Poor Performance

[Derrick]

Hello,

I've purchased a TBS 8921. After running some tests, I must say that I'm quite disappointed with the results. I'm interested in feed hunting and and in particular in stable reception of low bandwidht transponders. About blindscan I will make some comments in the other thread. But I will give a short resumé about the blindscan performance The tool in its present form is pretty useless and dangerous

I've performed my tests with transedit. The card is recognized and works well with my 17/8 ms. Scans on astra and hotbird showed no missing transponders.

Transponders with low SR showed a totally different behaviour. Apparently a very old driver bug (going back to the early days od B2C2 cards) has been copied again. An offset of 1-2MHz has to be added to the transponder frequency to get a lock. Here's a screen shot from the TBS 8921 and a TeVii 470S for the same transponder.





I've also tried AltDVB but the application couldn't get diseqc working:



The overall performance is rather poor and unstable (couple of BSODs).

Regards,
Derrick

[Derrick]

The altdvb diseqc problem has been solved with the BDA2 driver. With the latest TBSviewer_v20100921 I couldn't get diseqc working but dvbviewer-pro works ok, though for the recording service the latest Hardware_Database_2 has to be installed.

Remains the very poor reception performance. The result of a comparison scan with transedit on 42E for 52 transponders was 453 services in 120s with my tevii 470S, and only 408 services with the TBS8921 in 220s! I don't want by no means promote tevii. In fact the card has also quite a number of bugs but my resources are limited

Again mainly the low bandwidth SCPC-transponders are affected.

Will be continued..

[cody]

hello Derrick,

do i understand correctly - on Hotbird and Astra there are no performance issues, but on other satellites where there are low-symbol rate transponders you need to add 1-2 MHz frequency offset to get reception without missing packets? please, note that we have nothing to do with B2C2-based cards or their drivers and so no bugs are inherited from there.

regards,
cody

[Derrick]

@Cody,
Maybe not inherited but newly produced?

I was wrong about the offset frequency in one direction. In fact it's much worse. There's a kind of ripple or what you called birdies in the analog age;) Here's an example of a carrier with a nominal frequency of 11048MHz. On this frequency the TBS card won't lock. It will eventually lock if the frequency is varied. The scan produced with transedit uses transedit's "blind scan" function, which is just a simple try & error (fig1). In the example the frequency was increased in 1MHz steps between 11046 and 11050. The TBS scan yielded 2 hits on 11047 and 11050 (fig2). Fig3 shows the analyzer with the tevii card on the nominal frequency 11048.

This is a systematic bug that I could reproduce on many mor low_sr_carriers and which I have not ever seen before...

Regards,
Derrick

[cody]

Derrick,

it seems to me that either there is software (driver) bug with auto-frequency acquisition in TBS 8921 driver or just there is something wrong with your LNB frequency offset (LOF stability) on some frequencies. i will investigate with the same live signal on 42E east as on your screenshots. however, please, note that for me your tests with transedit's "blind scan" function and the tevii card don't show anything conclusive, because all current DVB-S2 demodulator have auto-frequency acquisition feature, which means that when you tell the tevii card to lock for example 11048 as on your screenshots if it's set to wider auto-frequency acquisition range it will get lock no matter of the actual frequency offset and return success on 11048 - i'm not sure if i'm explaining it clearly, but my point is that 11048 reported by transedit doesn't mean anything, because it's not the actual locked frequency reported by the DVB-S2 demodulator on the board itself. for example even if we assume there is some problem with TBS 8921 (either hardware or software) i can easily modify the auto-frequency acquisition range in the driver with increasing it to always return lock success on 11048, even if there is huge frequency offset and that is not the actually tuned frequency. in any case Linux driver for TBS 5921/6921/8921 can report actually locked frequency and thus frequency offset and read it directly from the DVB-S2 tuner/demodulator and i can tell you how to obtain that information - it's conclusive information and not like the test you ran in Windows.

regards,
cody

[Derrick]

..for example even if we assume there is some problem with TBS 8921 (either hardware or software) i can easily modify the auto-frequency acquisition range in the driver with increasing it to always return lock success on 11048, even if there is huge frequency offset and that is not tqCody,he actually tuned frequency.

@Cody,
You've got me wrong. Neither the ACBW (accquistion bandwidth) nor possible imperfections of my LNB (I get pefect locks on carriers with <500KS/s, alas not with pc-cards) are a concern. My 8921 locks narrow band transponders within a small frequency range (as it should be) but only if you tune either below or above the assumed correct frequency and not in the middle. A possible offset can only exist in one direction.

[Derrick]

..to make it clearer, another example. I've scanned 42E with transedit. 1st with my TeVii 470S. The card locked on 51 transponders (date and provider, if available). This list was rescanned with the TBS 8921. There were 13 transponders with no lock [X] (the number can vary). Then I took >19=11839,V,4444,34,1,|[X] 02.03: TERT and instead of transedit's blindscan I've created an ini with the same parameters in steps of 1 MHz. The scan yielded 2 hits: 1MHz below and 1MHz above the center frequency.

Code: Select All Code

[SATTYPE]
1=0420
2=Türksat 2A/3A 42.0°E
3=6
4=9750
5=10600
6=11700

[DVB]
0=51
1=10970,V,30000,56,1,|02.03: DOGAN TV
2=11012,V,30000,56,1,|02.03: DOGAN TV
3=11042,V,6400,34,6,|02.03: TRT
4=11048,V,2400,56,1,|02.03: XAZAR TV
5=11064,V,13000,56,1,|02.03: STV
6=11096,V,30000,56,1,|02.03: TURKSAT
7=11194,H,7200,34,1,|02.03: HABERTURK TV, Bloomberg HT, Service Provider
8=11465,H,15410,34,2,|02.03:
9=11495,H,15410,34,26,|02.03:
10=11521,H,10180,34,26,|02.03:
11=11527,V,2080,34,1,|02.03: Default Provider
12=11627,H,6665,56,1,|02.03: TURKUVAZ TEK. HIZ.
13=11651,H,2222,56,1,|02.03: TANDBERG
14=11658,H,2222,34,1,|02.03: MultiProgram
15=11716,V,2222,56,1,|[X] 02.03: OMG
16=11729,V,15555,56,1,|02.03: Digital Platform, DP 1
17=11804,V,24444,56,1,|02.03: DOGAN TV
18=11830,V,7200,56,1,|02.03: KANAL 7, YURDUM TV, ULKE TV, RADYO 7, ;, ,,,, tvt
19=11839,V,4444,34,1,|[X] 02.03: TERT
20=11844,V,2221,34,1,|[X] 02.03: EAS
21=11852,V,4340,34,1,|[X] 02.03:
22=11857,V,3148,56,1,|02.03: www.tgrt.com.tr, www.tgrt-fm.com.tr
23=11870,V,8888,34,1,|02.03: MNG TV
24=11879,V,2815,56,1,|[X] 02.03: PLUS MEDIEN TV
25=11887,V,9200,56,1,|02.03: Digital Platform, DP 1
26=11919,V,24444,34,1,|02.03: TRT
27=11962,V,3333,34,1,|02.03: T3A
28=11996,V,27500,56,1,|02.03: OMG
29=12130,V,27500,56,1,|02.03: TURKSAT, Scopus
30=12525,V,30000,56,1,|02.03: TURKSAT, HAYAT TV
31=12544,H,24000,56,1,|02.03: TURKSAT
32=12562,V,25000,56,1,|02.03: TURKSAT, SEBIL TV
33=12592,H,8888,34,1,|02.03: cine5
34=12596,V,3700,56,1,|02.03: TandbergTV
35=12601,V,2960,56,1,|[X] 02.03: Scopus
36=12602,H,4400,34,1,|[X] 02.03: KANAL A
37=12609,V,3700,0,0,|02.03: Tadiran Scopus
38=12615,V,4800,56,1,|[X] 02.03: TURKUVAZ TEK. HIZ
39=12618,H,2800,56,1,|02.03: SD Unnamed
40=12620,V,2400,56,1,|[X] 02.03: DMV
41=12623,H,2650,56,1,|02.03: D0GUS
42=12629,V,3740,56,1,|[X] 02.03: Tadiran Scopus
43=12640,V,2400,56,1,|02.03: DMV
44=12643,H,4800,56,1,|[X] 02.03: DOGUS
45=12650,H,4250,56,1,|[X] 02.03: Tadiran Scopus
46=12653,V,4444,0,0,|02.03: TVNET
47=12656,H,4444,56,1,|[X] 02.03: Scopus Network Technologies
48=12679,V,8888,34,1,|02.03: 24, KARAMEL TV
49=12685,H,30000,56,1,|02.03: TURKSAT
50=12729,H,30000,56,1,|02.03: TURKSAT
51=12729,V,30000,56,1,|02.03: TURKSAT


Code: Select All Code

[SATTYPE]
1=0420
2=TBS Test
3=6
4=9750
5=10600
6=11700

[DVB]
0=5
1=11837,V,4444,0,0,|[X]
2=11838,V,4444,0,0,|02.03: TERT
3=11839,V,4444,0,0,|[X]
4=11840,V,4444,0,0,|02.03: TERT
5=11841,V,4444,0,0,|[X]


[cody]

..for example even if we assume there is some problem with TBS 8921 (either hardware or software) i can easily modify the auto-frequency acquisition range in the driver with increasing it to always return lock success on 11048, even if there is huge frequency offset and that is not tqCody,he actually tuned frequency.

@Cody,
You've got me wrong. Neither the ACBW (accquistion bandwidth) nor possible imperfections of my LNB (I get pefect locks on carriers with <500KS/s, alas not with pc-cards) are a concern. My 8921 locks narrow band transponders within a small frequency range (as it should be) but only if you tune either below or above the assumed correct frequency and not in the middle. A possible offset can only exist in one direction.

there is no point arguing over this when there is definitive way to understand what's really going on and that's "actually/effectively tuned frequency" that every demodulator reports - as i told you the Linux driver for 8921 is already calculating and reporting the "actually/effectively tuned frequency", but in case you don't have Linux installation i will port the corresponding code to the Windows driver and output it as debug message in Windows and so you can test it in your case and until then there is really all just speculation. also, i know in full depth how your TeVii S470 card works and if you doubt that, you can look who made the Linux driver for the DVB-S2 tuner/demodulator on your so-good-working TeVii S470 card:

http://linuxtv.org/hg/v4l-dvb/file/e0cd9a337600/linux/drivers/media/dvb/frontends/ds3000.c

please, also note that i'm not saying there is no problem with your 8921 or even in general - that's under investigation and until there is something definitive - it could be general hardware or driver problem, it could be fault in your board, but until there are no numbers from the chips telling what's really going on it's as i already mentioned more or less just speculation what really is the root cause of the problem.

[Derrick]

there is no point arguing over this when there is definitive way to understand what's really going on

I agree. The frequency spectrum of a (qpsk) modulated carrier is symmetrical. If the pll manages to catch the carrier below and above the center frequency then there shouldn't be a hole in the middle!

My "so-good-working TeVii S470 " is just what I have for very low symbol rates. For 4444KS/s I can repeat the test with a FireDTV-S2. Same result Of course I don't know the exact carrier frequency in KHz. It depends on the LOF and stability of a normal off-the-shelf DRO-LNB. But I have also 2 STBs which are reporting back the frequency. I doubt that all are wrong and TBS is right For that I don't need linux

btw.
I still had 2 BSoDs today when did the TBS testing with TransEdit. I occurred when I closed the tool.

[Derrick]

Another test on Eutelsat W2A at 10.0°E (no feeds). With the reference a complete scan of 27 transponders was performed in 70s. It took the TBS 8921 3 times longer (210s) with 8 missing results [X] out of 27 ;(

Code: Select All Code

[SATTYPE]
1=0100
2=Eutelsat W2A at 10.0°E
3=6
4=9750
5=10600
6=11700


[DVB]
0=27
1=11144,H,2442,23,0,|02.04: Lebanon
2=11155,H,1630,56,1,|02.04: Network name
3=11158,H,2893,34,1,|[X] 02.04: ORTAS
4=11161,H,1320,56,5,|[X] 02.04: TandbergTV
5=11164,H,2308,34,0,|02.04: IGP
6=11167,H,3616,23,6,|02.04: NBC Russia, NBC Universal
7=11171,H,2532,34,0,|02.04: TandbergTV
8=11175,H,2532,34,0,|02.04: Service Provider
9=11178,H,2532,34,0,|02.04: TandbergTV
10=11184,H,1698,89,5,|[X] 02.04: SHANT TV
11=11222,V,27500,23,1,|02.04: BFBS
12=11279,V,2893,34,1,|[X] 02.04: WDS
13=11292,H,4820,12,1,|[X] 02.04: Thomson
14=11304,H,3642,12,1,|02.04: Thomson
15=11305,V,30000,56,2,|02.04: EBU
16=11346,H,27500,34,1,|02.04: Fileas
17=11429,H,30000,56,6,|[N] 02.04:
18=11513,V,30000,56,1,|02.04:
19=11554,V,3700,56,1,|[X] 02.04: Net 1
20=11622,V,2501,34,0,|02.04: TandbergTV
21=11623,H,2960,56,1,|[X] 02.04: IRIB
22=11627,H,2963,34,1,|02.04: IFILM
23=11631,H,1500,56,6,|02.04: Thaqalayn
24=11656,V,2592,34,1,|02.04: palestine
25=12596,H,2930,34,1,|[X] 02.04: Scopus Network Technologies
26=12610,H,9258,56,1,|02.04: RRSat
27=12620,H,5732,56,1,|02.04: DDS


According to the specs the card should be able to receive and demodulate DVB-S2 carriers between 1MS/s and 45MS/s. I could test one transponder with 45MS/s. The card locked within 3s which was only a fraction slower than the reference. The reason could be related e.g. to different diseqc timing. TBS sends a burst after the diseqc command according to the diseqc bus specs. Most other cards I know don't.

A totally different situation for DVB-S2 carriers with very small SR. Down to about 2MS/s the card will lock with an error free output mux, though it takes much longer than the reference (8s vs. 3s). Below around 1800KS/s the output in case of a lock is not reliable any more. It's impossible to determine the exact limits without a DVB-S2 multiplexer and modulator. I have to use real satellite transponders

Back to the 0100.ini (see above).
11161H, qpsk, 1320, 5/6 was locked by the reference in 8s with an error free output. After fiddling with the frequency (frequency bug) the 8921 showed an output signal after 15s, but this output was garbage:



the reference: