Speaker measurement tools

[Panomaniac]

If you have a soundcard with ASIO drivers, that can go a long way toward solving those types of problems.

[mah]

If your soundcard does not feature ASIO then you can add it with the following freeware.

ASIO4ALL - Universal ASIO Driver

Google ASIO4ALL to get more information. I intend to try it at a later date so this post is not a personal recommendation but just information that might save a DIYer from an upgrade. Your risk of course. :coffeedrinker:

Regards, Marshall.

[Panomaniac]

I didn't have good luck with ASIO4All. I mean it worked OK, but did not seem to be "exclusive" the way a real ASIO driver is. Windows volume and mixer still seemed to be in effect, and those were 2 of the things I especially did not want!

Any good experiences with it are welcome. I like the ASIO drivers because nothing else can address the soundcard when you are using them. Maybe my ASIO4ALL isn't working right.

[westend9]

I respectfully disagree. I think they want lab results at bargain prices and there's nothing wrong with that. It's actually doable if do your homework before you buy. That was my point. I learned this the hard way, going thru 3 different setups before I was satisfied.

Whether building speakers from scratch or upgrading and tweaking, DIYers are seeking to get that last bit of performance that simple economics preclude in most production systems. If that last bit of perfomance is shaving a 3dB peak at x frequency, you need the tools to reveal that detail accurately.

Just for posterity, I want lab results at bargain prices. Thanks, bfish.

Well the equipment might be able to do lab quality results on the cheap, but learning to use it is another matter. Skills take time. And guidance is not unwelcome.

You ain't akidding, there.

If your soundcard does not feature ASIO then you can add it with the following freeware.

ASIO4ALL - Universal ASIO Driver

Google ASIO4ALL to get more information. I intend to try it at a later date so this post is not a personal recommendation but just information that might save a DIYer from an upgrade. Your risk of course. :coffeedrinker:

Regards, Marshall.

Thanks, Marshall, I'll do a little research on that.

I guess measurement is going to be more than slinging a mic and stand in front of a driver and hitting the "enter" key. That's OK, I was a Windows 95 baby and am used to the software pitfalls. I've been looking around at some different gear and the M-audio Fast Track is going to be hard to better. It has the necessary phantom power, the correct interfaces, and seems to be priced right. I believe the Behringer ECM8000 electret condenser mic is also on the list. I happen to have three scrounged desktop boxes and might just make one of them a measurement only system. I'll have to mess around to see if the motherboards support a newer Windows OS.

Keep the good stuff coming guys, I've already learned a bunch and look forward to more.

[Panomaniac]

If you want the lab results at bargain prices, get a calibrated mic. That's something that is hard to do yourself.

[bfish]

Another remembered passage from the "Book of Hard Knocks";

Most PCI cards and onboard laptop cards operate on the consumer 0.7V = 0dBv standard, but all the software I've used operates on the pro 1.0V = 0dBv standard. Sometimes you can tweak the levels to work OK, sometimes not. The external cards I've investigated all use the pro std, so no problem.

[sba2]

Looks like I have some (a lot of) issues to sort out with the software, my soundcard, and its drivers. When I figure it all out, I?ll add it to ?The Book of Hard Knocks"

It turns out that all of my HOLM FR plots were contaminated (at 2 msec) by a very poor soundcard calibration file, which turned out to be not really a calibration at all, but a feedback loop between the ?line in? and the ?output?.

The problem went unnoticed for a while because I didn?t closely examine the IR plots, and didn?t even know what a proper calibration plot was supposed to look like.

The HOLM software has no documentation to help with such calibrations? hell, it even told me that I had made ?valid? cal measurements.

I only discovered the feedback issue by reading about it in the REW manual.
Disabling ?record monitoring? usually gets rid of the problem, but, I haven?t yet been able to access this function, either through the card directly, or via its latest ASIO driver, or via ASIO4all, or via Windows (XP).

I bought my card years ago simply to get high quality sound from my laptop, as well as to occasionally transfer something from vinyl (my laptop?s input is only mono mic). Recently I had recalled seeing the card on ARTA?s list, so I didn?t think twice when I began measuring. At the moment I would give the Echo card a ?no go? for acoustic measurements.

By removing the Echo card and using my laptop?s piddly internal SigmaTel card with MME 16bit/41k , I was able to get a proper cal file, but I?m not yet confident of that whole scenario. The latency with the SigmaTel was pretty high, too

I?ll probably have to get a usb card.

*******************

RightMark is pretty cool ! Right away I discovered some bad cables and tossed them. Too bad the ASIO version cost US 110 !

[Panomaniac]

Man, that's too bad. But we all go thru stuff like that. It takes a while to get it all running right.

I always check the loop measurement when I calibrate. I.E. , after all is set up and calibrated just run a measurement of the loop to see if there are any artifacts. The FR and impulse should be close to perfect. If they aren't, you've got problems.

[sba2]

At the moment I would give the Echo card a “no go” for acoustic measurements.

The ECHO soundcard was not the problem.
HOLM and REW were also not the problem.

I’ve narrowed it down to my Fujitsu laptop (XP). Actually, I have two identical Fujitsu laptops with Echo soundcards. Both setups could not produce a proper soundcard calibration measurement in either HOLM or REW. Analyzing the two setups with RightMark, I discovered “very poor” FR and IMD--

http://img819.imageshack.us/img819/1...cardsonfuj.png


I then tried the two ECHO soundcards in an old Dell laptop (XP). Voila ! Normal soundcard readings--

http://img198.imageshack.us/img198/4...ardsindell.png

And…. I was able to make proper soundcard calibrations in both HOLM and REW--

http://img580.imageshack.us/img580/8048/holm48cal.png

http://img94.imageshack.us/img94/2878/rew48cal.png


****************

I don’t know whether it’s a software or hardware issue in the Fujitsu, but the troubleshooting ends here. I’ll use the Dell & ECHO card.

I need to get my life back….I’m not going to be victimized by this!


*****************

A few observations on the free software I’m using:

HOLM detects & provides a choice of interfaces (ASIO, MME, DIRECT),
REW does not (I didn’t see any interface options, and was not sure which was used).

HOLM will sample to 96khz and above,
REW only to 48khz

HOLM plots a max of three measurements,
REW plots more… It also has waterfalls, spectograms, and other cool stuff.

********************

I always check the loop measurement when I calibrate. I.E. , after all is set up and calibrated just run a measurement of the loop to see if there are any artifacts. The FR and impulse should be close to perfect. If they aren't, you've got problems.

Good advice!

[bfish]

RE sampling rates and connectivity:

44/48kHz sampling rate is the minimum needed to sample 20kHz audio. Your hardware needs to support simultaneous in/out (duplex) at that rate at a minimum. Most cards advertise their maximum one-way sample rates, but you need to read the fine print to see what rate they'll duplex at, it's usually half of the maximum. If the duplex rate isn't specified, move on.

Sampling rate x bit depth x2 = the data stream (in bits per second) you need to be able to run from the card to your computer.

USB 2.0 offers 480 Mbit/s, a fortyfold increase over 12 Mbit/s for the original USB 1. 1394A (Firewire, ilink) ports support 400Mbits/s, making it comparable to USB 2.0. For desktop PCs, the low cost of a USB 2.0 or a 1394 PCI card (~ $40) makes it a no-brainer. If your laptop already has the port, pick an external card that can take advantage of it.

If you don't care to do the homework before buying your gear, you can join the guys that buy some crap, plug it in, run some tests, then splatter the web with their "scientific data", or at the least can't figure out why their results are unexplainable. Otherwise, read on...

Once you're setup with your gear...

...you should test it if you want any confidence in your measurements. Why go to the trouble of calibrating a mic if the rest of the system is unknown? Fortunately, there's a simple and freeware option for detailed souncard benchmarking called Rightmark audio analyser. I use it regularly, and it has exposed connection and settings flaws that I'd have overlooked many times;

Download. Audio Rightmark

Benchmark setup is stupid easy, just loop the soundcard output back to the input and run the level setting utility. Then click "run tests", and a few seconds later you get a detailed text/graphics report of your system performance.

For ASIO support you'll need to buy the Pro version. Well worth it if you're serious about accuracy (and your card is running ASIO).

An unintended extra function is the ability to insert nearly any audio device in the test loop so you can benchmark it too.

Since an excerpt from my post above seemed to strike a couple of nerves here, I thought it only fair to revisit it. I concluded that the "offensive" portion was taken out of context, and when standing alone might be offensive to some, but when considered in context was really what I meant;

...If you don't care to do the homework before buying your gear, you can join the guys that buy some crap, plug it in, run some tests, then splatter the web with their "scientific data", or at the least can't figure out why their results are unexplainable. Otherwise, read on...

Once you're setup with your gear...

This is a standard that I hold myself to as well. When I first bought some gear and ran some tests, I thought they were great. Later I found (by benchmarking and other self-checks) that they weren't. More gear, software, and fixing later, again I thought I was in good shape. Wrong. Now, I think all is good again, but considering everything, I'm very glad I didn't post my incorrect data.

Posting/not posting your data is really irrelavent though. My goal in that post was to inspire others to do some self-checking so they would end up with better results. Merely saying offhand you should self-test is likely ignored. The shock factor of my "offensive" quote, not so much.

So, if my being a little coarse results in even one DIYer having second thoughts that prompt repairs and better results, so be it. The post stands as written.