[pvrusb2] Noise

[Jan Ceuleers]

Mike,

Thanks for your reply.

Given that in the past it was possible (in fact necessary) to select the
audio input and that this is no longer possible/necessary now, am I
right in thinking that the driver programs the hardware to add all audio
sources together on the assumption that only the "correct"-one will
actually be producing an audio signal and the others will be silent?

If so, are there circumstances under which this assumption might be false?

Again if so, is there anything I can do about that? For example, in my
use case (where I'm only recording from line in and composite video in)
can I disable certain components that aren't being used? For example get
rid of certain firmware images, or use certain pvrusb2 module parameters?

Thanks, Jan



On 21/04/2019 14:54, isely at isely.net wrote:
> Jan:
>
> That's an odd symptom.  Getting audio from an RF signal actually has a 
> longer processing path than audio from the line-in port.  In fact, 
> there should be very little (if anything) involved with the line-in port 
> that isn't also involved with processing audio from an RF signal.  In 
> either of those two cases, the audio is processed by a separate chip 
> (cx25840 IIRC) and the result of that is what is fed into the mpeg 
> encoder chip.  Once past the cx25840, there's no difference in the 
> datapath.
>
> There's a separate driver for the cx25840, which the pvrusb2 driver 
> employs, which is part of the v4l2 subsystem as a whole, so any problem 
> involving audio like you describe should also be likely with any other 
> video capture driver that also happens to use a cx25840.  I know this 
> isn't being very helpful, but it does suggest that looking for other 
> online chatter involving that part might reveal another clue.
>
>   -Mike
>
>
> On Sun, 21 Apr 2019, Jan Ceuleers wrote:
>
>> Dear list,
>>
>> I am seeking some help with an audio defect that afflicts some
>> recordings made using Hauppauge PVR 1950 devices.
>>
>> I have two mythtv backends:
>>
>> - the slave backend has three 1950s that record from analogue cable.
>> Audio recorded from these tuners is fine.
>>
>> - the master backend has two 1950 PVRs that record from set-top boxes
>> (i.e. from the composite video input and line audio inputs). All
>> recordings made using these two tuners suffer from an audio defect as
>> described below.
>>
>> The audio defect is a clearly audible repetitive "whistling" noise, with
>> a periodicity of about 1 second. This noise is not present on the audio
>> signal that goes into the PVR.
>>
>> I have tried many variations of v4l2-ctl -d$device -f $freq without
>> effect (and have settled on 450 MHz as the frequency). The point of
>> doing this is that I want to exclude noise from the channel the analogue
>> tuner happens to be tuned to.
>>
>> Many moons ago a pvrusb2 device had multiple audio inputs from which one
>> could select the right-one using v4l2-ctl -d$device
>> --set-audio-input=$audiodev , with valid values for audiodev being 0, 1
>> and 2. The correct setting for my use case was 1. This is no longer the
>> case: the device presents only one audio input which is numbered 0.
>>
>> I'd be grateful for any hints.
>>
>> Thanks, Jan
>>
>> _______________________________________________
>> pvrusb2 mailing list
>> pvrusb2 at isely.net
>> http://www.isely.net/cgi-bin/mailman/listinfo/pvrusb2
>>