Line-in overload
« on: August 07, 2013, 11:42:12 AM »
Been looking all over and I can't find this answer. Can you help?

What is the typical USB soundcard line-in AC clipping voltage? I mean max voltage in with no headroom at all for a typical unbalanced line-in.

I figure that since they are all powered with 5V it can't be more than 5V P-P (1.78V RMS, +7.22 dBU, +5 dBV). But this is a gross assumption on my part. My guess is that max in is less than that.

Anybody have experience in this area?

Thanks
« Last Edit: August 07, 2013, 11:52:20 AM by oneoldude »


Kingston

Re: Line-in overload
« Reply #1 on: August 07, 2013, 02:38:46 PM »
It's not wise to estimate the headroom of anything based on just the voltage it's operating. Nearly always somewhere in the input stage the input audio signal is padded (attenuated) to interface with the actual A/D chip input. This way we can have 110dB headroom with even +22dBu max input before clipping with correct interfacing and design of the input analog stage into the A/D chip. All this is quite possible and very often done with the "weak" 5V USB power. Since this 5V is not a really a limitation for design, it's a bit difficult to define a typical operating spec. See the specifications of several hardware units of a certain "class" and you will find they will somewhat conform, but are probably not perfect equivalents.

Did you have a specific piece of hardware in mind?

Andy Peters

Re: Line-in overload
« Reply #2 on: August 07, 2013, 06:44:31 PM »
What is the typical USB soundcard line-in AC clipping voltage? I mean max voltage in with no headroom at all for a typical unbalanced line-in.

I figure that since they are all powered with 5V it can't be more than 5V P-P (1.78V RMS, +7.22 dBU, +5 dBV). But this is a gross assumption on my part. My guess is that max in is less than that.

There's no way to tell. Cheap units will just use the unregulated USB 5V as a rail, and that voltage can be as low as 4.6V. Some units will regulate the USB supply so the analog rail won't vary. Whether it's regulated at 5V or not is a design detail. Still others may use a switcher and generate a bipolar supply.

And of course there will be some attenuation because the ADC chip input range is often only about 2Vpp. So perhaps a pad is present in front of the input amplifiers, or perhaps not.

I suppose that the documentation for the particular piece of kit should tell you its input range.

-a
"On the Internet, nobody can hear you mix a band"

Re: Line-in overload
« Reply #3 on: August 08, 2013, 05:35:19 PM »
Problem solved.

Instead of trying to come up with a general solution that would apply in many situations, I decided to attack one problem and deal with that.

I bought a Behringer UCA222 that has a specified line-in max input of 1.25V RMS.

That 1.25V RMS = 1.768V P = 3.536V P-P = +4.157 dBU = +1.938 dBV

That will handle up to the Studio international level standard of +4 dBU.

Now I can design a mic-pre to match the mic sensitivity against SPL input against for the circuit overload point.

With that I will be able to measure loudspeaker drivers with less complications.

Thanks 

Re: Line-in overload
« Reply #4 on: August 08, 2013, 09:27:04 PM »
I bought a Behringer UCA222 that has a specified line-in max input of 1.25V RMS.

That 1.25V RMS = 1.768V P = 3.536V P-P = +4.157 dBU = +1.938 dBV

That will handle up to the Studio international level standard of +4 dBU.


"standard" +4 mean 0Vu read with continuous sin wave.... that mean your line in/out are set to +12dBu with same sin (considering Vu time integration). Add 10dB for headroom and you ended with +22dBu... this is the "max" in "standard" pro audio
your berhinger is far from this!


Re: Line-in overload
« Reply #5 on: August 09, 2013, 12:55:43 AM »
Where do you guys get this stuff?

The VU meter came about in the 1930's to help set levels for the installation of early telephones. Its purpose was to show a technician where to set the audio level so as not to get into distortion. That is why the numbers above 0 dB are in red. Red is to be avoided!

The recording guys picked it up from the telephone guys and look at the sorry state its in now.

In analog circuits there is some distorted head room above the ideal standard at 0 dB. But if you used an ADC, all information above 0 dB would not exist because it would exceed the bit capability of the ADC. There is no headroom above 0 dB for an ADC.

The real definition for 0VU (0 dB VU) is as follows and I quote:
 
"Technically speaking, 0VU is equal to +4 dBm, or 1.228 volts RMS across a 600 ohm load."  http://www.mediacollege.com/audio/monitoring/vu.html Look it up elswhere if you don't believe them.

Notice the 1.228V RMS. Any input that can handle 1.228V RMS or more will handle 0VU. The UCA222 (and I am sure many other lowly sound cards) will handle 1.25V RMS. That is more than the +4 dBU standard. So the UCA222 will handle +4 dBU (which is the voltage reference at any load not just at 600 ohms).

Designers of digital gear have designed the analog front end of ADCs with pads and gain circuits that are transparent to the user so as to mimic old fashioned analog gear. This allows the user to treat the system like old time gear and push signals into distortion to get that artistic sound they want. So be it. But for goodness sake, that has nothing to do with how the stuff actually works or the voltage it operates at.

You cannot put a 22 dBU signal directly through a circuit that is powered with 5V and get a good result. That is true. But with a 20 dB pad you can make it handle the signal and still have 2 dBU to spare. All that is happening is the signal is being crushed by 20 dB so the peaks are not overloading the ADC. But it is still a 5V system.

This is as far as I am going with this discussion. But for those of you who want to know how it actually works, here are some audio recording type URLs to get you started. To learn about how ADCs work and the voltages available to sound cards, you are on your own.

Good luck!

http://www.sengpielaudio.com/calculator-db-volt.htm

http://www.sengpielaudio.com/calculator-gainloss.htm

http://www.sengpielaudio.com/calculator-transferfactor.htm

 



audiomixer

Re: Line-in overload
« Reply #6 on: August 09, 2013, 08:01:09 AM »
That seems like a very angry responce to some helpful posts.... sengpiel is explicitly mentionning that there is no direct correlation between dBFS and dBu. Most AD converter do buffer and scale the input according to the ADs requirements and would have a powercircuitry to suit. No Problem to get a bipolar supply from +5v..

Cheers, michael

gyraf

Re: Line-in overload
« Reply #7 on: August 09, 2013, 08:21:02 AM »
Yup, let's keep this civilized..

I think your disagreeing point is whether you simply go for +4dBU (analogue) as 0dBFS (digital), or you follow the recommended (for 24-bit systems) 20dB-headroom-above+4, which requires some +26dBu of power handling from the system. But I may be wrong...

Jakob E.
..note to self: don't let Harman run your company..

Re: Line-in overload
« Reply #8 on: August 09, 2013, 10:16:34 AM »
Oneoldued !!!

What can I say?

I think your disagreeing point is whether you simply go for +4dBU (analogue) as 0dBFS (digital)

I think that too, I just can't let say +4dBu as 0dBfs is a "standard" Vu studio level...

Notice the 1.228V RMS. Any input that can handle 1.228V RMS or more will handle 0VU. The UCA222 (and I am sure many other lowly sound cards) will handle 1.25V RMS. That is more than the +4 dBU standard. So the UCA222 will handle +4 dBU (which is the voltage reference at any load not just at 600 ohms).

here we are, with all respect, there is a BIG confusion. Lot of cheap manufacturer call "+4" (+4 what? potatos?) hypothetically reminding "standard" 0Vu (which is what i describe before)
in you case it's the manufacturer extreme lying as the "+4" (marketing) IS the full scale minus 22mV (joking headroom)

Your Berhinger will only handle a 1.25V RMS pure sine wave that's it. No relation at all with the Vu meter specification or studio line level reference which is up to each studio choices, somewhere around what Jakob explained.

no pb with the idea of pad input, but what do you do with output when you need unity gain with you AD-DA converter?

Am I lucky ?  :)








JohnRoberts

Re: Line-in overload
« Reply #9 on: August 09, 2013, 11:08:45 AM »
It is difficult to interpret tone from written posts...  Lets not change his handle to angryolddude just yet.

It sounds like he thinks this is a simple question and only he knows the answer.

The 5V nominal power supply for USB poop only defines a max analog "output", nothing more, and even that can be gamed.

Hopefully he will warm to the group and get a better feel for the IQ around here over a little more time.

JR
It's nice to be nice....


gyraf

Re: Line-in overload
« Reply #10 on: August 09, 2013, 01:25:11 PM »
Thanks John - I fully agree...

Jakob E.
..note to self: don't let Harman run your company..

tv

Re: Line-in overload
« Reply #11 on: August 09, 2013, 04:13:09 PM »
otoh, based on my limited experience with "multimedia" soundcards, it is wise to PAD (and safety-clip) the signal yourself (i.e. outside the interface)

it can be done easily with 2 green leds, 2 schottkys and a couple of resistors

if you're lucky, it will be remarkably "close to spec"
If you sprinkle when you tinkle, please be neat and wipe the seat.

Andy Peters

Re: Line-in overload
« Reply #12 on: August 09, 2013, 07:43:48 PM »
In analog circuits there is some distorted head room above the ideal standard at 0 dB. But if you used an ADC, all information above 0 dB would not exist because it would exceed the bit capability of the ADC. There is no headroom above 0 dB for an ADC.

You conflate the ADC chip with a generic audio product called an "Analog to Digital Converter."

Yes, the chip has an absolute maximum input level. That level is called 0 dBFS. Its level is not determined by the analog voltage rails. It's set by the circuit's reference voltage. In some low-quality applications, the reference voltage may be the supply rail, but for any kind of real accuracy a proper reference is used. The device data sheet tells you the acceptable voltage reference range, and in pretty much all cases that reference voltage cannot exceed the rail.

That's just the chip. The circuit requires an input buffer/driver, and part of that design includes level matching between the product's expected maximum input voltage and the converter's 0 dBFS level.

If the expectation is that the product is to be used in a professional environment, the input circuit design will target an input level of +4 dBu to map to -12 dBFS or -18 dBFS or whatever. The "whatever" is a wiggle factor; the idea is that you also have to establish the analog clip point to be at or perhaps a bit higher than 0 dBFS. Mapping -12 dBFS at the converter to +4 dBu at the input means that the maximum input level is +16 dBu or 4.9 Vrms or 13.8 Vpp. If you want more input headroom, you map +4 dBu to -18 dBFS, so full scale input is +22 dBu, or 9.75 Vrms or 27.6 Vpp.

So obviously some amount of attenuation is required to ensure that the full-scale voltage doesn't exceed what is set by the reference.

-a
"On the Internet, nobody can hear you mix a band"


 

Related Topics

  Subject / Started by Replies Last post
1 Replies
2721 Views
Last post November 27, 2004, 07:14:20 PM
by Flatpicker
24 Replies
4547 Views
Last post April 21, 2006, 03:53:11 AM
by Purusha
1 Replies
1278 Views
Last post October 26, 2007, 03:28:34 PM
by CJ
21 Replies
4228 Views
Last post March 09, 2008, 03:18:44 PM
by JohnRoberts