream ID should be 1 not what your dvb-fe-tool returns.
# szap-s2 -q usually shows you all possible parameters you can set with command
switches. There is a hidden compressed format directly with the channel entries.
I had to write my own szap-s2 HOWTO to learn how to send all these parameters.
But I finally found out that some Linux drivers ignore the majority of those
parameters and find out themselves what they think is right. You can actually
ask for modulation+fec (MODCOD) and rolloff but it might have no effect at all.
5) The TBS-6925 did not work to begin with while the SR1 with the same chips worked.
It was again CrazyCat that found under Windows a solution with the FIFO parameter.
Konstantin dropped me a Mail and I had the first working PCI card by simply setting
1 single bit in the plenthora of ST0900 registers. That's why I know stv090x.c a bit.
6) At about the same time the SR1 received MODCOD filtering I added my own homebrew
MODCOD filtering to the TBS-6925 driver. Christian tried it out on a TBS-5925
later as well. If you reduce an SR1 or TBS-59/6925 on BAS only it means that you
payed a lot of money for a ACM/VCM capable device and use it in a way a 30 Euro
SkyStar 2 eXpress HD from eBay can work as well. EUMETSAT said it would never work
but praise CrazyCat who made it happen again. Also praise Georgios Potiriadis
at EUMETSAT who finally wrote a couple of HOWTOs for the use of PCI cards in
EUMETCast receivers even if some of these manuals are slightly outdated now.
7) When the TBS-6983 appeared (I still have two hand made prototypes here!) the SR1
was outperformed. The new ST chip allowed to receive at the same time T1 + T2 while
this asked for a second SR1. There was quite a bit of FUD on the forum concerning
this promise of the new PCIe card but it all came true. Arne van Belle demonstrated
with some help of me that he could receive 2 times (BAS + HVS-1) on a TBS-6983 and
keep up with all this data on a moderate powered PC even before T2 was in the air.
8) I'am no software ingeneer and I wrote my homebrew MODCOD filter in 2014 simply by
looking at Manu Abrahams and Konstantin Dimitrovs code of the stv090x driver. Technical
ST documentation asks for receiving and signing an NDA. I still have some TBS-6925 and
SkyStar 2 eXpress HD cards here and I guess my patch still works in the stv090x distributed
by TBS today. But this is now a 2020 in-tree version adapted by CrazyCat that is used by
TBS-59/6925 (CHIP STV0900 Advanced) but also works with a SkyStar 2 eXpress HD (CHIP STV0903).
BOTTOM LINE AND WHERE TO GO FROM HERE:
GNU/Linux drivers still do not allow for MODCOD filtering. With the C/N you showed
us you have almost 0 dB link margin for HVS-1. If you say it works fine under Windows
with the same Signal and C/N (SNR) it seems you can now really set MODCOD filters with
BDADataEx. Please confirm that by testing EUMETCast Windows reception with all MODCODs.
David Taylor and CrazyCat should certainly be able to shed some light on this point.
A) You should have about 12dB SNR for T1 if you cannot filter MODCODs.
I had an 80cm dish and I'am afraid your antenna is rather misspointed.
B) If we are sure you have a HVS-1 --> BS interaction reception problem I
could add my 2014 patch (that I just found somewhere on one of my receivers)
to the 2020 distributed version of stv090x.c and you can give it a try.
C) I cite from your first post in this thread: "I was quite confident that this would
work in a short amount of time, given my 35 years of experience working as an engineer
in the IT industry and beeing a Unix native...". If the patch from a simple Physicist
and Teacher works for you, then you could give CrazyCat and TBS a hand to integrate it
properly in stv090x.c but also in the very similar stv091x.c code for TBS-6983/03/08.
Hope this helps
P.S. My solution for the nasty HVS-1 --> BAS interaction was a Gibertini 1.25m dish
Ernst you are my hero, like my physics teacher 40 years ago in my Gymnasium. I switched on Modcode All in BDADataEx and immediately the error rate went up and looked almost as bad as with the linux setup. Thanks to the warm and beautiful weather today, I went outside, climbed on my roof and fixed not only my ADSB antenna (was lying almost horizontally because of the storm) but also my sat dish. With a slight movement to the east, the signal level was up to 10.6 dB. Now I'm back at solid zeros for the missing packet count with modcode All. Will go back to my NAS system soon.
I had a look at the stv090x.c code and it has provisions for setting and filtering MODCODs and PIDs. I would definitely be interested in your patch and the other szap-s2 version.
Thank you very much again.
If you are in Germany even a C/N = 12dB should be possible with your 80cm dish.
Here is my stv090x.c driver with 3 snipplets patched in for easy MODCOD setting.
I tested it in many ways with one of my old TBS-6925 last night and this afternoon.
I just tested with dvb-fe-tool that my version of szap-s2 correctly sets
STREAM_ID = 1
ROLLOFF = 20
So this might be a last problem. I never understood, why EUMETSAT
distributes an szap-s2 that apparently cannot set the STREAM_ID.
MODULATION = QPSK
is returned by dvb-fe-tool if you are in VCM (as far as I remember).
I'll make some more tests and possibly send you my szap-s2 source
tomorrow. I will have to UPDATE my own szap-HOWTO somewhat too.
Let's take the shortcut. Here is the Igor M. Liplianin, CrazyCat enhanzed szap-s2 version you better use.
All you have to do is add in channel.conf behind :hS1 more stuff: It should then read :hS1P1O20
P1 sets the STREAM_ID = 1 and O20 gives you ROLLOFF = 20 (AFAIK EUMETSAT talks about 5)
total success. For unknown reasons, my CN dropped to 10,0 - 10,2 since about 4 UTC and I had the errors again. Compiled and copied your versions, rebooted, run setmc and everything is perfect again.
One remark, copying the file stv090x.c to the ./media/... folder did not work, I had to copy it to ./media_build/v4l and ./media_build/linux/drivers/... to get it compiled and installed.
Btw. I have changed my BIOS settings back to SpeedStep enabled and C1,C6 and C5e enabled. The linux cpu governor runs in powersave mode. No packet losses even with all modcods if the CN is high enough (~>10,5).
Eumetcastview also runs on the NAS now and happily creates pictures. Out of curiosity, I will try SatPy/Pytroll now.
Here I have to change the driver stv090x.c in ./media as described. The point probably is
that this only works if the the patched driver copy is newer than the replaced one. This
is because make relies on time stamps. So if you downloaded the media tree from TBS again
your files could have been newer than my patched copy. Just checked this here again:
My downloaded TBS tree dates from Feb 12 10:47
This tree has been compiled and installed before (including my patches too)
I make sure in ./media_build I have again the stv090x.c from Feb 12 10:47
My patched stv090x.c driver copied to ./media dates from Feb 14 19:26
I go to ./media_build and do a make that compiles just one module stv090x.o
Now in my ./media_build I have a patched todays copy of stv090x.c Feb 16 13:45
Every thing works as discribed in the README_FIRST.txt if the patched driver is newer
than the files in the TBS tree. I guess you downloaded the GIT tree again and it was
newer than my patched driver? You can open the patched source with mc and save it again
to get the newest file on the planet. Maybe I should explain that in my README_FIRST.txt.
Nice to hear your GNU/Linux receiver is now up and running fine. I have a problem here
with my Debian 10 and latest 64bit tellicast client. Some 10 minutes after I start tellicast
there is a very small but slowly increasing use of swap space. Is there a memory leak?
How does your NAS with a far newer kernel handle that? What does top (memory) show?
P.S. SatPy/Pytroll is powerful and extreme fun. But this is not an overnight thing either ...
I can observe an increase in memory usage. Right now the service allocates 5.1GB of memory:
root@openmediavault:~# service tellicast-client status|grep Memory:
the main running tc
root@openmediavault:~# pmap 4679
4679: /usr/local/bin/tc-cast-client -c /etc/cast-client_bas.ini
0000000000400000 2760K r-x-- tc-cast-client
00000000007b1000 3036K rw--- tc-cast-client
0000000000aa8000 492K rw--- [ anon ]
0000000001a3f000 83204K rw--- [ anon ]
00007f661828e000 664K rw--- [ anon ]
00007f6618334000 12K r---- libnss_files-2.28.so
00007f6618337000 28K r-x-- libnss_files-2.28.so
00007f661833e000 8K r---- libnss_files-2.28.so
00007f6618340000 4K ----- libnss_files-2.28.so
00007f6618341000 4K r---- libnss_files-2.28.so
00007f6618342000 4K rw--- libnss_files-2.28.so
00007f6618343000 36K rw--- [ anon ]
00007f661834c000 136K r---- libc-2.28.so
00007f661836e000 1312K r-x-- libc-2.28.so
00007f66184b6000 304K r---- libc-2.28.so
00007f6618502000 4K ----- libc-2.28.so
00007f6618503000 16K r---- libc-2.28.so
00007f6618507000 8K rw--- libc-2.28.so
00007f6618509000 16K rw--- [ anon ]
00007f661850d000 52K r---- libm-2.28.so
00007f661851a000 636K r-x-- libm-2.28.so
00007f66185b9000 852K r---- libm-2.28.so
00007f661868e000 4K r---- libm-2.28.so
00007f661868f000 4K rw--- libm-2.28.so
00007f6618690000 4K r---- libdl-2.28.so
00007f6618691000 4K r-x-- libdl-2.28.so
00007f6618692000 4K r---- libdl-2.28.so
00007f6618693000 4K r---- libdl-2.28.so
00007f6618694000 4K rw--- libdl-2.28.so
00007f6618695000 8K rw--- [ anon ]
00007f6618699000 72K rw--- [ anon ]
00007f66186ab000 4K r---- ld-2.28.so
00007f66186ac000 120K r-x-- ld-2.28.so
00007f66186ca000 32K r---- ld-2.28.so
00007f66186d2000 4K r---- ld-2.28.so
00007f66186d3000 4K rw--- ld-2.28.so
00007f66186d4000 4K rw--- [ anon ]
00007fff5f5bb000 232K rw--- [ stack ]
00007fff5f5fc000 12K r---- [ anon ]
00007fff5f5ff000 4K r-x-- [ anon ]
root@openmediavault:~# grep VmPeak /proc/4679/status
VmPeak: 94168 kB
It slowly creeps up, will check again in a few hours after maybe some garbage colection has taken place.
# service tellicast-client status|grep Memory:
# grep VmPeak /proc/4679/status
VmPeak: 185332 kB
the peak allocated virtual memory value increases, the allocated memory for the service fluctuates between 2 and 5 GB.
Received Packets on Data Channel: 31654570
Missed Packets before FEC: 3
FEC Recovered Packets: 3
Received Bytes on Data Channel: 36929653635
I see no reason not to upgrade to the 5.4 kernel. It's the new LTS kernel in the 5 series.
Your NAS GNU/Linux receiver seems to work perfectly now. I attach what my 11 years
old Fujitsu/Siemens CoreDuo receiver io did when I tested the TBS-6925 two days ago.
I also attach the patched driver with updated installation instructions once more.
If anybody is encouraged by this thread to give GNU/Linux a try there is a second
hand TBS-5925 for sale on www.eBay.de currently. A couple of people use it 24/7.
Let's close that thread now. I want to go back building my third Beowulf Cluster.
If your curiosity for PyTroll/SatPy shows results MSG-1 already started a thread.