help to impedence match audio - avoid distortion on capture

[FramedMe]

Hi All,

Many thanks to all who've shared they're knowledge on this site!!! It's pot-o-gold for the new digital videographer.

I'm having trouble with audio distortion on capture. Here's what I'm doing.

VHS capture:
from:
Panasonic AG DS540 - SVHS VTR with Time Base Correction
(video out: s-video;
audio out L & R channel XLR (-6 dB, 50 ohms) -> 1/4" phono -> merged to 1/8" stereo)

to:
HW
ATI AIW Radeon 9800
AOpen AX4SPE MAX 865PE w/ 1 GB DDR400 & 2 x 120 GB SATA HD
(with 8x AGP, onboard audio [Realtek AC'97], 1 Gb LAN & IEEE-1394)

[s-video into the ATI AIW + 1/8" stereo audio into the onboard line-in]

SW
ATI MMC v8.6
AviSynth v2.5
VirtualDub v?
CCE v2.67
Ulead Movie Factory v2.12

My overall goal is to transfer 100s of SP VHS tapes containing mostly b/w telecine movies to DVD for TV viewing. I'm using a nice, semi-pro VTR deck to be able to get the best possible analog video signal I can to the ATI AIW digitizer.

This VTR has high-quality audio outputs that are set by default to the typical professional audio out impedence of 4 dB. If I had a professional-quality audio sampling system - either PCI board or USB/Firewire system - then this signal would be perfectly matched to what those input devices expect.

The standard, commerical PC audio card - and certainly the onboard audio input I'm dealing with - expects the kind of signal you'd typically get from a consumer audio device line-out which is -10 dB. Luckily, the Panasonic AG DS540 audio output circuitry has a switch enabling you to set the output impedence to other values to accomodate just such a situation. Unfortuately, the other settings are 0 dB & -6 dB. Since decibels are a log scale, being off by 4 (-6 dB instead of -10 dB) is still quite significant. I should probably shell out $200 or so on semi-pro, digital audio input hardware, but most of what I intend to digitize will have very lo-fi audio.

So - what happens when I capture video. As I mentioned, I'm just getting started doing this work, so I've been trying different options. One of the first methods I settled upon using was sampling into VirtualDub and saving the video + audio Huffyuv encoded as an AVI for further processing. When I went this route initially, the audio output was on the default 4 dB setting. No matter how I adjusted the audio output gain (the AG DS540 had meters to indicate audio output levels) or the audio input levels, I got terribly distorted sound - lots of crackling & popping, especially in the 'louder' passages. I then dug deeper into the AG DS540 manual and discovered this audio output impendence range switch. Once I set this to -6dB. I could then adjust the audio input gain to get very nice audio capture quailty into VirtualDub.

I was having other problems with VirtualDub, however (the ever present audio sync problems). When I happened upon LordSmurf's lucid and helpful site, he made clear how to make use of the included ATI MMC app to capture AVI, so I switched to that. I do like what I'm getting via that route as far as the video is concerned, but now the audio is once again distorted. This time, however, my audio output impedence is set to -6 dB, so I can't adjust that to better match it to the -10 dB the onboard audio line-in expects.

Does anyone know why VirtualDub would be able to grab well tempered audio, while the same signal captured via the same hardware into ATI MMC gives terribly distored audio.

I have several processing steps following capture - AviSynth filtering, CCE MPEG2 encoding, ULead MF authoring - so I don't want to add an additional filtering or demux/mux step to the process to fix the audio. It seems like there should be a way to fix this on capture, since the audio coming into VirtualDub is quite reasonable. Besides, if the input signal is truly out of range or pourly matched, no amount of digital futzing will bring back the original fidelity - garbage-in-->garbage-out, right.

Many thanks for any advice or feedback you have to offer.

Cheers

[FOO]

I'll probably get in trouble for this . but the impedance
doesn't matter with audio. And impedance is in OHMs not dB.
The level is in dB and the reference needs to be specified, e.g dBm.
The desired situation is to get the highest level possible into
the A/D converter without clipping. Looks like your -6dB setting is
OK with the sound card hardware.
ATI has had a number of problems in the past setting levels.
It's a pure software problem. Have you tried just
setting the record level lower ? That's not the same as line in
level displayed on the standard volume controls.

[BJ_M]

And impedance is in OHMs not dB. true

impedance for audio works as low impedance into high impedance (not matched as in video) .. we are out of the tube and transformer era now (for the most part - where its more importaint) ...

most capture cards have a very high impedance so it really doesnt mater to much .. but as stated the levels must be set //

[dphirschler]

Try an attenuator cable. You can probably get one at Radio Shaft. Also, www.core-sound.com has them.

I use a DAT recorder in line. The DAT recorder's pre-amp can handle the load. I then set recording level on the DAT deck, and go S/PDIF out to the sound card. No more audio distortion.


Darryl

[Capt_Diode]

@FramedMe,

XLR is typically 150 Ohms whereas line input is typically 10,000 Ohms. The impedence mismatch commonly causes distortion due to the mismatch in voltage levels in the impedence mismatch. I'm guessing you live in the States. Radio Shack sells an impedence matching plug for XLR to 1/4 inch stereo jack (Tip, Ring, Sleave). If you don't live in the States you can use an impedence matching transformer. Be careful with grounding. Improper grounding causes terible 50/60 Hz hum.

Good Luck,
Captain Diode

[Gazorgan]

You can buy a little transformer the size of a marble for low-high conversions. Just solder on desired connectors. Maybe a $7 project if you can't find one prewired.

The difference between -6 db and -10 db is NOT an issue worth bothering about, your dynamic range will far exceed the 4 db difference. Trust me.....I used to balance Cable TV headends (80+ channels of video and audio). High-Low impedence issues are another problem (most of my stuff was high impedance audio, but some emergency items aren't).

[FramedMe]

Many thanks to all for your very helpful replies.

Thanks for correcting my electronics gaffs, too. I'm afraid my E&M course is 25 years in the past and as a biologist & software engineer, I've not had a chance to use it much - though I do my fair share of circuit hacking, when necessary.

Capt_Diode: You hit my main concern right on the head - voltage levels. I'm expecting what I'm sending in is not quite what the typical PC sound card line-in expects. 10 kOhm is pretty high, but not high enough to compensate for a big mismatch. The output impedence is very low, too - 50 ohms.

Anyway - I did pick up my ALR -> 1/4" phono plugs from radio shack and was hoping they were handling a certain amount of the matching. I'll check the labels on the package again.


dphirschler: Thanks for the recommendation on using DAT player electronics to condition the signal. I wish I had a spare one sitting around.

I picked up a nice & relatively cheap signal conditioner/pre-amp from Turtle Beach to use for my turntable. That's how I did all my LP -> MP3 or AC3 transfers. That worked great, but it was particularly suited to match the output from a turntable to the PC's audio line-in and won't work on the amplified, ALR outputs from my VTR.

Howdy Foo: I've read many of your helpful comments amongst the 100s of threads I've parsed for advice as I've been getting up and running. Thanks for your input here. I think you've also hit to the heart of the matter. I remember specifically adjusting the input levels in VirtualDub. I can't seem to figure out how to do that in ATI MMC apart from using the system sound panel line-in control. Does the TV capture component of MMC have an audio input level adjustment setting somewhere? I've poked around everywhere and searched/read through what little they have in documentation. I can't find anything. I'd greatly appreciate any guidance folks can provide to uncover this widget, if it exists.

Thanks again for all the helpful feedback.

Cheers!

[The village idiot]

Just to clear something up. XLR can be many different impedances. Typical is 150 ohms for microphones, or other low impedance equipment. 600 ohms for most line level equipment.

Some active balanced amplifiers do not like having one of the balanced signals tied to ground, as has been a common practice when making balanced to unbalanced connections. They will over heat and distort! Some handle it just fine. Now to you problem, the -6db may still be incredibly high for your card. The card is most likely to be set for a -10db input at 0dbm, -6db would be 4 to many, and it is likely that the card will clip if it is one db too high. You can try changing your adaptor cable so that it may provide less signal. Connect XLR pin 1 to shield, and pin 2 to signal, leave pin 3 un-connected. Then shield goes to the shield of the mini connector, and signal to the tip or sleeve connection (depends whether it is left or right). You could also buy an adaptor that is made for converting balanced to unbalanced audio. Some of these will have an attenuator in them too, but not all. If it is still too loud, or you don't want to buy a balanced to unbalanced adaptor, you could simply put a "T" pad on the unbalanced side. You can find the formula here:

http://www.csgnetwork.com/decibelattenuatecalc.html

Impedance of 10000 input and output gives you 2 resistors of (R1 & R2) 5.2K ohms and one of about (R3) 7K ohms.

----R1--R2-----signal
      |
      R3
      |
---------------shield