Unreliable tuning with HVR-950q

Unreliable tuning with HVR-950q

[Scott Bronson]

I set up a Hauppauge HVR-950q on Ubuntu Intrepid, kernel 2.6.27-2.  It
works out of the box except that it can only achieve a lock maybe 50%
of the time.  Once it gets the lock, it's perfect.  Never loses it, no
signal issues whatsoever.  I live so close to the tower, though, that
I don't see why it wouldn't tune 100% of the time.

Is there any way to debug this?  Is the transmitter overpowering my
card?  I tried both the crappy antenna that came with the 950q and a
6.5 dB indoor picture frame antenna, no difference.

Thanks!


Log from the last 3 days:

CHAN  TIME            DISTANCE according to antennaweb.org

11_1    03:00   fail    12 miles
11_1    03:00   fail
11_1    03:00   ok

2_1     00:00   fail     1 mile
2_1     00:00   ok
2_1     00:00   ok

44_1    00:30   fail    1 mile
44_1    00:30   ok
44_1    00:30   ok
44_1    18:30   fail
44_1    19:00   fail
44_1    21:00   ok
44_1    22:30   fail
44_1    22:30   ok

54_3    10:00   ok    maybe 35 miles?
54_3    10:00   ok

65_3    11:30   fail    5 miles maybe?
65_3    11:30   ok
65_3    11:30   ok
65_3    12:00   fail
65_3    12:00   fail
65_3    12:00   fail
65_3    14:00   fail
65_3    14:00   ok
65_3    23:00   fail
65_3    23:00   fail

9_2     05:00   fail    1 mile
9_2     20:00   ok
9_2     21:00   fail
9_3     09:00   fail

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Devin Heitmueller]

On Fri, Sep 19, 2008 at 5:25 PM, Scott Bronson <bronson@rinspin.com> wrote:
> I set up a Hauppauge HVR-950q on Ubuntu Intrepid, kernel 2.6.27-2.  It
> works out of the box except that it can only achieve a lock maybe 50%
> of the time.  Once it gets the lock, it's perfect.  Never loses it, no
> signal issues whatsoever.  I live so close to the tower, though, that
> I don't see why it wouldn't tune 100% of the time.
>
> Is there any way to debug this?  Is the transmitter overpowering my
> card?  I tried both the crappy antenna that came with the 950q and a
> 6.5 dB indoor picture frame antenna, no difference.
>
> Thanks!

What application are you using?  Kaffeine?  Scan?

Devin

-- 
Devin J. Heitmueller
http://www.devinheitmueller.com
AIM: devinheitmueller

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Scott Bronson]

On Fri, Sep 19, 2008 at 2:28 PM, Devin Heitmueller
<devin.heitmueller@gmail.com> wrote:
> What application are you using?  Kaffeine?  Scan?

mplayer and MythTV.  They seem to be about equally unreliable.

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Scott Bronson]

On Fri, Sep 19, 2008 at 8:05 PM, Scott Bronson <bronson@rinspin.com> wrote:
> On Fri, Sep 19, 2008 at 2:28 PM, Devin Heitmueller
> <devin.heitmueller@gmail.com> wrote:
>> What application are you using?  Kaffeine?  Scan?
>
> mplayer and MythTV.  They seem to be about equally unreliable.

And scan of course.  I haven't noticed any problems with it missing channels.

Does scan try to acquire a lock?  Would it be useful for me to run it
5 or 6 times in a row and see how consistent its results are?

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Steven Toth]

Scott Bronson wrote:
> On Fri, Sep 19, 2008 at 8:05 PM, Scott Bronson <bronson@rinspin.com> wrote:
>> On Fri, Sep 19, 2008 at 2:28 PM, Devin Heitmueller
>> <devin.heitmueller@gmail.com> wrote:
>>> What application are you using?  Kaffeine?  Scan?
>> mplayer and MythTV.  They seem to be about equally unreliable.
> 
> And scan of course.  I haven't noticed any problems with it missing channels.
> 
> Does scan try to acquire a lock?  Would it be useful for me to run it
> 5 or 6 times in a row and see how consistent its results are?

I guess it's possible that the frontend is being overwhelmed. What 
happens if you discinnect the antenna and hold it very close to the 
antenna connector on the usb device and try locking on a channel that 
doesn't normally lock (on a very close transmitter).

- Steve

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Scott Bronson]

On Fri, Sep 19, 2008 at 11:37 PM, Steven Toth <stoth@linuxtv.org> wrote:
> I guess it's possible that the frontend is being overwhelmed. What happens
> if you discinnect the antenna and hold it very close to the antenna
> connector on the usb device and try locking on a channel that doesn't
> normally lock (on a very close transmitter).

Yes, that does seem to help.  I can't find a distance that gives locks
100% of the time, but 1mm distance seems to take it from 50% to
80-90%.

It seems to always get a lock if I just keep trying.  I just quickly
keep hitting return and escape until I get a picture.  It rarely takes
more than 3 attempts.  However, if I just leave it alone as it tries
tune, it will sit there forever (well, at least 1/2 hour) and never
get it.

Thanks, tell me if there's anything more I can do.

    - Scott

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Steven Toth]

Scott Bronson wrote:
> On Fri, Sep 19, 2008 at 11:37 PM, Steven Toth <stoth@linuxtv.org> wrote:
>> I guess it's possible that the frontend is being overwhelmed. What happens
>> if you discinnect the antenna and hold it very close to the antenna
>> connector on the usb device and try locking on a channel that doesn't
>> normally lock (on a very close transmitter).
> 
> Yes, that does seem to help.  I can't find a distance that gives locks
> 100% of the time, but 1mm distance seems to take it from 50% to
> 80-90%.
> 
> It seems to always get a lock if I just keep trying.  I just quickly
> keep hitting return and escape until I get a picture.  It rarely takes
> more than 3 attempts.  However, if I just leave it alone as it tries
> tune, it will sit there forever (well, at least 1/2 hour) and never
> get it.
> 
> Thanks, tell me if there's anything more I can do.

This is really starting to sound like a signal level issue. Try 
attenuating, this will probably help. Radio shack probably have a 
selection of 5 and 10db inline connectors that should help.

I don't know where our lab guys from ours from.

This isn't something that we can compensate for in software.

- Steve

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Scott Bronson]

On Sun, Sep 21, 2008 at 7:28 PM, Steven Toth <stoth@linuxtv.org> wrote:
> This is really starting to sound like a signal level issue. Try attenuating,
> this will probably help. Radio shack probably have a selection of 5 and 10db
> inline connectors that should help.

I'll try that.  It will attenuate distant stations too, right?  I
might have to buy two tuners, one for close stations and one for far
away ones...?

> This isn't something that we can compensate for in software.

I'm not sure about that...  Rapidly re-trying seems to always produce
a lock.  I suppose the driver could do that automatically when needed.
 (yes it's a hack, but drivers have always needed to hack around
hardware limitations)

Is there any way to log the tuning metrics, like signal and noise
level, etc?  There may be a pattern here.

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Steven Toth]

Scott Bronson wrote:
> On Sun, Sep 21, 2008 at 7:28 PM, Steven Toth <stoth@linuxtv.org> wrote:
>> This is really starting to sound like a signal level issue. Try attenuating,
>> this will probably help. Radio shack probably have a selection of 5 and 10db
>> inline connectors that should help.
> 
> I'll try that.  It will attenuate distant stations too, right?  I
> might have to buy two tuners, one for close stations and one for far
> away ones...?

That's fairly unusual, unless you're trying to receive TV at the low 
extremes of the tuner. I guess never say never.

> 
>> This isn't something that we can compensate for in software.
> 
> I'm not sure about that...  Rapidly re-trying seems to always produce
> a lock.  I suppose the driver could do that automatically when needed.
>  (yes it's a hack, but drivers have always needed to hack around
> hardware limitations)
> 
> Is there any way to log the tuning metrics, like signal and noise
> level, etc?  There may be a pattern here.

Watch the SNR value, it's measures in db and expressed in hex with one 
decimal place. So, 0x102 = 25.8db

- Steve

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Scott Bronson]

On Mon, Sep 22, 2008 at 12:22 PM, Steven Toth <stoth@linuxtv.org> wrote:
> Watch the SNR value, it's measures in db and expressed in hex with one
> decimal place. So, 0x102 = 25.8db

OK.  Watching the display in MythTV, when I see this I immediately
know that the tuner will never lock.

    Signal 0% | S/N 2.4dB | BE 0 | (L__) Partial Lock


And when I see this, I know that in 1/2 second I'll see a TV picture:

    Signal 0% | S/N 2.4dB | BE 0 | (LAM) Lock


It's always the same: 0% signal, 2.4dB S/N and BE 0.  Does "L__" offer
any hints as to why locks are so elusive?

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Steven Toth]

Scott Bronson wrote:
> On Mon, Sep 22, 2008 at 12:22 PM, Steven Toth <stoth@linuxtv.org> wrote:
>> Watch the SNR value, it's measures in db and expressed in hex with one
>> decimal place. So, 0x102 = 25.8db
> 
> OK.  Watching the display in MythTV, when I see this I immediately
> know that the tuner will never lock.
> 
>     Signal 0% | S/N 2.4dB | BE 0 | (L__) Partial Lock
> 
> 
> And when I see this, I know that in 1/2 second I'll see a TV picture:
> 
>     Signal 0% | S/N 2.4dB | BE 0 | (LAM) Lock
> 
> 
> It's always the same: 0% signal, 2.4dB S/N and BE 0.  Does "L__" offer
> any hints as to why locks are so elusive?

Scott, thanks for the info.

I'm not actually comfortable trusting the myth display without digging 
into it's code, I don't know what it does with the values. Try using the 
azap statistics, which I'm happier to discuss.

At this point I don't see any software fixes, we're not going to make 
the driver retune a number of times. You're overwhelming the tuner/demod 
frontend with high RF levels.

If you have SNR numbers they would be good to see, but they are for 
informational purposes at this point. They are a curiosity.

You likely need an attenuator.

- Steve

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Scott Bronson]

Another data point...

I bought a used Pinnacle PCTV HD and set it up last night.  It works
perfectly -- tunes every channel 100% of the time.  No glitches at
all.  Meanwhile, the HVR-950q is still getting a lock only 50% of the
time on good days.  The difference between these devices is dramatic.


On Tue, Sep 23, 2008 at 8:14 AM, Steven Toth <stoth@linuxtv.org> wrote:
> You're overwhelming the tuner/demod
> frontend with high RF levels.

The attenuators are in the mail, 3dB and 6dB...  I'll report back when
I've tried them.

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Scott Bronson]

On Fri, Sep 26, 2008 at 9:52 AM, Scott Bronson <bronson@rinspin.com> wrote:
> On Tue, Sep 23, 2008 at 8:14 AM, Steven Toth <stoth@linuxtv.org> wrote:
>> You're overwhelming the tuner/demod
>> frontend with high RF levels.
>
> The attenuators are in the mail, 3dB and 6dB...  I'll report back when
> I've tried them.

I've now tried the attenuators.  Verdict: didn't help.

3dB: no difference that I can detect.
6dB: probably a little worse
3dB+6dB in line: definitely worse

If I was overwhelming the tuner, wouldn't it have a hard time keeping
the lock too?  Because once my 950q locks, it keeps it rock-solid
forever, perfect video.  It's just getting it in the first place
that's hard.

My Pinnacle PCTV HD works 100% perfectly connected to the same
antenna.  Locks every time, excellent video for hours.  No problems.

Does anybody else have a 950q?  Does it work?  To me, it appears that
either its tuner is junk or the Linux driver needs some work.  I guess
my next step is to set up a Windows machine and try it there.

    - Scott

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Steven Toth]

Scott Bronson wrote:
> On Fri, Sep 26, 2008 at 9:52 AM, Scott Bronson <bronson@rinspin.com> wrote:
>> On Tue, Sep 23, 2008 at 8:14 AM, Steven Toth <stoth@linuxtv.org> wrote:
>>> You're overwhelming the tuner/demod
>>> frontend with high RF levels.
>> The attenuators are in the mail, 3dB and 6dB...  I'll report back when
>> I've tried them.
> 
> I've now tried the attenuators.  Verdict: didn't help.
> 
> 3dB: no difference that I can detect.
> 6dB: probably a little worse
> 3dB+6dB in line: definitely worse
> 
> If I was overwhelming the tuner, wouldn't it have a hard time keeping
> the lock too?  Because once my 950q locks, it keeps it rock-solid
> forever, perfect video.  It's just getting it in the first place
> that's hard.

No, not necessarily. My comments to you were based on the fact that 
you're 1 mile from the broadcast antenna. Assuming your antenna is in 
good shape then I'd expect the hvr950q to be a great choice. It's got 
great sensitivity (xc5000 + AU8522) and I use it all the time (for 
digital cable).

Good result btw.

We've also done extensive testing on ATSC and we have no obvious issues.

> 
> My Pinnacle PCTV HD works 100% perfectly connected to the same
> antenna.  Locks every time, excellent video for hours.  No problems.

Hmm. Maybe we really are talking about a bad stick. Can you try this in 
a windows box?

I'd expect the performance to be the same under Linux and Windows. If it 
isn't then we have driver issues.

Can you try this?

> 
> Does anybody else have a 950q?  Does it work?  To me, it appears that
> either its tuner is junk or the Linux driver needs some work.  I guess
> my next step is to set up a Windows machine and try it there.

We _do_ have rock solid HVR950q support for digital, on lots of 
platforms in various locations.

Based on the facts so far it's probably something specific to your 
stick, your location or your testing. Try windows and report back if you 
can.

Regards,

Steve

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Devin Heitmueller]

Hello Steven,

On Mon, Oct 6, 2008 at 11:43 AM, Steven Toth <stoth@linuxtv.org> wrote:
>> Does anybody else have a 950q?  Does it work?  To me, it appears that
>> either its tuner is junk or the Linux driver needs some work.  I guess
>> my next step is to set up a Windows machine and try it there.
>
> We _do_ have rock solid HVR950q support for digital, on lots of platforms in
> various locations.
>
> Based on the facts so far it's probably something specific to your stick,
> your location or your testing. Try windows and report back if you can.

Just an FYI:  I'm actively debugging an issue with xc5000 signal lock
that looks like some sort of timing issue.  I found that some fixes I
made in my local tree to optimize time to lock resulted in getting
lock *much* faster in many cases, but failing entirely in others.
When it works, I get lock in 330ms; when it fails I don't get lock
even when polling the demod at 100ms interval for 2 seconds.

I'm doing this on a Pinnacle 801e (xc5000/s5h1411) and not a HVR-950q,
but it's possible it's the same issue.

Will report back when I have isolated the problem.

Regards,

-- 
Devin J. Heitmueller
http://www.devinheitmueller.com
AIM: devinheitmueller

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list

Re: Unreliable tuning with HVR-950q

[Steven Toth]

Devin Heitmueller wrote:
> Hello Steven,
> 
> On Mon, Oct 6, 2008 at 11:43 AM, Steven Toth <stoth@linuxtv.org> wrote:
>>> Does anybody else have a 950q?  Does it work?  To me, it appears that
>>> either its tuner is junk or the Linux driver needs some work.  I guess
>>> my next step is to set up a Windows machine and try it there.
>> We _do_ have rock solid HVR950q support for digital, on lots of platforms in
>> various locations.
>>
>> Based on the facts so far it's probably something specific to your stick,
>> your location or your testing. Try windows and report back if you can.
> 
> Just an FYI:  I'm actively debugging an issue with xc5000 signal lock
> that looks like some sort of timing issue.  I found that some fixes I
> made in my local tree to optimize time to lock resulted in getting
> lock *much* faster in many cases, but failing entirely in others.
> When it works, I get lock in 330ms; when it fails I don't get lock
> even when polling the demod at 100ms interval for 2 seconds.
> 
> I'm doing this on a Pinnacle 801e (xc5000/s5h1411) and not a HVR-950q,
> but it's possible it's the same issue.
> 
> Will report back when I have isolated the problem.

Good to know, thanks Devin.

- Steve

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list