The analog RS SPL meter is an inexpensive instrument that can beused to measure the frequency response of audio systems. Two modifications are described that improve itsaccuracy at the upper and lower extremes of the audio spectrum,and make it more convenient to use.
The analog RS SPL meter (cat.no.33-2050) is an effective,useful, and cheap tool that should be in every DIY speakerbuilder's arsenal. When coupled with a test CD or other signalsource, accurate anechoic-like bass measurements can beperformed using this meter in a "close-miked" setup; andin-room response can also be measured easily.
In many ways the analog RS meter is better than the digitalmeter RS offers: The resolution of the readout is higher(fractions of a dB, while the digital meter only reads to thenearest dB), the analog meter movement provides moreinformation when analyzing dynamic signals, and finally it isless expensive than the digital meter. If the digital meterdispensed with the range selector switch and read fractionaldBs, I would consider purchasing one, but with it, I considerthe analog meter more useful and the better value.
The electronics of the stock analog RS SPL meter rolls off thebass (many poles) somewhere around 35Hz, and rolls off thetreble (one pole) at 10kHz.
This is roughly comparable to the "C" weighting curve:
20HZ =6.2DB 25HZ =4.4DB 31.5HZ =3DB 40HZ =2DB 50 =1.3DBThe frequency response mod described makes the *electronics*flat over the entire audible range for the "C" setting of theA/C switch. After perfoming the mod, direct meter readingsshould be accurate for LF SPL measurements down to 20Hz or sousing the internal mic *without* using any correction factorslike those listed above -- in other words, pure WYSIWYG forLF readings.
63HZ =.8DB 80 HZ =.5DB 100HZ =.3DB 125HZ =.2DB 160 =.1DB
200HZ to 1250HZ Flat
1600HZ = .1DB 2000HZ =.2DB 2500HZ =.3DB 3150HZ =.5DB
4KHZ =.8DB 5KHZ =1.3DB 6.3KHZ =2DB 8KHZ =3DB 10KHZ =4.4DB
12.5KHZ =6.2DB 16KHZ =8.5DB 20KHZ =11.2DB
The HF response of the stock meter is another matter. Thepoor "roller coaster" response of the meter in the HF region (as seen in the instruction manual for the meter) iscaused (this is in fact informed speculation, butspeculation nonetheless) partly by the physical mounting of theinternal mic and partly by the mic module itself. The externalmic mod (coupled with the frequency response mod) *should*correct for this if you use a Panasonic mic module mounted inthe end of some tubing ala the MMII. I have no way of checkingthe frequency response of the mic element, but I have used theadded external mic input to check the frequency responseof the modified electronics. Basically, if the external miccartridge you use has a flat frequency response, then themodified meter electronics will be able to keep up with it andnot introduce any additional response irregularities.
Recommended external mic cartridge:
Panasonic electret mic capsules from Digi-Key; 1-800-344-4539. They carry several, but I think you want WM-60AY, Digi-Key part P9959-ND, $2.24 in quantity 1-9; $19.88/10.This is the same cartridge used in the Mity Mike II, andtypically has a +1dB hump around 10kHz, but is very flatotherwise.
Both mods are simple and may be free if you have some caps inyour junk box and some plumbing supplies in your closet /garage.
Click here for the schematic.
Instructions are as follows:
- To extend the LF response: -
CHANGE (either replace or solder in parallel to the old value,your choice, use electrolytic caps):
- To extend the HF response: -
CHANGE (replace, use ceramic or NPO cap):
Case Disassembly:
After this mod, I (or the meter, rather) exhibited someinstability (hf oscillations). Making C12 12pF cured this (itwas 10pF in the inital mod). If you experience similarinstability (needle jumps around, stays pegged on the lowerranges, is influenced by the position of your hand over theneedle area) increasing C12 is the way to fix this. Be awarethat the larger this cap is, the lower the hf rolloff will be,though of course the meter is useless if it is not stable.
Regardless of whether you do the external mic mod or not, afterthe basic electronics mod is done, you will notice that themeter needle pegs a couple of times at turn-on before settlingdown. This is normal since the circuit now has many very longtime constants. As I stated before, on the most sensitivesettings, the meter will seem somewhat jumpier. This is due tothe amplification of low frequency data such as wind and casethumps. In fact, this is an indication that the mod is workingcorrectly. My meter, on the 60dB setting, registers my movinghand from several feet away. And this isn't the breeze from myhand "blowing" on the mic, this is my hand moving in a rythmiccloser-farther-closer motion relative to the mic at a couple ofHz rate, Xmax ~3" p-p (sorry, ~1.5" peak :-). It is thesensitivity of the mic cartridge to very low bass in this testthat lead me to believe the cartridge is probably pretty flatto a few Hz.
I just want everyone who is contemplating performing this modon their meter (the frequency response correction mod, with orwithout the external mic mod) to be well aware of the behaviorafterward. I am quite happy with the results, but if this sortof behavior (i.e. the double pegging and settling at turn-on)drives you crazy, perhaps you should decide against surgery andgo with the old therapy, i.e. equalization charts.
The SPL meter electronics (including the meter needle itself)now respond in the following way (regardless of the rangeswitch, external mic input, C12=12pF):
"Fast" response switch position:
+/-3dB: 6Hz to 50kHz
+/-1dB: 28Hz to 23kHz
-1.2dB @ 20Hz
-0.8dB @ 20kHz
Flat otherwise.
"Slow" response switch position:
+/-3dB: 4Hz to 50kHz
+/-1dB: 8Hz to 23kHz
-0.2dB @ 20Hz
-0.8dB @ 20kHz
Flat otherwise.
The "Slow" response setting looks like the way to go foraccurate LF measurements. Based on the -0.2dB @ 20Hzfigure, we can now pretty much ignore the frequency response ofthe electronics for most LF measurements.
Also, I assume I would match the type of caps ie. electrolytic, tantulum, mica etc....
A: The idea is only to lower the poles so that they do notsignificantly alter the meter response at the frequencies ofinterest (>10Hz or so). So it is really just a matter ofincreasing the capacitance of C1, C2, C3, C4, C7, C8, C9, andC15 until this condition is true.
There is no need to remove the original caps, and no need tofind special values to make the resulting parallel capacitancevalues exact. For example, soldering a 10uF in parallel with a1uF only makes the resulting value off by 10%, and this isusually within the tolerance of the 10uF in the first place!
The resulting pole may be 10% lower, but all the better. Ifyou use values that are significantly larger than those that Ispecifiy for the signal path, the meter will take a lot longerto settle down at turn-on, which might drive your crazy if youare sensitive to such things. Mine takes about 3-5 seconds tocalm down after turn-on, and is stable during and afterchanging the range switch. It is very sensitive to wind andcase shock after the mod, which makes sense if you think aboutit.
C12 is the only capacitor that needs to be replaced, since itsvalue has to be reduced.
If you find that your meter is unstable after the highfrequency mod (needle moves erratically, pegs or stayssomewhere in the midrange even with no sound input), try addingsmall values of capacitance (2pF or so) across C12 until themeter stabilizes. This will make your meter less accurate inthe high-end (the high-end response will start rolling offsooner). Better to err on the side of stability, since themeter is useless if it is not stable.
Use electrolytics in parallel with the electrolytics (which isevery modified cap except C12). I used some that I scroungedfrom an old VCR. These caps were very small for their size,making them an easy fit. Any voltage level at or above 10V isfine. Observe polarity when installing them (s/b same as theoriginals).
A ceramic or NPO will do for C12. R/S sells an assortment ofsmall value caps (1-30pF) that will do nicely for C12, and willallow you to find the critical value that will yield bothstability and maximum flat hf response. As I stated above,starting with 12pF would probably be wise.
Q: You had previously written :The easiest way to change these caps (except for c12, which must be replaced) is to leave the originals on the PC board, and solder the new caps to the back of the board, where there is plenty of room.
I assume from the above that you meant to add the new parts to the old ones (in parallel). So C1 goes from 1 uF to 11 uF not to 10 uF. This looks safe enough since you're just lowering the pole frequency even more than if the old cap were removed.
A: Right. The tolerance on electrolytics is usually rather wide,and all we are trying to do is lower the poles below some(rather arbitrary) frequency (I picked ~1Hz).
Q: FYI, I made the meter mods last night - just the 9 caps.So far as i can tell, it's acting as you described. The needle jumps a lot on power-up and when changing ranges, especially in "fast" mode. Hand motion is detectable. Next step is to run some tests on it and compare it to previous results.
Also, I think I'll keep it as an internal mic device only if the existing mic module seems to be pretty sensitive. It's probably not flat all the way to DC, but there's a decent chance it's flat to at least 20 (the Panasonic module I bought from DigiKey is supposed to be rated 20-20k; I wonder how far off it is below that ?) and that's good enough for me. Then I'll use the other mic for something else I guess.
A: Mine pegs twice on power-up, but then settles down. Switchingranges doesn't affect it very much when in the "slow" mode,which is where mine stays now.
Note that the "slow" setting will give much better lowfrequency readings, -0.2dB @ 20Hz vs. the "fast" settingresponse of -1.2dB @ 20Hz. This also keeps the needle fromjumping around during range switching.
I guess you changed high frequency cap C12 also. Did you makeit 10pF or 12pF? I had to go to 12pF for stability reasons,outlined in my second post. If *holding* your hand near themeter area causes readings, you have oscillation and shouldincrease C12 until they go away.
Adding the external mic option is a snap. Have you consideredthis? I think that one of the reasons the HF response sucks sobad (in the RS manual, the roller coaster action in the highs)is due to the funky mounting of the internal mic. That hugeplastic flange (I thought it was aluminum until I took thething apart; some things I would rather not know :-) and thatfunky "PC board" screen *have* to be altering the responsecurve in a major way. An external mic (like the wand-type MMIIcartridge at the end of a tube type thing - without all thefancy electronics) would make this unit accurate all the way upto 20kHz, where the response of the electronics is down only-0.8dB (with C12=12pF). Getting my body out of the measurementarea and the meter nearer to my eye would be added bonuses(boni?).
HF response aside, I bet dollars to donuts that the miccartridge itself is, as you say, flat to at least 20Hz based onthe "waving ~3Hz hand" experiment. Be nice to have some way tocheck this. All we need is a calibrated sub good to 3Hz (nextproject...just kidding) or a specral noise source that doesn'temploy speakers (maybe released compressed air?). Till then,I'll trust that the bass readings I get are fairly accurate.
Q: Just to clarify, am I correct in assuming that this mod will correct the deficiencies of the internal mike, but that the impact of C/A weighting will still have to be cranked in the the readings?
A: Just the opposite! After the mod, the electronics will be flat(including the meter needle itself) but the internal mic willcontinue to contribute its poor hf response.
The internal mic seems to be fine for bass readings, andcontinues to work well for those situations when you needcalibrated "A" weighting readings (rock concerts and jet planesand such). The external mic mod should easily get aroundthe high-end response problem, and also will get your body outof the measurement field and place the meter nearer to youreyes where you can more closely scrutinize the trembling littleneedle.
Q: I've just found your instructions for modifying the meter's response and was wondering if this also affects the signal sent to the output jack?
A: This mod should correct for the output jack response also,since the meter amplifier ic actually feeds the output jack.So you should do both the hf and lf mods if you want thefrequency response of the signals at the output jack to beflat.
Q: Have you done any tests on the output? I was wondering if it was constant if a part of the signal drops off the current range (eg range = 90dB, but at say, 500Hz the signal is only 80dB)?
A: I have checked the fr of the output jack, and the response ofthe electronics after the mod follows the actual meter needleresponse over the entire 20Hz - 20kHz range. Everything*after* the built-in mic in the signal chain is flat over thisrange, including the meter response and the output jackresponse.
Q: Will the output jack give a sensible (linear) reading if I'm measuring a signal where 1kHz reads 90dB, 10kHz reads 80dB and 100Hz reads 100dB with the range set to 90dB?
A: I know the meter scales the reading (ie switching the 80dB range in this example would make the output higher) but would signals that are out of the meter's current range be output at the correct relative value?
Since the single transistor pre-amp circuit is class A (and Iassume the op-amp used for meter drive / output jack drive isclass A as well) then signals at the output jack should belinear for input signals outside of the meter range. Thecaveat here would be for signals that are so large that theycause clipping, and thus overload the op-amp. Smaller signalsshould be no problem as long as they are not so attenuated thatthey are buried in noise.
