Blindscan on linux with tbs cards

[EnoSat]

ROOT 16416 = GOLD 131070
The question is whether the card supports both inputs

 

[Llew]

ROOT 16416 = GOLD 131070
The question is whether the card supports both inputs

So maybe it doesn't according to my unsuccessful attempt with GOLD 131070 and success with the ROOT PLS. Any ideas why? Maybe just this si2183 demod ?

 

[deeptho]

So maybe it doesn't according to my unsuccessful attempt with GOLD 131070 and success with the ROOT PLS. Any ideas why? Maybe just this si2183 demod ?

According to the driver both should work, but it is possible there are bugs

 

[JDavid22]

Now it works better! It finds and blocks more MUX, only as a bug maybe that some mux when it blocks, it stays there and does not continue with the other mux. As said, the scanning is still as slow.

 

[deeptho]

It is a pity you cannot receive the same satellites as I do. Otherwise I would ask you for an example.
Or perhaps you can send "neumo.log" and tell me on which mux it gets stuck?

I have been working on completely new code that has less problems like these, but it has not yet been released.

 

[JDavid22]

Here is more evidence...

 

[deeptho]

I have released new versions of the drivers. The main recent novelty is
to provide support for compiling under linux 6.1 and 6.2 kernels. It is important to
follow the instructions on github when compiling and installing.

You will need

GitHub - deeptho/linux_media: neumoDVB blindscan drivers media_tree

neumoDVB blindscan drivers media_tree. Contribute to deeptho/linux_media development by creating an account on GitHub.

github.com

and

GitHub - deeptho/media_build: media_build for compiling the neumoDVB media_tree on this site

media_build for compiling the neumoDVB media_tree on this site - deeptho/media_build

github.com

 

[JDavid22]

Is this enough?

cd ~/blindscan_kernel/media
git reset --hard
git pull
cd ../media_build
make -j8
#check for error
sudo make install
#then reboot

Or should I upgrade the kernel to 6.1-6.2 ?

I have Fedora 36 and kernel 5.17.5-300.fc36 .x86_64

 

[deeptho]

Is this enough?

Yes if you have installed all prerequisites.

Or should I upgrade the kernel to 6.1-6.2 ?

I have Fedora 36 and kernel 5.17.5-300.fc36 .x86_64

 

[JDavid22]

Thanks

 

[Bobben_no]

I hava a little problem after my ubuntu auto updated to kernel 6. I am using TBS6903 card. The drivers modified for kernel 6 has installed correctly and I can see symbolrate and modulation is detected automatically. The only thing not working is DiSEqC control when I try tools like tsduck.io. I am using simle Diseqc 1.0 four ports (AA AB BA B. It is working in the drivers from tbsdtv and also it has been working in the old blind scan drivers I installed in March 2023.

 

[deeptho]

I hava a little problem after my ubuntu auto updated to kernel 6. I am using TBS6903 card. The drivers modified for kernel 6 has installed correctly and I can see symbolrate and modulation is detected automatically. The only thing not working is DiSEqC control when I try tools like tsduck.io. I am using simle Diseqc 1.0 four ports (AA AB BA B. It is working in the drivers from tbsdtv and also it has been working in the old blind scan drivers I installed in March 2023.

Can you give me a command line that goes wrong for you?

 

[Bobben_no]

##Example with working TBS drivers tsduck 13 EAST diseq 1.1 switch the fourth port ( BB )
## tsduck sends diseqc command E0 10 38 FE:

root@dvbtest:/home/dvbtest#tsp -I dvb --device-name /dev/dvb/adapter0:0 --satellite-number 3 --frequency 11,137,000,000 --symbol-rate 27,500,000 --polarity horizontal  --delivery-system DVB-S2 --modulation 8-PSK --fec-inner 3/4 -P bitrate_monitor --periodic-bitrate 5 -O drop
* bitrate_monitor: 2023/09/22 08:51:43, TS bitrate: 37,972,992 bits/s
* bitrate_monitor: 2023/09/22 08:51:48, TS bitrate: 38,045,184 bits/s
* bitrate_monitor: 2023/09/22 08:51:53, TS bitrate: 37,972,992 bits/s

I have added logging of diseqc commands in tsduck and I can see this four diseqc 1.0 commands supported:
E0 10 38 F2 ; port 0 (--satellite-number 0)
E0 10 38 F6 ; port 1 (--satellite-number 1)
E0 10 38 FA ; port 2 (--satellite-number 2)
E0 10 38 FE ; port 3 (--satellite-number 3)

I have also tested dvbtune which only works with DVB-S in orginal tbs driver:

root@dvbtest:/home/dvbtest#dvbtune -c 0 -D 4 -f 11137000 -s 27500 -p H -m                                                                                     Using DVB card "TurboSight TBS 6903 DVB-S/S2 "
tuning DVB-S to L-Band:791621423, Pol:H Srate=27500000, 22kHz=off
polling....
Getting frontend event
FE_STATUS:
polling....
Getting frontend event
FE_STATUS: FE_HAS_SIGNAL FE_HAS_LOCK FE_HAS_CARRIER FE_HAS_VITERBI FE_HAS_SYNC
Event:  Frequency: 11137000
        SymbolRate: 27500000
        FEC_inner:  3

Bit error rate: 0
Signal strength: 34768
SNR: 42640
FE_STATUS: FE_HAS_SIGNAL FE_HAS_LOCK FE_HAS_CARRIER FE_HAS_VITERBI FE_HAS_SYNC
Signal=34768, Verror=0, SNR=42640dB, BlockErrors=0, (S|L|C|V|SY|)
Signal=34768, Verror=0, SNR=42640dB, BlockErrors=0, (S|L|C|V|SY|)
Signal=34768, Verror=0, SNR=42640dB, BlockErrors=0, (S|L|C|V|SY|)
Signal=34768, Verror=0, SNR=42640dB, BlockErrors=0, (S|L|C|V|SY|)

There is no lock when I use the same command lines with the blindscan drivers installed. It is working great apart from the diseqc issue.

 

[deeptho]

I do not have a 6903 card, but I tried the closest thing available: tbs5927 and it seems to work
fine:
```
tsp -I dvb --device-name /dev/dvb/adapter12:0 --satellite-number 3 --frequency 11432000000 --symbol-rate 29900000 --polarity vertical --delivery-system DVB-S2 --modulation 8-PSK --fec-inner 3/4 -P bitrate_monitor --periodic-bitrate 5 -O drop
* bitrate_monitor: 2023/09/22 21:56:25, TS bitrate: 65,117,184 bits/s, net bitrate: 64,101,082 bits/s
* bitrate_monitor: 2023/09/22 21:56:30, TS bitrate: 65,189,376 bits/s, net bitrate: 64,160,339 bits/s
^C* tsp: user interrupt, terminating..
```
Possible reasons why it does not work for you
1. not using the most recent drivers
2. an lnb which has a large deviation in local oscillator frequency: neumodvb compensates for this problem
and then tunes, but other software does not and is therefore more critical for tuning. For example, with a
"corrected" frequency it may work
3. some specific bug for 6903.
4. some problem with the settings.
5. some problem with your diseqc setup: weird problems occur at powerup. Therefore neumodvb inserts
some small delays at specific points during tuning. Normally tsp will also do that, but the delays may not be same
and perhaps not enough.

You could compare the calls made to the driver by neumodvb and by tsp, e.g., using strace or by activating kernel debug logging and checking the messages. That could tell if there is a real bug.

 

[Bobben_no]

I just checked out the exact versions for the kernel and the user space program which are on github and tested with the TBS 5972
All works fine (apart from the printf problem). Clearly the "FE_GET_EXTENDED_INFO unknown error..." is what goes wrong.
Why I don't know. Perhaps not using the correct kernel code, or something else. Does the kernel log show anything interesting?

I understand if you give up. I think I will also check in my version of media_build. Perhaps it helps to compile the proper kernel.

I notice my kernel is 6.2.0-31-generic while I can see on the github it is mentioned kernel 6.1.
Kernel sources is installed and can not see any errors when running ./install.sh

Can see this kernel warning message :

[    3.737123] WARNING: You are using an experimental version of the media stack.
                As the driver is backported to an older kernel, it doesn't offer
                enough quality for its usage in production.
                Use it with care.
               Latest git patches (needed if you report a bug to linux-media@vger.kernel.org):

I can also see this at the end of running ./install.sh:

Hmm... distro kernel with a non-standard place for module backports detected.
Please always prefer to use vanilla upstream kernel with V4L/DVB
I'll try to remove old/obsolete LUM files from /lib/modules/6.2.0-31-generic//updates/dkms:
find: warning: '-name' matches against basenames only, but the given pattern contains a directory separator ('/'), thus the expression will

To make sure I clean old driver install and install blindscan drivers i run this commands:

rm -rf /lib/modules/`uname -r`/kernel/drivers/media/
rm -rf /lib/modules/`uname -r`/updates/extra/media/

mkdir ~/blindscan_kernel
cd  ~/blindscan_kernel
git clone --depth=1  https://github.com/deeptho/linux_media.git ./media
git clone https://github.com/deeptho/media_build
cd media_build
git checkout deepthought
git reset --hard
make dir DIR=../media
make distclean
./install.sh
wget http://www.tbsdtv.com/download/document/linux/tbs-tuner-firmwares_v1.0.tar.bz2
sudo tar jxvf tbs-tuner-firmwares_v1.0.tar.bz2 -C /lib/firmware/
reboot

When I list files in ~/blindscan_kernel/media_build/linux/drivers/media/dvb-core I can see some odd extensions .orig and .c~.
I assume that is normal ?

-rw-r--r--  1 root root  36568 sep.  23 10:00 dmxdev.c
-rw-r--r--  1 root root  36554 sep.  23 10:00 dmxdev.c.orig
-rw-r--r--  1 root root  47950 sep.  23 10:00 dvb_ca_en50221.c
-rw-r--r--  1 root root  34223 sep.  23 10:00 dvb_demux.c
-rw-r--r--  1 root root  34209 sep.  23 10:00 dvb_demux.c.orig
-rw-r--r--  1 root root  26784 sep.  23 10:00 dvbdev.c
-rw-r--r--  1 root root  26372 sep.  23 10:00 dvbdev.c~
-rw-r--r--  1 root root  26358 sep.  23 10:00 dvbdev.c.orig
-rw-r--r--  1 root root 109989 sep.  23 10:00 dvb_frontend.c
-rw-r--r--  1 root root 109975 sep.  23 10:00 dvb_frontend.c.orig
-rw-r--r--  1 root root   5212 sep.  23 10:00 dvb_math.c
-rw-r--r--  1 root root  42508 sep.  23 10:00 dvb_net.c
-rw-r--r--  1 root root  10140 sep.  23 10:00 dvb_ringbuffer.c
-rw-r--r--  1 root root  10342 sep.  23 10:00 dvb_vb2.c
-rw-r--r--  1 root root   2447 sep.  23 10:00 Kconfig
-rw-r--r--  1 root root    350 sep.  23 10:00 Makefile

I also notice that my TBS5520SE (5521) does not get detected with blindscan drivers running and kernel log show errors related to RC. I edited ./install.sh to solve that problem:

#sed -i -r 's/(^CONFIG.*_RC.*=)./\1n/g' v4l/.config
#sed -i -r 's/(^CONFIG.*_IR.*=)./\1n/g' v4l/.config
#sed -i -r 's/(^CONFIG.*_BT87X.*=)./\1n/g' v4l/.config
sed -i -r 's/(^CONFIG.*_MXL603.*=)./\1n/g' v4l/.config
sed -i -r 's/(^CONFIG.*_MTV23X.*=)./\1n/g' v4l/.config
#sed -i -r 's/(^CONFIG.*_GX1503.*=)./\1n/g' v4l/.config

Not sure if all of the lines needed to be removed.

I also get this message but I assume that is expected:

Bug: mm88rs6060 cards error attaching lnb control on adapter

The TBS5520SE diseqc do not work in orginal tbsdriver or blindscan driver but diseqc has never worked in linux or windows. Also my TBS6504 has never worked with diseqc in linux and windows . I think TBS5520SE and TBS6504 uses same hardware (Si2183). I have tested spaun diseqc and EMP diseqc switch. I think I once saw something written in the TBS forums regarding diseqc tone level being to low for some diseqc switches and that it can be adjusted in the drivers. Unfortunately I do not know where in the drivers it can be tested/adjusted .

I also found this on the TBS forums:

22khz wired only from first demod from each si21832 dual demod chip.
si2183 frontend driver set 22khz tone and send diseqc cmd for each demod instance

For TBS5520SE that can not be a the reason for the diseqc issues but perhaps it does for the TBS6504.

Not sure if it is similar with stv091x but I have tested both inputs on my TBS6903 card with blindscan kernel but diseqc (Spaun and EMP) does not work.

 

[deeptho]

tbs6504 should work with diseqc. It works for me (but not tested recently).
I am not sure with which switches that I used at the time, but I also have some spaun and emp ones.

The errors you report suggest that you are not using the recommended installation procedure.
The ".orig" files suggest that patching may not be successful.

The error "mm88rs6060 cards error attaching lnb control on adapter" obviously prevents diseqc from working,
but should not affect tbs6504.

Why not stick to a kernel that has been tested?
6.3.8-200 works for me, but there have been various changes before that in the linux kernel which
could cause compilation problems.

 

[Bobben_no]

tbs6504 should work with diseqc. It works for me (but not tested recently).
I am not sure with which switches that I used at the time, but I also have some spaun and emp ones.

The errors you report suggest that you are not using the recommended installation procedure.
The ".orig" files suggest that patching may not be successful.

The error "mm88rs6060 cards error attaching lnb control on adapter" obviously prevents diseqc from working,
but should not affect tbs6504.

Why not stick to a kernel that has been tested?
6.3.8-200 works for me, but there have been various changes before that in the linux kernel which
could cause compilation problems.

Thank you for the feedback. I would like to stick with a kernel that has been tested so I will need to find out how to install a spesific kernel on ubuntu 22.04. But first I will wipe the hdd and make a new ubuntu install and see if it works better.

Btw, I have used the installation procedure from GitHub - deeptho/linux_media: neumoDVB blindscan drivers media_tree :

mkdir ~/blindscan_kernel
cd  ~/blindscan_kernel
git clone --depth=1  https://github.com/deeptho/linux_media.git ./media
git clone https://github.com/deeptho/media_build
cd media_build
git checkout deepthought
git reset --hard
make dir DIR=../media
make distclean
./install.sh
wget http://www.tbsdtv.com/download/document/linux/tbs-tuner-firmwares_v1.0.tar.bz2
sudo tar jxvf tbs-tuner-firmwares_v1.0.tar.bz2 -C /lib/firmware/
reboot

 

[deeptho]

Thank you for the feedback. I would like to stick with a kernel that has been tested so I will need to find out how to install a spesific kernel on ubuntu 22.04. But first I will wipe the hdd and make a new ubuntu install and see if it works better.

Btw, I have used the installation procedure from GitHub - deeptho/linux_media: neumoDVB blindscan drivers media_tree :

mkdir ~/blindscan_kernel
cd  ~/blindscan_kernel
git clone --depth=1  https://github.com/deeptho/linux_media.git ./media
git clone https://github.com/deeptho/media_build
cd media_build
git checkout deepthought
git reset --hard
make dir DIR=../media
make distclean
./install.sh
wget http://www.tbsdtv.com/download/document/linux/tbs-tuner-firmwares_v1.0.tar.bz2
sudo tar jxvf tbs-tuner-firmwares_v1.0.tar.bz2 -C /lib/firmware/
reboot

Wiping the hard disk is never the right solution: it takes a lot of time, and even if it fixes some problem (unlikely) then it will equally "unfix" any solutions found so far.

Specifically: you can install more than one kernel at the same time. I have less experience with ubuntu myself.

 

[Bobben_no]

Wiping the hard disk is never the right solution: it takes a lot of time, and even if it fixes some problem (unlikely) then it will equally "unfix" any solutions found so far.

Specifically: you can install more than one kernel at the same time. I have less experience with ubuntu myself.

I replaced the harddisk with an empty one just to keep my ubuntu disk untouched. I have now installed fedora 38.
Kernel is 6.2.9-300.fc38.x86_64.

Followed the installation procedure 100%. Got message about missing vmlinux so I installed it. After that everything seems to compile.
Diseqc still do not work. I can tune ok with auto SR etc. if diseqc is not in use.

dmesg still show "Bug: mm88rs6060 cards error attaching lnb control on adapter". Does my TBS6903 use a m88rs6060?

I can see files installed in /lib/modules/6.2.9-300.fc38.x86_64/updates/extra/media/

 

[Bobben_no]

I got it to work now.

I edited stv091x_drv.c :

static int wait_dis(struct stv *state, u8 flag, u8 val)
{
    int i;
    u8 stat;
    u16 offs = state->adapterno ? 0x40 : 0;

    for (i = 0; i < 10; i++) {
        stat = read_reg(state, RSTV0910_P1_DISTXSTATUS + offs);
        if ((stat & flag) == val)
            return 0;
        msleep(100);  // changed from 10 to 100
    }
    return -1;
}

With the same change diseqc also works in ubuntu .

It is probably my EMP and Spaun diseqc switches that did not work with current timing of diseqc commands.
Perhaps it is a better way to solve it but for now it works fine in both fedora 38 kernel 6.3.8 and ubuntu 22.04 kernel 6.2.0 .

I will probably have to do something similar for si2183 as well to get my TBS6504 and TBS5520SE to work with diseqc.