My Computer System, Vintage Audio for New Tech

My main system is in a constant state of flux, it was different this afternoon from what it was in the morning and different this evening than it was this afternoon. So, I'm not going to list it just yet. My computer system, however, has been largely stable for many months now.

I'm running Winamp 5.24 with FLAC for the most part and the MAD plug-in for MP3s going out to a Sound Blaster Audigy sound card. This feeds a Pioneer SX-727 that then powers a pair of Dynaco A-25s. Sometimes I run a modified Sonic Impact T-amp (powered by a 20 amp Astron regulated PS) off of the 727's pre-amp section. There's a powered DIY ported 10" subwoofer crossing in at 60Hz.

I very much enjoy listening to this system. It's not "audiophile" by any means (that's what my main system is for), but it's just plain very musical. I can listen to it by the hour with zero fatigue and if anything it sounds even better after several albums.

Often I've wondered why this is so. As I had some spare time this weekend, I took a little time to do some quick measurements and this is the resultant graph:


Each of the X-axis numbers correlates to the frequency (Hz) following:

The downwards slope is a major clue as to the musicality of this setup as it's nearly identical in slope of the graph published in Brüel & Kjær's psychoacoustics study.

This is the frequency response determine to sound most natural/correct to listeners:


The only weakness aside from a slight lack of transparency in this system is that the tweeter rolls off around 15kHz, but that's very minor considering the overall musicality of the tone.

- JP

someday you'll have to come back over and measure the bedroom system with the T-Amp and ADS L-810's. from the standpoint of pure listening pleasure, its hard to beat this system. i am curious to see how close it comes to the Brüel & Kjær slope.

Measuring a system can be a fairly lengthy process as a multitude of tones are required and several seconds spent on each to get an accurate reading and record that measure. It took about an hour to put together that graph, double check the measures, and apply the meter correction values. I believe the Radio Shack in Auburn (on the right headed South to the Goodwill from Hwy 18) is closing out, so maybe you could snag a SPL meter from that store on a discount? I can set you up with the proper meter correction values so you can do this on your bedroom system and even your big system. I suspect the ADSs are a better match to the slope as I don't believe they roll off as fast on the top end (or bottom end of the A-25s without the sub).

- JP

precisely why i want you to come over and do it. plus i want to get your take on the sound.

you're a lazy dog dingus

Indeed. Measuring a system is just a simple matter of sitting down and writing down what the SPL meter reads then add the correction value. Not technically difficult at all, just time consuming.

These are the correction values as posted on Audiogon and verified against several other independent sources:

You need only download one of the many computer software programs that can output WAV files of specific frequency test tones and burn them to a CD.

Now, get measuring.

- JP

I've come to realize that I neglected to mention one of the most important components in this computer system, the power conditioner. I'm currently using a Topaz Line 2 Laboratory Grade Power Conditioner that I picked up a hamfest a couple of years ago for a few bucks (intermittent operation fixed by 5 minutes with a soldering iron. To me, this is an absolute must due to the line noise generated by the computer. To me, it provides superior midrange clarity and, as the reviewers like to say, 'a dark velvety background against which the music is highlighted.'

It's not audiophile approved nor does it carry a Monster brand tag, but this little no nonsense unit will not leave my computer system. Oddly enough, though, it made no difference on my main system, so I'm fairly certain it takes care of computer noise alone in this instance.

- JP

check the stickied thread in the speaker forum, you can make a much more dynamic measuring system for not much more than a ratshack spl meter. measuring is much less tedious with the proper tools.

Just the other day I discovered Room EQ Wizard as hosted by another fairly new forum, Home Theater Shack. Just playing around with it reveals some great possibilities including automated system measurement and the best feature -- it's free.

Yesterday, I downloaded and installed the program and plugged in the calibration sheet they had for my relatively new RS digital SPL meter (cat. no. 33-2055). It's accurate and done at a resolution of 1 Hz, but only covers from 10 Hz to 105 Hz. Above that, you're on your own. I then set up my SPL meter on a tripod where I normally sit with the computer chair positioned awkwardly in about the same place and went through the calibration sequence the software requires. Here are my preliminary results:

10 Hz to 105 Hz @ 1 Hz resolution corrected meter values with no smoothing:

Yes folks, that's within +/- 0.5 dB including that little peak at 25 Hz. You can see the woofer unloading some around 25 Hz due to the ported design and I disabled the parametric equalization plug-in I normally use to goose the deep bass just a bit at a little sacrifice of tight response. I'm surprised at how relatively smooth this is considering this is a DIY sub. The speakers (Dynaco A-25s) are crossed over at 70 Hz and I have adjusted the levels by ear a bit over time. [BRAG]Looking at this graph, the crossover point has a barely measurable difference of ~0.2dB. I've never claimed to be a golden ear, but man alive I must've been incredibly lucky to get that close of a match up just by ear and doing so while crawling under my desk to an odd place trying to remember just about how much of a difference was needed. [/BRAG]

Here's the rest of the response measured at 1/60th octave spacing, no meter corrections implemented nor curve smoothing:

You can very easily see the slight rise before then sudden crash afterwards caused by the cheap opamps used in this meter and thus nulling any chance at accuracy above 10 kHz. It's a measured phenomena and modifications are published online to fix it, but then the bass meter corrections (most important due to the very unflat response of this meter down low) become worthless. So, I'm definitely going to need to look into a better mike that comes with a good calibration sheet.

I don't know what the rest of those little dips are about. I was in the next room to avoid adding noise to the measures and those 1/60th octave dips were very audible and just sudden dropouts. I'm not sure what's causing it if there's something goofy going on with the drivers. However, the power conditioner on my receiver has been acting up in this hot weather dropping out for a couple of seconds and then coming back on. It's most likely that this is the cause of those little dips.

Interestingly, the "Law of 400,000" may be part of the explanation for why I find this system so musical. It's roll off points are 25 Hz and 16000 kHz (top determined by ear though there is still some reduced output up to 18.5 kHz) and when run through that theory, they multiply to 400,000. Hmm, some food for thought.

All in all, this software impresses me and I'm amazed at just how well my system is performing. Of course, I listen and measure in the near field so room effects are mostly avoided.

- JP

Hmm, funny, thinking through this trying to categorize all that could be possibly leading to an incorrectly flat reading (never happy am I with things as they are), an old incident I had with another A-25 owner came to mind. Some guy very loudly proclaimed that the flat setting on the Dynacos was very much not flat and that the only way to achieve good sound was to bypass the control to brighten the tweeter to a "flat" level. Of course, being the combative spirit that I can be I stated my case that the flat setting actually was flat, that the only issue was a 16kHz roll-off of the tweeter, and that messing with the control was going to disrupt the balance of the speaker. This gent then promptly launched into slamming me for my poor hearing and took the defense that as a musician he knew far better than I what a flat response sounds like and that I was very much wrong. Weeelll, looks like I've been vindicated, I was correct at least about the tweeter being a level match for the woofer with the stock levels switch.

Yes, it is a good feeling to be found right after taking a heaping of abuse. Shame I can't contact the person and gloat some.


- JP

Post scriptum: I personally believe that musicians possibly are a poor choice for speaker design because they are accustomed to the sound of the instrument/music with it in their hands while amongst the band and not from the perspective of the audience where the rest of us are familiar with how music sounds. Of course, that's a matter for discussion and likely argument elsewhere.

Well, I figured out the main cause of those random one frequency dips that plagued most of the measured frequency response, my Java Runtime Environment was one version older than recommended for the software. Evidently the issue caused by this version difference is manifested in occasional brief drop-outs of the sig gen function. Beyond that, a better mike with calibration sheets is definitely needed to settle the 3.5kHz gradual rise and the 10kHz crash. There simply are no decent correction factors out there that are universal for this model of budget SPL meter. Ah well.

- JP