Software design challenge for Klippel method testing

[augerpro]

Saurav> http://www.gedlee.com/distortion_perception.htm

Dennis> So 4x louder is 6dB though right? I'm not sure if I'll use SE or ARTA yet I may have to set this manually. I'm going to for sure have to step it down when switching to two tone. Chances are pretty good I'll have to step down the level when I switch to two tone. If I'm already at 10% HD I wouldn't think the IMD would be too good when flipping on that second tone? If I do that I have I would back down F1 one millimeter of excursion and re-adjust F2 so F1 is still 4x louder. So that's 6 dB right? And we're all sure about that it's F1 is 4x louder than F2? Sorry to beat this to death

Brian> how does ARTA figure the "composition" of THD? You mean the higher order components? That is one thing I like about spectrum plots. You can see a lot the higher order stuff, some drivers are just crazy with the high order stuff. Are you familiar with Wigan's and Geddes distortion metrics?

[brianpowers27]

Saurav> [url]

Brian> how does ARTA figure the "composition" of THD? You mean the higher order components? That is one thing I like about spectrum plots. You can see a lot the higher order stuff, some drivers are just crazy with the high order stuff. Are you familiar with Wigan's and Geddes distortion metrics?

I am pretty sure they are using the standard IEC THD methods. (I think there is a standard anyway.) THe manual is very verbose and likely details the specific methods.

I have read over the various theory from Geddes. I am not sure which specific metric you are describing. I think that his work seems fairly sound. From what I could tell I had a few issues since some of his test couldn't be performed past 80db, due to lab safety issues. Who listens at 80db?

Regardless of that metric I believe that the general principle for weighting THD/IMD is sound.

[brianpowers27]

From P.22 of the ARTA Users Manual

THD - total harmonic distortion – defined as percentage of the square root of ratio of power
sum of higher harmonics (H
,H3,..) to the power of fundamental signal harmonic (H ).
2 1

222
..
HHH
32
n
(%)100

THD
2
H
1

An alternative definition is frequently used:

222
..
HHH
32 n

(%)100
THD
2222
..
HHHH
n
321

In a denominator a full distorted signal power is used. This definition is closest to the value of
harmonic distortions that are measured by analog instrumentations in low noise systems. In a
system with a high noise a better quality measure, is a THD+N.

[Dennis H]

So 4x louder is 6dB though right?

12dB. Every time you double the voltage or sound pressure, it's 6dB.

[augerpro]

Thanks Dennis!

Saurav I emailed you some SE info.

SE seems to be able to do what I need. IMD calculation appears to include IM2, IM3, and IM5. Not sure I can change it, but I like including the higher orders so I don't think I want to change it. SE is so feature packed I forget all teh things it does.

[ch83575]

augerpro,

I really like the idea here. Sounds very useful. The thing that worries me is the demands this type of testing will put on measurement equipment. I don't know what equipment you have, but my concern is that when you start to adjust power instead of just frequency in a distortion measurement you could end up inadvertently measuring your mic more than your driver. I recently did a test of a drivers distortion at 10" then moved the mic to nearfield without changing the voltage to the driver. The measured 2nd harmonic went up 12db! The driver should have been producing the same distortion, so it had to be introduced by the microphone/pre. My mic is DIY, but I doubt the Beringer is drastically better (i don't really know, maybe it is). The only reason measurements like Zaphs are useful is because they are all done with the same mic at the same level.

I know the thread is more about software than process, but I for one would love to hear how you plan to overcome mic overload/distortion. Really expensive mic? Large measurement distances that keep the mic in its ideal SPL range? I would also like to be able to do similar tests, so I would love to hear your plans.

-Chad

[Dennis H]

Good point, Chad! Herb at Cross-Spectrum did some Behringer distortion tests.

http://www.hometheatershack.com/forums/spl-meters-mics-calibration-sound-cards/15951-cross-spectrum-microphone-calibration-service-2.html

I just ran a quick test comparing the distortion of my one of my reference mics (ACO Pacific 7052, rated to 144 dB) to an ECM8000 at 1 kHz.

At 115 dB, my ref mic gets 0.13% distortion, the ECM8000 reads 0.54%

At 120 dB, my ref mic gets 0.22% distortion, the ECM8000 reads 0.81%

At 125 dB, my ref mic gets 0.46% distortion, the ECM8000 reads 1.37%

Tests at 30 Hz:

At 115 dB, my ref mic gets 1.8% distortion, the ECM8000 reads 3.1%

At 120 dB, my ref mic gets 2.95% distortion, the ECM8000 reads 5.0%

At 125 dB, my ref mic gets 4.2% distortion, the ECM8000 reads 7.4%

FWIW, this is a different ECM8000 than the one used for the 1 kHz test.

[brianpowers27]

I remember reading informally that the ecm8000 is spec'd for 138db. I would test this theory by keeping the driver under load at a constant volume and moving the mic inward. This should allow you to relatively test the spl linaerity of the mic over a given range. I am guessing that the test Auger is desiring will have to be performed at least 2 ft away, in order to keep SPL reasonable.

Edit: It doesn't look like 138db would be acceptable at all! :E It seems like 110db is a good target for low F measurements.

[augerpro]

Well, there is the possibility of this being easier on the test setup. The same concerns are true with the sweeps Zaph (and myself, look at my sig) does. Here though you use a tone at Fs which generates maximum excursion with minimum power input. Also the SPL level is down considerably for most drivers at these frequencies, so not *too* much SPL is being generated. Now because we already have plenty of excursion happening the second tone won't have to be all that loud. Not like a normal sweep where you have to really crank it up for the entire sweep. At least that's my off the cuff take on the issue.

One thing that does concern me is the mic distortion profile either being linear or low enough to not be a problem. That may not be true as the info Dennis posted above shows. The HD may be artificially high when running the the tone at Fs and may show a 10% limit at a lower excursion than the IMD will when running two tones. I'll have to look into it more.

[Dennis H]

At 125 dB, my ref mic gets 4.2% distortion, the ECM8000 reads 7.4%

The way I read that, the speaker is distorting about 4% on the 30 Hz test, since the ACO shouldn't be distorting all that much at 125 dB, and the Behringer is adding another 3% of its own.

[augerpro]

SE v16 looks like it should be able to everything I need. I really have to hand it to Bohdan, SE just gets better and better. I'll be trying a test run this week on the distortion and see how it works.

Another issue: how would you guys compare the power compression of two drivers? I'm thinking a pink noise signal at the same voltage for some amount of time, followed by a quick check of the DCR with a multimeter. I could do a Z sweep too, but that would cause a bigger gap between pink noise and measurement, and I'm not sure that it would show anything more informative. My question is using voltage as the standard. Complex impedance will change the actual amount of current to be different. MAybe use SPL as the standard?

I have a TD12M coming and I've been curious how their method of heat removal (magnet heatsink) compares to the VC venting on my B&C 15NW76's. The 15NW76 has the exact same motor as the 12NW76 so I think I can make a apples to apples comparison.

[Dennis H]

MAybe use SPL as the standard?

That's how I've seen it done. Plot a bunch of curves, maybe smoothed RTA with pink noise, and you can see where they start to get closer together. If you double the voltage each time, the curves should be 6dB apart.

[ch83575]

Brandon,

How are the measurements coming? Have you nailed down a procedure for effectively testing x-max with tools available?

I was just thinking about how helpful a testing method like this will be for testing midrange domes. I realize it is not the main thrust of the evaluation, but as of right now we have no real way to evaluate dome x-max at all. I was looking at the new Tang Band mid-dome and wondering how it will hold up in the low end compared to the RS-52 or the ATC for that matter.

I know you are focusing on woofers, but I just thought that once you perfect your technique this would be another useful application for it.

-Chad

[augerpro]

Hi Chad. Tomorrow I start measurements and it'll probably take me the day to see if the methods I'll be using work well. I'll be trying the little Linkwitz device for measuring excursion. I'll then match this to the predicted excursion at the given frequency with the given voltage. This will be a little tricky. I can measure both voltage and current, so I have to normalize one of these to 1 watt to compare to SE's box modeler. This involves knowing the impdedance which gets tricky when measureing at Fs. I may try a different modeler that allows for a 2.83v input to make things easier.

I *may* also use a standard SPL. We'll see how things looks using all the methods, and how well they agree.

I think it would work fine for mid domes. Low frequency distortion can give you a good idea how two drivers compare in excursion also. The new 3" TB mid is head and shoulders above the RS52 in Zaph's latest testing.

[augerpro]

Anyone know how to do a DC offset on a driver, while still being able to run a signal through it?

[augerpro]

Anyone know how to do a DC offset on a driver, while still being able to run a signal through it?

No electrical gurus out there have any ideas?

[ThomasW]

Buy a Klippel system.....

[augerpro]

Come on now Tom, where is the DIY spirit?

[ch83575]

I would have to guess that it will involve building an amp. I am not an electronics expert, but I would imagine that any commercial amp will have some sort of DC protection or servo circuit. Or, perhaps you could use an isolation cap and apply the opposite transfer function in the computer afterwords? Or build an amp with adjustable DC offset like the Kippel system uses (I think CLIO also has this built into their amp module). The amp would need to be discrete, any chip amp solution would have the DC protection built into the chip. Again, this is not my strong suit, just trying to stimulate some ideas. I don't know of the validity of either technique and there may be a more simple solution that I just don't see.

-Chad

[Saurav]

I was thinking of attaching a battery across the driver terminals, but that would probably cause a rift in the space time continuum, and possibly other even more harmful side-effects like killing the amp.

[augerpro]

It seems to me what Klippel does is pretty simple and straightforward, but yeah, I don't want to start mixing AC and DC and acid and get a surprise.

[benchtester]

Anyone know how to do a DC offset on a driver, while still being able to run a signal through it?

Actually, I am pretty sure that chip amps are DC coupled and they depend on external caps or servos to remove DC offset.

It sounds like you need is a mixer (between source and power amp) and an amp that is DC coupled. Here is a simple passive mixer:

http://www.epanorama.net/circuits/linemixer.html

You would put the signal in Line 1 and a DC voltage on Line 2. Remember that the DC voltage will also be multipled by the amp gain.

[ch83575]

edit: benchtester posted while I was typing... I defer to him.

Is there any problem applying a large offset to a DC coupled amp? ie. cause oscillation in the amp or some other problem?

-Chad

[benchtester]

I am pretty sure using any dc source across the terminals would activate the amps dc protection. I am not positive, but I would think that almost all commercial amps have some sort of DC protection... from the simple 'turn of at the presence of x Vdc' to more advanced servo circuits that correct for small DC drift on the fly.

-Chad

I am just shooting from the hip, but I think there are two different types of protection. One is for DC offsets and most amps this is simply blocking capacitors. The second is currently limiting that kicks in only at maximum current.

Ideally you can jumper the blocking caps and the amp should operate correctly. I would not suggest this without a circuit diagram and some preliminary testing to make sure the amp is DC stable.

It is hard for me to think of the reverse engineering since years ago I made chip amplifiers which were DC coupled and I rarely got around adding the DC servo.

Brandon, how much power, offset and frequency range do you need?

[benchtester]

edit: benchtester posted while I was typing... I defer to him.

Is there any problem applying a large offset to a DC coupled amp? ie. cause oscillation in the amp or some other problem?

-Chad

I am cross posting too. Fundamentally, the amp doesn't care about the DC offset. I would be a little worried that a component like a feedback resistor or transistor might be boardline on the power rating and the DC offset might put it over the edge. But for short periods I doubt it would be a problem and I think there would be a greater chance of damaging the speaker driver.

[augerpro]

I was going to use this amp, but now that I've calculated the voltages I need to reach the specified excursion on some of the drivers, I'm thinking about using this amp. I have both, I usually use either the Onkyo or the Yamaha receiver's own amp. These will be driven at Fs, so as I understand it, the voltage required may be quite high, but the actual current flow is much lower.

What I really need is a way to measure DC offset produced by K or BL nonlinearities. But that's just icing on the cake, the distortion measurements will still find the limit, it's just the DC offset is unique to the K and BL nonlinearity so it's a dead giveaway. Alternatively I thought about inducing an opposing offset of the K or BL nonlinearity and see how distortion cleans up. Then possibly calculate the actual DC offset caused by the K or BL nonlinearity from this improvement.

The other (primary) reason to cause a DC offset was so I could measure Le at various +/- excursions without having to use a mechanical device to hold the cone in position.

So there is no way to induce the offset downstream of the amp?

[benchtester]

So there is no way to induce the offset downstream of the amp?

All my DC coupled chip amps have less power/voltage than you what are looking at.

One could take the ground lead of the driver and attach it to a DC coupled amp, then the current would go through the amp to the ground of both amps. The second amp would only need to generate the DC offset volts, but would have to be capable of the full current. This is essentially a bridged amp except one side is the signal and the other side is the DC offset. The two problems is you still need one side to be DC coupled and I don't know if it will screw up the measurements. I assume the probes will be attached to the driver and now the probe ground has to be floating with respect to the amp grounds.

[JonP]

Hmmm... some interesting ideas here.

I had a few... but it's pretty late, I cant guarentee them...

One, you could get some really enourmous caps to AC couple the driver under test, then attach a DC offset source on the other side. Still don't know what the AF voltage would do to the supply.

Other one, if you had a DC servo corrected amp, you could adjust the servo to create an offset rather than correct to zero, and thereby dial up a DC voltage with the amp output itself, while still applying the audio... :B

[Wilk]

I think what Auger may be looking for is a Bias-T. It allows you to couple a DC voltage onto the line with out it getting back to the amplifier. Only the equipment after the T would get the DC. You can also use a second on to couple it back off the line after a UUT too if needed. They are used alot in RF testing. I have never seen one used at audio frequency or in 4-8 ohm enviroments, but i don't see why the concept wouldn't work. I'll look around and see if I can find one more suitable to what you are doing.

Auger do you have Voltage, Current specs for the AC, and DC signals you intend to be using?