Sean, you've obviously gone to a lot of trouble to research this mod and to write it all up. Sharing this with us via this medium is what being an audiophile is all about. I, for one, deeply appreciate it and also try to assist others whenever possible. Kudos to you!
23 responses Add your response
Sean, to offer a "thank you" seems unworthy. I have had the privelage of reading countless informative posts during my stint on Audiogon. This has to be the most instructive, informative, well researched post I have had the pleasure to read. You certainly deserve something above a commendation. THANK YOU! Joe.
That was Awesome! I did a test using my new Terrasonde The Audio Toolbox, albeit in nowhere near lab conditions, nor do I suspect the audio toolbox qualifies as lab grade equipment, and my data roughly agrees with your data above.
I thought that some of the bass roloff had to do with the C weighting of the meter, do you have any insight in to that?
Folks, i'd like to take credit for all of the info presented above, but i can't. I simply combined information that i've picked up from various sources and presented it together as a package for convenience sake. Some of the websites that i garnered all of this "well documented" info from are no longer operating. Since that is the case, i can't forward you to some of them. If your truly interested in more background on the data presented here, email me directly and i'll forward what i have. Take my word for it though, all of it presented here has been well tested and is quite accurate when used in the manner specified. Keep in mind that when comparing readings between one of these SPL meters and a computer program, you are at the mercy of the sound card and microphone used for the puter. As such, most "sound blaster" type cards along with all of the "generic" versions are HORRIBLY non-linear. Combine this with a non-calibrated microphone and you would be COMPOUNDING the results, giving you nothing but error filled data at best. Glad you "guys" ( and gals too.... ) liked this and can put it to use. Sean
I have found two other sources of corrections to the RS meter that are very similar in the bass and midrange as reported by Sean. However, these corrections indicate that the RS meter is also rolled off in the higher frequencies. My own testing (I did it once all the way to 20k hz) with a RS meter and Stereophile test CD on my system seemed to verify the following:
Add to your reading:
20k = 11.2
16k = 8.5
10k = 4.4
8k = 3
5k = 1.3
3150 = 0.5
2500 = 0.3
2000 = 0.2
1600 = 0.1
1250-200 = flat
160 = 0.1
125 = 0.2
100 = 0.3
80 = 0.5
63 = 0.8
50 = 1.3
40 = 2
31.5 = 3
25 = 5
20 = 7.1
Sorry, I wrote the above on a sheet of paper a year or two ago so I don't have the original source. The bass region is the most important to me. I never check beyond 200 Hz anymore. Besides, a tweeter chirping at 75 db is annoying if not a little scary. Roland
Lightningman, thanks for posting your results. Your comments about rolled off highs were echoed in some of the original info that i culminated this from. The change to C12 addresses that problem. Due to production tolerances, some mic elements and / or circuitry do have slightly different measurements. Lowering the value of C12 helps to restore more uniform high frequency coverage to the unit even if you happened to have one that "nose dived" from the factory. Don't get me wrong, i'm not saying that these mods turn the RS meter into a piece of laboratory grade test gear, just that it will alleviate the need for factoring in major corrections in terms of non-linearity. In plain English, it will give you more of a "what you see is what you get" type of reading. As stated previously though, the biggest problem with the RS meter is the factory mic element and how they have it mounted. Doing the above listed mods and then substituting a calibrated microphone will give you excellent results. After the mods have been done, the electronics themselves are basically flat from below 20 Hz to appr 20 kHz. Any major variations in the response would be due to the microphone. I would expect pretty linear results using the factory stock mic with the modified electronics to give solid results from appr. 20 Hz to 10+ kHz as long as your taking your measurements directly on axis. Some may extend beyond 10 kHz with good accuracy, but that would just be icing on the cake. Sean
For those who are lazy, or are arithmetically challenged, the Rives audio Test CD has a set of tracks that are compensated for error of the Radio Shack meter.
For those who are really, really, lazy, get a spectrum analyser and calibration mic.
Sean...Why do you think that the meter is made with so many wrong-value capacitors?
Bluemax: I do not know if the 33-4050 is built / designed in the same fashion. Given that RS is primarily interested in increasing profits via cosmetic changes rather than re-designing products, my "guess" would be that it is probably quite similar.
El: I have no idea why this or many other products are built and designed the way that they are. My guess is that many manufacturers simply trust the engineering department to get things right without ever really testing any prototypes and market it as is. This would explain why new models come out quite frequently, why there are Mk II and Mk III revisions and why products are recalled by those same manufacturers.
What i can't figure out is how there is always enough time and money to do something twice ( or more ) in order to correct problems that were overlooked in the initial design phases when there is never enough time or money to do it right the first time. Like the song says, "Kinda makes you go hmmmm"...... Sean
Sean...In my old business, where I was responsible for test design (software definition) the problem was like this...
1...You must get the software delivered by this date. Don't worry about making it perfect.
Then, after the SW is delivered, and some problems or potential upgrades becomne evident..
2...What we have is working. Why change?
In between there are about 100 millisconds of time to make changes.
Sean, your attention to detail and in offering this up are appreciated. A perhaps dumb question, though. If there are this many tweaks needed, why doesn't The Shack just build it right in the first place? With the "fixes", is it still a relative bargain in comparison to a meter one could buy without a need for "fixes"?
The parts to modify this meter are quite inexepensive, making it a good deal even after investing the money. The fact that RS is convenient and many people already have these units makes such an approach even more acceptable. In comparison to many other SPL meters that cost more money ( $70 - $100 ), the modified RS meter is more accurate. One can find a very good quality SPL meter on Ebay for not that much money, but the calibration of such a unit may be questionable due to having a "rough life".
As to why they haven't built it right, maybe they have. I know the original analogue meter circuitry remained pretty much the same for a couple different incarnations of the meter and the digital meter basically copied the existing circuitry ( from what i've been told ). Like El stated above though, "if it works, don't fix it". Their circuit worked, it just wasn't as accurate as it could have been.
For more info on this and other RS SPL meters, try taking a look at Eric Wallin's website. He's got mods for a few different RS SPL meters along with several other audio related projects. Sean
Thanks Sean, I guess that would explain why my subs measure down 10db or more at the extream low freq's yet still shake the room.
I did raise their output level at one time to bring it into line with the higher bass freq's and found this to excessive to blend smoothly.
Your chart confirms what I was hearing (and feeling) in my room.
I just purchased a 33-4050 model (7 scale selector switch, Tandy catalog Q1461 here in Australia), and I have not seen any other models. I opened it up, and just from the description of the component values, I'd say it is different from the model described above. The original values I see are:
C1, C2, C3: 1 uF
C7, C12: 22uF
C15: 100 uF
C5, 13, 14, 16, 17 are electrolytics on the component side of the board but I forgot to get their values before putting the case on again, mainly because they were not mentioned in the ones to modify.
C6, 8, 9, 10, 11, 18, 19 are surface mounted tiny capacitors on the rear side of the board.
So it would be goot to know if anyone has information on the calibration of this model. It does have a trim-pot called "cal" for calibration, but no instructions other than to take it to an expert for that.
Bob: The calibration pot simply allows one to adjust for factory tolerance at one reference frequency and spl. Once that is adjusted, the non-linearities of the circuitry / mic still come into play at other frequencies and spl's.
As to whether or not these mods apply to the newer digital meters, i have no idea. I always prefer to use "analogue" ( i.e. "ye olde school" ) meters as they tend to be a little more revealing of trends, etc... They might not give the absolute resolution of a digital read-out, but at the same time, we are basically looking at the average frequency response here, not correction down to the .1 dB level. One could only hope to achieve anywhere near that type of linearity in-room and if they did, i will guarantee that there would be no sense of "ambience" what-so-ever. Sean
So now I have both the analog and digital meters, and they both read the same with every frequency and any test CD, therefore, the corrections apply to any RadioShack meter (and I am only going to keep the digital). Moreover, the capacitors in both are different than the original post, therefore, it's currently questionable whether the surgery will achieve the desired results. BTW, thanks for the pointer to the Rives CD, which I also obtained.
So I believe the Rives CD indeed corrects the curves of both of these meters as advertised (I can only really confirm this by ear at the very low end of the spectrum, < 30Hz, where in fact 20Hz and 25Hz tones do indeed sound much louder than measured with the uncorrected tones), however, there is a major difference and issue when compared with the Stereophile CD... Namely, I get smoother readings with Stereophile's warble tones than Rives' pure sinusoidal (uncorrected and corrected) tones, and in fact, if I move my head around I get cancellations in lots of frequencies with the Rives, and this is verified with the meters as well (wild fluctuations, especially at the reference tone of 1kHz - as much as 10dB).
So despite the fact the Rives CD corrects the errors in the SPL meter, positioning of the mic is extremely critical, down to fractions of an inch, in my room.
So, the question is - in general, are warble tones the better approach to measuring level, or not, and why? Any idea why Rives didn't use warble tones? As good as Rives' idea was to correct the meter's curve, I think they missed the mark by not using warble tones, and in that case, we can just play the Stereophile warble tones and adjust the readings manually by applying the diffs in the original post herein.