Just Sharing This Using TBS6983/6903 Below 900MHz (STV6120)

[s-band]

It seems that the STV6120 has responses at fRX/3 and fRX/5. The /3 response is only about 12dB below the main and /5 is 20dB down. The result is that the spurious, high frequency, responses can interfere with low frequency signals. It doesn't matter if the upper channels are clear but when using something like a Sky Q LNB, channels below 11200 were up to 4dB below the expected SNR (tested on 1W). Using an ~850MHz LPF for low channels should fix it but I've not tried it.

 

[tom55]

Faced the same issue here with m88ts2022. Only difference is that it actually matters whether the upper band is clear. When it is there is no SNR degradation in the low band.
Also found your post on satellitescommunity.
STV6120 has the RF lowpass filter, maybe there is a chance it's bandwidth can be programmed?

 

[Channel Hopper]

Why would anyone use a SkyQ LNB with such a tuner ?

 

[tom55]

Just so it happens that i have 3 such devices. And i wasn't very happy seeing that unicable LNBs need extra programmer 5x more expensive than LNB itself.
Also it is not Sky Q, but inverto wideband LNB that i got

 

[s-band]

Why would anyone use a SkyQ LNB with such a tuner ?

Curiosity? Because it's possible? Because you have one?
I use the lower range on some Ka LNBs to get 2GHz span. In those cases there isn't usually anything strong in the upper part.

STV6120 has the RF lowpass filter, maybe there is a chance it's bandwidth can be programmed?

I didn't think it was programmable but am not 100% sure. I used an external anti-alias filter but that means routing the DC around it to power the LNB.

Using this comb of signals with a 20MHz spacing and 5dB level difference the alias products can be easily seen:



Centre frequency of comb set to 2.3GHz:

 

[tom55]

I suppose the actual cause of these responses has not been identified?
While thinking about it, it occured to me, maybe there is a tuner chip exhibiting this issue that has the possibility to power gate unused VCO.
I think manufacturers set these up to be constantly on, even when unused, probably to speed up channel switching time.

I will add that i put 2x4 voltage controlled multiswitch (high pass cutoff at 900mhz), it dampened low band, and all the responses from high band got SNR bump from 3-5dB to >15dB
That's how i was able to lock transponders on these harmonics and confirm the services match those at 3x IF

 

[s-band]

It's a zero IF RX where the LO is at the RX frequency. The odd harmonics of the LO (either from the LO or generated in the mixers) are high enough to give a response. e.g. If it's tuned to 800MHz then LOx3=2400MHz will also generate a response in the IF. Hence if you look at the example comb, the spacing of 20MHz reduces to 20/3MHz for the comb at around 800MHz. I used the trapezoid comb shape to check if there was a frequency inversion when I first looked at the problem.

strannik, on satellitescommunity.de, showed that it was possible to demodulate these responses and get the STV6120 to receive things above 2.4GHz.

 

[tom55]

I didn't think it was programmable but am not 100% sure. I used an external anti-alias filter but that means routing the DC around it to power the LNB.

Seems like it is. Take a look at this: longmynd/stv6120.c at master · myorangedragon/longmynd
You might get full wideband properly without external filters on the stv6120

 

[s-band]

Seems like it is. Take a look at this: longmynd/stv6120.c at master · myorangedragon/longmynd
You might get full wideband properly without external filters on the stv6120

Interesting, thanks. I'm not convinced that the front end filter is being set though. It looks like it is the low pass filter on I & Q which is being set to select the IF bandwidth.

 

[tom55]

Looking at the diagram, the 4 filters on I/Q channels are BB filters. They can't help with the issue. But these two after the LNA and before mixer are RF filters, and in datasheet they refer to these as HF filters which is also a description in the code.
The lowest value in lookup table is 854mhz, and just setting that value every time the channel tuned is <11.2ghz should do the job.

 

[s-band]

I missed that, I should stick to hardware! I see what you mean. It's interesting that there are filters up to 6796MHz so maybe it was designed to do harmonic mixing. Unfortunately the data sheet is only available under NDA and I don't have it. The short data sheet seems to have trimmed down specs e.g. frequency 250-2150 but it will do 140-2300+

I wonder if the standard drivers set the filters anyway, maybe @CrazyCat can help?

 

[tom55]

I guess it depends on what you consider 'standard drivers'. No idea if your tests were carried out on windows drivers, or linux drivers, but i would not count on any of the two, just as i wouldn't count on help from crazycat.
At least looking for sources in linux_media/stv6120.c at latest · tbsdtv/linux_media there is no HF filter code implemented.

From what I've found thus far, it is likely the bug occurs in PLL module, not in mixers.
That module also contains a filter, but not every diagram will show it

 

[tom55]

In case of montage ts2022, i wait, contact at hauppauge said to ask manufacturer, it's been over a month, probably they got nothing.

As a remainder, from another site "Spectral inversion happens when you mix an RF signal with a high-side local oscillator. "High side" in this context means that the LO signal is higher in frequency than the RF signal"

Spectrum appears to be compressed most of the times, but on several occasions it was inverted. Inverted spectrum confirmed now on 7e after a lot of transponders shut down

 

[tom55]

Trying to write 0x00 or 0xff to most of the tuner registers while doing spectrum scan leaded absolutely nowhere, except that some registers create other types of anomalies...
A scan without LNB or anything attached to the card, while having signal generated at 2252 Mhz in a close proximity of the card clearly shows the harmonic response at ~800 Mhz

 

[Channel Hopper]

What happens to the plot if you put a standard / universal LNB on the dish ?

 

[tom55]

If it wasn't obvious, total width reduces and both these frequencies fall out of it's range.
For the first time, i actually got a reply from montage, more precisely, their senior sales director. The question was if the new RS6060 chips support wideband.

"We can’t meet your requriement,because the frequency only support 950 ~2350MHZ."

 

[Channel Hopper]

If it wasn't obvious, total width reduces and both these frequencies fall out of it's range.
For the first time, i actually got a reply from montage, more precisely, their senior sales director. The question was if the new RS6060 chips support wideband.

"We can’t meet your requriement,because the frequency only support 950 ~2350MHZ."

Understood, but since I use a downconverter in line with the LNB to provide another 500MHz, you could see if the harmonic issue at 800 MHz (now 1300MHz) remains.

 

[tom55]

You can't manage downconverter dynamically so it ends up shifting entire band 500 mhz up. The tuners already have degraded levels at 2350mhz, and even if they could reach 3ghz then would probably need an amplifiers.

 

[Channel Hopper]

The converter I have can shift 500MHz up, or down, so a 2800 MHz signal should be within the manufacturer's range.

 

[tom55]

Do you know if there exist any external filter to workaround this problem?
Something like tracking lowpass filter, that would sense tuned frequency and based on programmed cutoff frequency switched filtering on/off ?