[LAD] [ot] rme fireface: weird balanced output measurements

[gene heskett]

On Friday, February 17, 2012 11:06:47 AM Jörn Nettingsmeier did opine:

> hi everyone!
> 
Greetings from your lurking retired broadcast engineer & CET.

I won't address the RME as I have no experience with it, what I will say is 
more akin to debunking and good practice stuff.

> yesterday, i visited my friends' new studio, to help them shake out some
> bugs in the patch and fix a monitor problem. there i came across a
> really weird issue with the line outputs on two rme fireface 800s:
> 
> we put a test tone out in logic (yeah, they run a mac shop), and i went
> to measure the outputs. in addition to really high fluctuations from one
> output to the next (with identical digital input signals), i measured
> huge differences in voltage on the hot and cold side, such as
> 
> hot to ground: 2.00 V
> cold to ground: 1.82 V
> 
First of several questions: Meter set for AC or DC?  Many, but not all, 
meters will read a DC offset as AC on the AC ranges.  Sort of hard to read 
the AC when its also reading .2 volts DC and adding (or subtracting) it to 
the AC reading.

> despite the fluctuations across several channels, this trend was pretty
> constant. so i figured, maybe this box has a problem with its negative
> voltage rail. they had another ff 800 in there, which we measured for
> comparison. same issue.
> 
> i figured, maybe apple wrote an oscillator with a dc offset, so we
> applied a known-good test tone .wav file. same result.
> 
> is my thinking flawed, or are rme really delivering such crappy output
> stages on such a pricey box?
> 
> my multimeter is not a calibrated one, but it's in the 100+€ range,
> and i used AC true rms measurement mode.

There are quite a few meters out there that claim true RMS, but few (if any 
in the price range of a hand held) are capable of that on a broadband audio 
basis, usually tweaked to be accurate within the boxes side panel claims 
ONLY at the line frequency of the local power, 50 or 60 hz.

The ONLY way to measure true RMS power over wide frequency ranges is with a 
bolometer, a teeny little resistor capable of about 10 milliwatts of power 
absorption whose temperature rise is how the power is measured.  Or by very 
accurate thermometers, one at the input of a resistors cooling flow, and 
one after the coolant has passed the resistor and been heated by that 
power.  There are some formulas about that will give 2% accuracy if the 
coolant is water and the flow rate is known to an equal accuracy.

Either of those methods costs 500-5000 USD to accomplish.

DVM's that claim true RMS use either a digital sample & store at many times 
the expected frequency, then do the math to discover the RMS, or a 
semiconductor based log converter.  

The digital method obviously doesn't fit in the price range or battery 
budget of a hand held.  The alternative is a quite fussy semiconductor log 
converter, which deals in currents so low near the bottom of the scale 
range that it is severely speed limited, often to less than the local line 
powers frequency.

Curious, playing with one of mine I paid about $125 for, I fed it from an 
audio oscillator whose output amplitude was monitored with a 100 mhz o-
scope.  It was passably accurate at my local power frequency of 60 hz, read 
about 25% high at 30 hz, read about 5% of what it should at 400 hz input, 
and wasn't able to see a 4000 hz signal at all!

Essentially, such a meter is worthless at common audio frequencies.  Its a 
sales gimmick that doubles the price of the DVM with zero real benefits.

Use an o-scope, it, unless you need distortion measurements, is far more 
informative than the meter, plus if you do see distortion on the scope 
screen, the format is often very explicit as to the cause when one is 
familiar with the scope enough to recognize the distortion.

> i understand my meter has a very high input impedance, and that's what
> line level connections should have, right? operating more or less
> open-loop, without significant currents flowing. or should i use a shunt
> resistor and measure across that? if so, what value is recommended?

That is another can of worms entirely.  The ideal is of course to design 
for an arbitrary impedance, often in the past, 600 ohms, an unfortunate 
artifact inherited from Ma Bell about 80 or 90 years ago.

Works for local runs of a few feet, but fails miserably on long runs.

Why?  The common two wire & foil shielded audio cable, used in broadcast 
and studio facilities in miles per studio quantities, actually has an 
impedance in the 60 ohm area!  Feed it with a 600 ohm source and 300 feet 
of cable later its rolled off like a Ma Bell telephone circuit.  Your audio 
DA's, to drive that, need to source terminate at 30 ohms per wire, from a 
very low impedance amplifier.

These can get hot & use lots of power to do a +20 dbm signal without 
clipping, and maintaining the usual 16 db of headroom established by the 
normal VU meter reading zero db at +4 dbm output.

Its also a good idea to load the far end with this same 60 ohms total to 
prevent any echos.  However, _if_ the line is properly source terminated, 
the lack of a load termination will likely be detectable only under very 
critical test conditions and can generally be ignored under the normal 
usage conditions of a recording studio.  You'll have 6 db more signal at 
the load without it.

In short, use an oscilloscope and leave the meter in the tool kit unless 
tracing power cable stuff.  You may have to learn what the scope is doing 
from scratch, but when you have done that, you'll be fixing things that 
will get you accused of being able to walk on water.  Or asked how you can 
sit so low in a chair with gonads the big.  Both have happened to me. ;-)

> thanks for any insights, best,
> 
> 
> jأ¶rn
> 
> _______________________________________________
> Linux-audio-dev mailing list
> Linux-audio-dev at lists.linuxaudio.org
> http://lists.linuxaudio.org/listinfo/linux-audio-dev


Cheers, Gene
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
My web page: <http://coyoteden.dyndns-free.com:85/gene>
If people are good only because they fear punishment, and hope for reward,
then we are a sorry lot indeed.
		-- Albert Einstein