Fs Measurement - Baffle, low ohm amp?

[kimbo]

Hi all,

I am having problems reconciling the Fs that I am actually measuring with the figure I have seen published for the vifa p22wp-01 8 inch drivers.
I have been measuring them loosely suspended vertically in the middle of the room facing diagonally across the room, without a baffle. They have been broken in for 20 or 30 hrs.

The three or four books I have consulted, including "the loudspeaker design cookbook" all recomend a similar method, and I have been feeding the speaker from the signal generator via a 1000 ohm resistor to create a constant (almost) current setup. The figures for Fs that I am getting are a few Hz higher than the 29.5 quoted in several sites and the 31 in the magazine article that I spotted.

Anyway, I sent off an email to the mag and the response was that it was measured using a baffle. I have not seen this before, but when I looked into it I noticed that the 1971 paper by Small p.275 does say that a baffle can be used, but that it will change the measured Fs. The response from the magazine (see below) also stated that an amp with a low resistance less that one ohm should be used. This sort of makes sense to me.
Anyway, now I am really confused........
* Do I use a baffle? or suspend it loosely or mount it rigidly?
* Does using a 1000ohm resistor change the measured value of Fs? (are the books wrong?)
* How do speaker manufacturers perform these measurements?
Can any members shed some light on this or have any comments on Leo's reply?

My original email to Silicon Chip
Hi, some time ago, after reading your article on the JV80 speaker system, I purchased the speaker kit with the intention of constructing my own boxes. I have finally begun, and for interest I thought I'd check the resonant freq (Fs) of the drivers. I was most surprised to find that they all measured between 34.5 and 35.4 Hz. Your article gave it as 31Hz and a search of the net came up with 29.5Hz. I am just wondering how sensitive the box design is to such a variation in Fs and would like to know how it was measured for the October 2003 article.

Leo Simpsons reply
There are a number of significant points about the measurement of speaker resonance. First, the speaker should be mounted on a baffle, typically 1 metre square. Second, the speaker needs to be driven from a low impedance source, less than 1-ohm. Driving it from a high impedance source will typically raise the measured resonance by a few Hertz. Third, the quoted resonance figure is not an absolute and has a tolerance which could be plus or minus 5 or 10% (whatever the manufacturer specified).

The resonance could also be expected to drop by a few hertz after the speaker has been used for a few hours or more.

Fiinally, we would not expect the speaker box performance to be affected by a small shift in the resonance.

[SY]

First, manufacturers' specs are often works of fiction. Believe your own measurements.

Second, a baffle will not affect Fs by a large amount compared to free air- maybe 0.5Hz at most. But... when making free-air measurements, make sure that the speaker frame is clamped or otherwise held rigidly. If you let it swing free, it will react to the cone motion and affect the measurement significantly.

The current-source method is the best bench way to measure Fs. If you've got an FFT setup capable of low sampling rate and long sample time, you can get good results that way, too.

[richie00boy]

How have you broken the speakers in? It's best to use some lowish frequency noise or sweeps as the suspension needs to work.

If you mount it on a baffle then the Fs changes and you need to apply a correction to get the true Fs. This is covered in the Loudspeaker Testing section of the Loudspeaker Design Cookbook.

Using a current source is how I measure my speakers and I have not heard of this method raising Fs before. I prefer this method as it makes the peak more pronounced (but not in relative terms)and thus some points are easier to measure.

One issue could be that you are clipping the signal generator and this will screw up your measurements. Another key issue that is probably overlooked by all DIY speaker testers unless they use an accurate, true RMS meter, or take the measurements from an oscilloscope, is that as frequency changes the meter reading does not stay linear. I got around this with my old meter by performing a calibration sweep with a 10 ohm resistor at 5Hz intervals from 20-100Hz and using this data to correct my speaker measurements.

[kelticwizard]

About the baffled versus unbaffled measurement.

Peerless HDS 164 6.5" speaker. 850439-official specs.

Fs:
Free air: 44.9 Hz
Baffled: 43.6 Hz

So there is only a small difference.

Your measurements are only a little higher than the specs anyway-about a quarter octave.

One thing you should understand is that Fs, Qts and Vas all work together to determine the Fc in a closed box, or the bass behavior in a vented box.

If your Fs=35 Hz woofer is a quarter octave above the 29.5 specs, do not be surprised if the Vas of your woofer is smaller than the specs. An Fs=35 Hz woofer, with a Vas 25% less than stated specs, will perform similarly when placed in the box. Ideally,the Qts of your woofer should be a little higher than spec, but it's okay if it is the same.

[kimbo]

Great Guys! The members out there are so quick to assist newbies. Your responses give me confidence in what I am doing. Thanks! Yes I thought the idea of a baffle was a bit unusual as it amounts to almost putting the driver in an enclosure. My impression is that manufacturers generally quote the 'free air' resonance ie. no baffle, so Silicon Chip seem to be going it alone.

Yes, I did break them in, without baffles.....about 10 hrs of loud program with the loudness on and about 20-30 hours of sine wave at about the resonance to give about 50-70% Xmax. I also tested them every 5-10 hours to track the change in Fs. There was little change at first, but after about 10 hrs I started to see a drop which finally leveled out to be about one Hz from new. Temperature was pretty constant when tested and I let them rest overnight before testing.

I originally mounted the driver on a tall and pretty flimsy stand as I wanted it to have as little stuff around it as possible. I took your advice SY and beefed up the mounting by running a brace from the speaker to a wall to really stiffen it up. After averaging 20 measurements I saw about a 0.2 Hz drop when braced. If the speaker was originally actually suspended on cables to start with the drop may have been greater.

I am measuring the peak by using a scope. I measure the impeadance by using a simple jig with a 500 ohm pot with a parallel 37ohm resistor and a double pole switch. I view a trace from the speaker, then flick the switch which substitutes the resistor/pot and adjust the pot to get the same trace level. When I flick the switch back to the speaker it also switches in a digital multimeter so I can directly read off the resistance of the resistor/pot. I then subtract the resistance of the leads etc to get the true impedance. I'm now reasonably confident that the setup is accurate.

But does anyone else have an opinion with regard to low (less than one ohm) amp as per Leo's email to me?

Thanks again to you all. I'll believe what I am measuring and give up trying to massage the figures to match what I have read.

[SY]

A low output Z amp is useful if you want to quantitate the peak, i.e., know if it's 30 ohms or 31 ohms. If you've got a high source Z, you'll be able to get the frequency OK, but the magnitude will be off. However... nearly any competent solid state amp will have sufficiently low source Z to work splendidly in this application- I use a simple LM1875-based circuit.

[wintermute]

I think you will find that what they mean by less than 1 ohm amp is the output impeadance of the amp, I think any solid state amp should have an output impeadance of well under 1 ohm.

This makes me wonder about the measurements I did recently with speaker workshop and my Audigy II ZS sound card. I did the measurements driving the speaker directly off the soundcard, I don't know what it's output impeadance is...... might pay to set up a test using my amp for comparison.

Tony.

[richie00boy]

Originally posted by SY
A low output Z amp is useful if you want to quantitate the peak, i.e., know if it's 30 ohms or 31 ohms. If you've got a high source Z, you'll be able to get the frequency OK, but the magnitude will be off. However... nearly any competent solid state amp will have sufficiently low source Z to work splendidly in this application- I use a simple LM1875-based circuit.

The issue of high Z causing slight under-reading of the peak amplitude is because of potential divider action, isn't it? So if your source Z is high enough then the error is minimised.

[SY]

Well, in a usual test setup, you've got a resistor in series with the output of the amp to make a current source-ish drive. The idea is that you want the amp's output Z low so that the driving impedance can be closely approximated to the series resistor. Any Z from the amp will cause some error unless that Z is accurately known for all conditions of drive. It usually isn't, hence the desirability of a source Z that is negligible compared to the size of the resistor.

[richie00boy]

I think we were coming from different angles there, both valid points