Unbenchworthy

front

either

you?

addressed

total

can

ended

do

designed

and

pretty

only

PC

level of

with

budget

OK, being but a humble tech I am getting confused - looked at the glossy websites mentioned, sounds great - and, cause its on the internet, it must be true - eh?

And if you use a PC based (whatever) for test/measurement, for a $35 soundcard, free software, you get pretty impressive results - ie, you can SEE whats going on within the bandwidth limits of the soundcard- pretty cool.

And most of my "sophisticated" test gear is old, and out of calibration

- thats when its werking and doesnt need "routine maintenance" - would love a new 100Mhz Tek cro, but no way - the electricity bill comes first.

So, any chance of a tutorial on this subject rather than claim and counterclaim - its OK for you engineers, but some of the rest of us just get bloody confused....

73 de VK3BFA Anderw

snipped-for-privacy@hotmail.com wrote in news:1105418222.089503.72700 @c13g2000cwb.googlegroups.com:

Please tell that to Sound Technology and Audio Precision.

shows a picture of their PCI card which you so clearly disdain.

r

is a web site based on measurements based on computer sound cards.

is a web site based on listening tests set up and auditioned using computer sound cards.

OK, that's your opinion, Cal. So far your posts have been full of holes like Swiss Cheese if not Muenster. Let's see if you have the facts to back this one up.

Wrong. The audio interface can be wherever you want it to be, its just a matter of picking the physical configuration you want. Obviously Cal doesn't know about audio interfaces at all, even common ones like a Delta 1010 seem to have eluded his fractional-second technical *study*.

Obviously Cal, you are not well-informed about audio interfaces or PCs.

Wrong, Audio interfaces are available with a variety of common audio input and output connections including XLR, TRS, and RCA.

Wrong. In fact PC power supplies are as a rule built inside shielded steel or aluminum boxes with some kind of RF bypassing on all of the inputs and outputs.

Wrong. In fact some of the quietest audio interfaces that exist have their circuitry entirely inside the PC, on a PCI card.

Wrong, computer audio interfaces are widely used for audio production and professional audio purposes. They are widely availble with true balanced input and output connections. While audio interfaces are typically AC-coupled, so is virtually every other piece of audio production equipment on the market today.

You want quality audio? You pay for it. True for computer audio interfaces as well.

You still want quality audio? You still pay for it. Still true for computer audio interfaces as well.

Actually, there's a wide variety of computer PCI cards on the market that are used for test and measurement purposes.

You forgot the part where these supposedly unsuitable products are widely used and have excellent performance.

Right now its a two-option feasibility study. You buy an Audio Precision System Two for $10K (used) -$25K and up (new), or you pop a $850 LynxTWO into a $500 PC and run Audition ($300) and Spectra Lab ($800) software on it for a total of about $2500 brand new. BTW the prices I'm quoting are highly approximate, YMMV.

Finally, there are some problems with PCs + sound cards as test equipment that weren't covered in Cal's post, and as soon as Cal admits this post of his is like totally wrong, I'll tell what they are.

BS.

You are assuming that PC = Desktop PC. There are plenty of industrial PCs that have everything on the front.

You are assuming that Audio processor = Desktop PC Soundcard. There are PC-Based audio processors that use XLR connectors.

You are assuming that PC = Desktop PC. I havesitting in front of me a PC104 system that uses a very quiet linear power supply.

So? Good tools cost more. Do you imagine that going from an industrial PC with a sophisticated, instrument grade PCI card will cost more than a GPIB stack?

I suggest that you do some pricing, keeping performance and capabilities the same.

You are comparing cheap tools with expensive tools and then complaining because the cheap tools aren't as good as the expensive ones.

I have many bench instruments with connections at the back too, don't you?

Only for toy grade, even then they can be replaced.

Yes.

True, some may even matter.

Not will, MAY. As in all cases.

And yet a Lynx soundcard in a suitable PC, with suitable software, is pretty close to test equipment costing tens or hundreds of times as much. (When used within it's designed range of course.)

But the advantage of direct data aquisition combined with a high level of computing power and display/print/network abilities.

One always has a choice. Often many choices, depending on budget. Pretty handy eh?

MrT.

And very good too, if you were shopping for a soundcard 4 or more years ago :-)

We're sadly in need of a similar comparative review of current cards.

equipment

post of

Ever worked with a *real* THD analyzer? You probably have, right?

If you have, you know that they have input and output level controls that allow them to be used over a wide range of signal levels. Usually the output level control is more limited, but the input level controls on real THD analyzers can handle signals from less than a volt to several hundred volts. They also typically have a pretty good audio voltmeter function that you can use to set levels.

If you use a sound card for audio measurements, you pretty well have to add a good audio voltmeter and some kind of attenuator, particularly if you want to work with power amps.

tell

Admittedly, no. My day job application uses the sound card and my own program to analyze some non-audio circuits that happen to work in the audio frequency range. I am more concerned about shortcomings of the sound card as a measurement device, and not about audio specific test software. Your comments are reassuring.

Naturally I have undertaken to ensure that my results are meaningful. For my use, I have to apply a calibrated physical stimulus to the sensor, and this make my results somewhat self-verifying.

that

output

THD

volts.

you can

to add

you want

For my home hobby, I have mainly looked at preamps and speaker response. My main pet peeve is the lack of a calibrated voltage scale on the sound card. This requires me to stick with relative measurements, or check levels using my scope or voltmeter.

I could see how a sound card plus additional test equipment and ingenuity would be adequate for casual audio testing, but that a real test rig would make the whole process a lot quicker and easier.

OK, well I'm an old-timer who has worked with a bunch of different distortion analyzers, and lusted in detail over many of the rest.

And of course that is the most important thing.

To me one intersting game to play is to figure out how to confirm a measurement by independent means or means that make the setup;'s performance less ambigious.

This can work well with sound cards. As a matter of fact speaker analysis is about audio level signals in and out, plus a mic and a mic preamp. Slam-dunk and highly effective.

There you go.

The good news is that sound cards are typically stunningly accurate if you can compare them to something else that is accurate enough.

Mostly the output attentuator and a good audio voltmeter can suffice. I admit it, I used to use an external voltmeter with my old THS analyzer even though it had one built in. So its pretty close.

BTW I tend to build attenuators into my dummy loads. Usually, a 5K ohm 2 watt AB-type carbon comp potentiometer does the job, maybe with a switchable resistor in series with it for really large voltages. My dummy loads also tend to have XLR jack outputs from the attenuator to the sound card, and test points for the Fluke meter to plug into.

After you have checked it once, the levels should be nearly as consistant as your scope or voltmeter. Nether of which have a current calibration certificate I bet? Just make sure you don't use any variable software attenuation.

Quick, easy and wrong is simple to do, even with expensive equipment :-)

MrT.

I have owned a HP 339, an 8903, a Sound Tech box (which was a piece of shit like all ST boxes) and finally an AP ATS-1. The 339 was actually the most convenient in a lot of ways but it wasn't rugged, I had to replace some knobs and the dB switch assembly. High dollar. The ST is a gussied up 339 as is the Potomac pair (which splits the box in two , at a higher price singly than the 339....yet, and pay attention, Kroo, it sells....).

Possibly the chipset used in soundcards and an embedded SBC with a nice front panel, coupled up to a low distortion oscillator and analog back end on the gen side, would make a stunningly good test box....indeed this is essentially what AP has done.

The bottom line has little to do with electrical specs and a lot to do with form factor and setup. If the desktop or notebook PC were acceptable, VXI and PXI wouldn't exist, nor would the dozens of digital test boxes built into overgrown ruggedized notebooks.

My other beef with PC test is that Windows is such a disaster. Several actual test instruments have WinPCs embedded into them and you have to manage them just like every other PC in the house. I know a lot of people who have had to format and reload their Infiniium scope or R&S CSM because of viruses, worms, or crap installed by third shift techs.

All I ask if anyone out there does build a box like this...use a soundcard chipset that is an embedded-designated product, and make the front end part with through-hole parts, so we can fix it when we do something dumb.

performance

Well Cal, you see reality through a very dim lens.

There is no such thing as "the chipset used in soundcards". There are many chips that are used, and they break down into two functional areas.

(1) There is a chip of several possible standard chips or a proprietary chip that incorporates a PCI interface with perhaps a SP/DIF or I2C interface, relevant control registers, and maybe some DSP functions. Either that, or the function is implemented with some general purpose programmable chip. This chip may also include all or most of the active parts of the audio interface, but this is true only for low end audio interfaces.

(2) Higher end audio interfaces include one or more standard ADC and/or DAC chips or so-called CODEC chips that in this context include both ADC or DAC.

(3) Higher end audio interfaces also include an appropriate number of high-quality audio operational amplifiers.

Obviously, a wide range of audio functions and audio performance levels are possible, including the highest levels of performance that are currently available in any chips or discrete components.

In fact what most people do is simply drive the DAC with a low distortion signal from the digital domain. This can provide the function an incredibly low-distortion oscillator.

Most audio gear can be exercised nicely by the direct output of a good quality audio interface. As a practial matter typical signal levels are around 7 volts rms with a standard 600 ohm-compatible impedances, and residual distoriton well below 0.001% 20-20 KHz.

Pretty close. The fact of the matter is that an AP System Two is roughly matched in terms of signal accuracy by the best sound cards. In fact they are too good to be properly measured by the best, most recent AP hardware. IOW the AP test equipment does not have 20-20K THD+N residuals that are 10 dB better than the best audio interfaces.

Acceptable for what? The function and performance of an audio interface rarely come close to exploiting the performance capabilities of the standard PCI bus.

This stuff exists for many reasons including history, portability and convenience. These may have nothing to do with audio perforamance limitations.

I guess you never heard of MacIntosh and Linux computers, Cal. There is more to the audio world than PCs. The good news Cal is that not everybody shares your inability to manage windows PCs or even yout inability to conceive of how to manage a Windows PC.

Cal, you've already read testimonials from people who are sucessfully using PCs and PC audio interfaces for audio and audio-like analysis purposes. If you're not buying, fine. But if you are trying to impress us with your command of the capabilities of modern audio interfaces, you're failing miserably.

Inasmuch as I'm typing this on a Tadpole SPARCle running Solaris, I have heard of Unix (Linux is Unix). And PXI/CompactPCI is the PCI bus put on a different and more benchworthy form factor. Despite using the same bus, chipsets and firmware, the prices are pretty harsh. VXI is the old VMEbus and the prices are really out there. Sorry.

Putting Linux on the PC, or using a Mac, SPARC or SGI doesn't change the fact these are desktop computers and not test instruments. PC test-as opposed as PC _control_ of test, which is what AP and Agilent and Tek have done for decades-is awkward, electrically iffy, and a pain in the ass and it sucks. It's as simple as that.

Be that as it may, I would love to have a Collins 95S-1.....although not really an audio box, it would be a fun toy. Would be more fun if it had a front panel, though.

Of course not, unless you can operate it near absolute zero :-)

MrT.

Actually "Linux" ins an acronym for "Linux Is Not UNIX."

Linux cannot fix 'electrically iffy.' It also can't fix 'unreliable' and 'not rugged.' You can, however get industrial PCs that don't have these flaws.

Linux can fix 'awkward', 'a pain in the ass', and 'it sucks', but only if the programmer understands human interfaces. Most programmers don't - they spend years learning the fine details of an environment where one comma in place of a period makes the entire program useless. This ruins them for human interface work. EEs have a different problem; we tend to build what is easy to build rather than what is easy to use. Marketting droids have yet a third problem; they tend to specify what is easy to learn rather than what is easy to use, with sidetracks into what looks impressive rather than what is easy to learn or use.

Linux is simply the kernel developer/project instigator's first name used in the tradition of Unix based OS projects, which all end in -ix, or -ux. (AIX, A/UX, Ultrix, Irix, and Minix, most importantly), not a recursive acronym in the GNU tradition.

The OS isn't the problem. M$ Windows, a poor operating system, is used in many excellent test instruments on a fully or semi-embedded basis. PC desktop hardware just is not designed to work as test equipment. It's like making a modern Japanese motorcycle engine work in an airplane-it has an integrated gearbox and the mechanical arrangement is all wrong, plus, all of them are first designed by cartoonist/animator types and the engine is then designed within that visual envelope. Making cards with proper connectors and putting them in a "backwards" case (or putting the optical drives and power and reset butons on the back) is a little start: going to a PXI/CompactPC form factor in a test instrument like backplane/mainframe is a big improvement. You are talking a lot of money now and you still don't have a physical user interface,i.e. knobs!