Got the signal strength to report correctly in dBm now. Had to pull it from the tuner chip.
I still don't think the Windows Driver really does anything with strength. If it does, it doesn't do it correctly.
Posts by majortom
-
-
Also sold re-branded by 'Mohu'.
I am only posting this here in the off chance it is useful to someone...
In this linux-media post,
someone mentioned this cheap atsc tuner that didn't work in linux yet..
So I bought a couple to play around with and got it working, fairly decent.
Only issue so far is that haven't been able to get any meaningful i2c reads from the demod's AGC registers.
So the 'Strength' will always report 0%. Not even sure if the windows driver reports strength properly, so
hard to troubleshoot, with no documentation other than source code. As in a windows application used to
test and capture logs, smartDVB, it was always reading 97% strength, whether locked to a good signal
or attempting to tune an known channel with no expected signal.
But, at least it can work now on a modern linux kernel.
This branch, 'stock' should build an out of tree cxusb module u can install on a stock kernel / media_tree.
The 'master' branch, is only meant for users of UDL's v4l-updatelee tree.
-
I have never read this thread from beginning to end, but I think if you do, you will find there is a solution to m88ds3103b and some folks with m88ds3103b reporting it working. I have never seen such chip so can't help...
-
I gotcha, so now there is another revision of m88ds3103 chip, that's gonna need among other things some updates to m88ds3103_priv.h in the sense of another table of initial register settings, etc???
ie like an additional table of init vals, and some way to trigger loading them upon recognizing a new chip id?
and guessing with no documentation, the only way we are gonna know what they should be set to is to have the newest hardware in
hand, and either do some i2c sniffing, or usb snooping of an updated windows driver that actually works?
or wait to see if dvbsky implements the new chip also, and actually contributes an updated m88ds3103 module to the kernel...
Just a guess, that we might see something from dvbsky developers before geniatech? seein as they actually went thru the hoops to implement an entire usb driver for the mainline linux kernel based on their hardware.. Assuming dvbsky has to buy the same new chip from Montage, eventually.
ughh
static const struct m88ds3103_reg_val m88ds3103b_dvbs_init_reg_vals[] = {
static const struct m88ds3103_reg_val m88ds3103b_dvbs2_init_reg_vals[] = {
-
If using the raspberrypi, IMO, the best way to get that device working properly is to use the raspberrypi tree, that darksky an I maintain....
that has UDL's drivers merged in, located here.
If anyone uses arch linux, or archlinuxarm, there are package builds u can download, to make compiling/installing easier.
No, we don't contribute to aur, so you won't find it there. I can compile install the pi kernel in about 20 mins
on an old laptop here with the pkgbuild, crosscompiling (see makepkg-arm).
v4l-updatelee pkgbuild for x86_64:
The raspberrypi tree was just updated the other day, and also has the ts2020 Loop Thru mod in it,
in case ur bored and wanna install the missing Coax jack on your hdstar's PCB.
-
Couple of snapshots to demonstrate how well the geniatech hdstar Hardware revision 3 works here.
I don't use crazycat's tree here, but when i first got the hdstar device working in linux, using the m88ds3103 demod,
instead of the older ds3000 module, I noticed it was working MUCH better than the ds 3000 module ever did.
I sent my patches to crazycat, and he graciously modified it so as to attach the m88ds3103 / ts2022 chip a little more gracefully than what I was doing... his way autodetects the demod chip id and loads the appropriate module (ds3000 old vs m888ds3103 new), whereas my way was kinda brute force to make it work just for me.
Anyhow we then implemented his version in to UDL's tree, located here. That was a few years ago...but to this day still works like a champ.
for the hdstar:
First of all, to determine whether you have the SU3000 hw V3 or not, see the photo of the PCB that I uploaded to the linuxtv wiki,
around the time we first got this device to function properly in linux.
geniatech_hd_star_dvb-s2_usb2.0
If yours doesn't say SU3000 V3.0 silkscreened on the PCB then it's probably an older hardware version with the older ds3000 / ts2020 chips.
If it does say SU3000 V3.0 on the PCB, then look closer at the chips, they should be m88ds3103 demod / ts2022 tuner, like mine in that photo.
That would be a good thing...
The hardware IS different, but both old and new have the same USB Vid/Pid which is why ya have to use the detection of the demod chip id via i2c to determine which module should be loaded. That's the trick crazycat implemented, and we have included in UDL's tree.
I would enable the real stats in your tvheadend configuration to display the signal stats in dBm for strength, and dB for Quality.
The percentages ur looking at for comparison are useless... If you notice in my screenshots, the strength dBm and Quality dB are identical to
the updateDVB program I am using.
If interested, Since then I had noticed the empty Coax connector pads there in the rear of the PCB, and installed a new 75 ohm coax jack there, and modified the code a little bit in the TS2020 module to enable the Loop Thru (IF Loop) on that previously empty coax path. I then use that to feed my dvbsky S960 at times. IF Loop path is only active when the hdstar is active however.
