Legendre,
Creep is really complicated.
The primary suspension element in the driver, the spider, has the majority of the total of friction in it. This is because it is composed of many interwoven fibers. These fibers have friction between them. It takes a certain minimum force to overcome the friction and displace the suspension to a new position. Once at that new position, the cone assembly may stop at this new position because the new orientation of the fibers may reach an equilibrium of forces. The spider portion of the suspension has a lot of hysteresis. In fact it has multiple hysteresis loops, not just one. This makes its stiffness very uncertain.
For a blocked impedance measurement, you need to keep the vibrating portion of the driver from moving. The only really good way to do this is to have the coil separate from the cone assembly and lock it in place. If you lock the coil in place at different distances from the motor, you can look at some types of drive nonlinearities, such as inductance variance.
To get good impedance measurements, you need to minimize any environmental vibration and sound. Remember, the driver works just as well turning sound into current as it does current into sound. For best results it is normally always better to mount the driver so its cone is perpendicular with the ground.
The measurement settling issue alone, severely limits how consistent the impedance measurements can be. To get around this, some measurement systems take a wide band measurement, then do a curve fit on the impedance data to a driver model of what the impedance curve should look like. Other systems use a sine wave technique with programmable settling parameters.
Creep is really complicated.
The primary suspension element in the driver, the spider, has the majority of the total of friction in it. This is because it is composed of many interwoven fibers. These fibers have friction between them. It takes a certain minimum force to overcome the friction and displace the suspension to a new position. Once at that new position, the cone assembly may stop at this new position because the new orientation of the fibers may reach an equilibrium of forces. The spider portion of the suspension has a lot of hysteresis. In fact it has multiple hysteresis loops, not just one. This makes its stiffness very uncertain.
For a blocked impedance measurement, you need to keep the vibrating portion of the driver from moving. The only really good way to do this is to have the coil separate from the cone assembly and lock it in place. If you lock the coil in place at different distances from the motor, you can look at some types of drive nonlinearities, such as inductance variance.
To get good impedance measurements, you need to minimize any environmental vibration and sound. Remember, the driver works just as well turning sound into current as it does current into sound. For best results it is normally always better to mount the driver so its cone is perpendicular with the ground.
The measurement settling issue alone, severely limits how consistent the impedance measurements can be. To get around this, some measurement systems take a wide band measurement, then do a curve fit on the impedance data to a driver model of what the impedance curve should look like. Other systems use a sine wave technique with programmable settling parameters.
Hardly sounds like the old KLH company that made classy gear in the "New England sound" period. Lots of classy old HiFi names have been sold to definitely unclassy new factories in emerging nations.
Yes, blocked impedance is the impedance when blocked from moving.
Jack Hidley is right, proper impedance measurements can't be done just sitting on the bench on the magnet. But slight stiction or failure to return to static position wouldn't affect impedance differentially or as you seem to have described it.
If the suspension was really non-linear or had some kind of lazy, break-in, dead cockroach, or other weird behaviour unsuitable for a speaker, you could get some kind of impedance change from trial to trial. But unlikely to be consistent and repeatable.
Strange.
Can you simply tell us if the speaker (or raw driver sitting on your bench) makes clean sound or not.
B.
Yes, blocked impedance is the impedance when blocked from moving.
Jack Hidley is right, proper impedance measurements can't be done just sitting on the bench on the magnet. But slight stiction or failure to return to static position wouldn't affect impedance differentially or as you seem to have described it.
If the suspension was really non-linear or had some kind of lazy, break-in, dead cockroach, or other weird behaviour unsuitable for a speaker, you could get some kind of impedance change from trial to trial. But unlikely to be consistent and repeatable.
Strange.
Can you simply tell us if the speaker (or raw driver sitting on your bench) makes clean sound or not.
B.
Last edited:
@bentoronto
Both of the KLH speaks play just fine, last time they were in a system - they are one of several pair of small utility speakers that I keep around. I use them when I'm in the early shake-down phases of an amp rebuild / test job, and so forth. They have a reasonable power limit, are very small, and cheap enough that a destructive accident won't break the bank. The sonic qualities are such that I can quickly tell if something is way off.. and that's their job.
I've never put music through the bare driver, though the various test signals (sine, tri, square) all sound out as you would expect.. without any weird defects / buzzes etc.
Both of the KLH speaks play just fine, last time they were in a system - they are one of several pair of small utility speakers that I keep around. I use them when I'm in the early shake-down phases of an amp rebuild / test job, and so forth. They have a reasonable power limit, are very small, and cheap enough that a destructive accident won't break the bank. The sonic qualities are such that I can quickly tell if something is way off.. and that's their job.
I've never put music through the bare driver, though the various test signals (sine, tri, square) all sound out as you would expect.. without any weird defects / buzzes etc.
Image that, a complex material property!
But yes, I totally get what you're on about.. and my sense is that the 'creep' factor plays into the driver break-in phenomenon. If we can't count on the home position of the voice coil within the gap, neither can we count on a given impedance or response at any frequency. Sure, there will be common trends, but absolute values must shift a bit.
If that's true, then it would seem that a well-broken-in driver would have fewer issues with creep, due to material break-in. Is that the case, or am I off?
KLH LS #1 - after 48hr rest
Instruments: Fluke 8050A DMM (cal'd), Wavetek 111 funct. generator (sine) monitored on Racal-Dana 9904 frequency counter. The impedometer circuit uses a 1K 1% 1/2W metal film current limiting resistor and an 8R 1% 1/2W metal film resistor hand selected for precision 7R997 value.
(Note: Due to possible 'creep' factor, 15 sec settling time is allowed from the moment the LS is switched into the meter circuit until the DVM measurement is taken. This figure was chosen somewhat arbitrarily, based on the RoC of the reading after the DUT is switched into the circuit.)
Freq - Impedance (R)
20 - 12.13
25 - 12.17
31.5 - 12.18
40 - 12.25
50 - 12.30
63 - 12.51
71 - 12.70
80 - 13.0
90 - 13.4
100 - 14.2
113 - 16.16
126 - 20.63
143 - 26.23
159 - 20.32
180 - 15.45
200 - 13.76
250 - 12.97
315 - 12.38
400 - 12.30
500 - 12.36
630 - 12.48
800 - 12.60
1000 - 12.74
And this is an 8R speaker!? Wow.. seems more like 16R to me, but then again, we'll see what happens when I shake it up a bit.
Instruments: Fluke 8050A DMM (cal'd), Wavetek 111 funct. generator (sine) monitored on Racal-Dana 9904 frequency counter. The impedometer circuit uses a 1K 1% 1/2W metal film current limiting resistor and an 8R 1% 1/2W metal film resistor hand selected for precision 7R997 value.
(Note: Due to possible 'creep' factor, 15 sec settling time is allowed from the moment the LS is switched into the meter circuit until the DVM measurement is taken. This figure was chosen somewhat arbitrarily, based on the RoC of the reading after the DUT is switched into the circuit.)
Freq - Impedance (R)
20 - 12.13
25 - 12.17
31.5 - 12.18
40 - 12.25
50 - 12.30
63 - 12.51
71 - 12.70
80 - 13.0
90 - 13.4
100 - 14.2
113 - 16.16
126 - 20.63
143 - 26.23
159 - 20.32
180 - 15.45
200 - 13.76
250 - 12.97
315 - 12.38
400 - 12.30
500 - 12.36
630 - 12.48
800 - 12.60
1000 - 12.74
And this is an 8R speaker!? Wow.. seems more like 16R to me, but then again, we'll see what happens when I shake it up a bit.
Last edited:
If you have an oscillator and an amplifier handy set it to 40c/s and drive your speakers at a low level (just below Xmax of your bass driver) for about 20 minutes.
It should then be ready to be measured after about a 30 minute cool down period. That should stabilize measurements for you. Cheap loudspeaker drivers can vary as much as 20% from their advertised specs, that's the reality of manufacturing that DIY'ers face.
C.M
It should then be ready to be measured after about a 30 minute cool down period. That should stabilize measurements for you. Cheap loudspeaker drivers can vary as much as 20% from their advertised specs, that's the reality of manufacturing that DIY'ers face.
C.M
This one is quite a bit different from the first series. And again, this is the self-same speaker, post-disturbance..
KLH LS #1 - Series #2 - Gently pushed, pulled, banged, bounced, shouted at in a disaproving tone and generally disturbed & perturbed - after 48hr rest.
Instruments: Fluke 8050A DMM (cal'd), Wavetek 111 funct. generator, Racal-Dana 9904 frequency counter. (Note: Due to possible 'creep' factor, 15 sec settling time is allowed from the moment the LS is switched into the meter circuit until the DVM measurement is taken. This figure was chosen somewhat arbitrarily, based on the RoC of the reading after the DUT is switched into the circuit.)
20 - 7.55
25 - 7.69
31.5 - 7.63
40 - 7.70
50 - 7.85
63 - 8.08
71 - 8.31
80 - 8.66
90 - 9.20
100 - 10.15
113 - 12.38
126 - 16.68
143 - 21.85
159 - 15.81
180 - 11.21
200 - 9.53
250 - 8.55
315 - 7.93
400 - 7.86
500 - 7.87
630 - 8.06
800 - 8.19
1000 - 8.35
As you can see, the variations are large.. to the point that they render the measurements almost pointless. We can see the same trends in Z, but the absolute values are shifted significantly.
KLH LS #1 - Series #2 - Gently pushed, pulled, banged, bounced, shouted at in a disaproving tone and generally disturbed & perturbed - after 48hr rest.
Instruments: Fluke 8050A DMM (cal'd), Wavetek 111 funct. generator, Racal-Dana 9904 frequency counter. (Note: Due to possible 'creep' factor, 15 sec settling time is allowed from the moment the LS is switched into the meter circuit until the DVM measurement is taken. This figure was chosen somewhat arbitrarily, based on the RoC of the reading after the DUT is switched into the circuit.)
20 - 7.55
25 - 7.69
31.5 - 7.63
40 - 7.70
50 - 7.85
63 - 8.08
71 - 8.31
80 - 8.66
90 - 9.20
100 - 10.15
113 - 12.38
126 - 16.68
143 - 21.85
159 - 15.81
180 - 11.21
200 - 9.53
250 - 8.55
315 - 7.93
400 - 7.86
500 - 7.87
630 - 8.06
800 - 8.19
1000 - 8.35
As you can see, the variations are large.. to the point that they render the measurements almost pointless. We can see the same trends in Z, but the absolute values are shifted significantly.
Last edited:
Correct. As I said previously, this KLH 9900 is a small 2-way HT satellite / bookshelf speaker. I suspect the 'crossover' is just an NP electrolytic cap in series with the tweeter leads. Isn't that typical for a basic 2-way? The woofer is like 3" or so, with a 3/4"-1" dome tweeter. Nothing at all special or fancy, just one of several pair of utility speakers I keep around for testing work.
Last edited:
Here in a free loudspeaker impedance testing program which I occasionally use, I find it quite useful for quick impedance checks. Making up a simple jig easily makes testing that easy, not like that with SpeakerWorkshop.
It would require a full duplex soundcard though, I use a SoundBlaster Audigy Rx soundcard which is readily available. As I say the program is free but donations are appreciated.
https://www.acustika.org/
C.M
It would require a full duplex soundcard though, I use a SoundBlaster Audigy Rx soundcard which is readily available. As I say the program is free but donations are appreciated.
https://www.acustika.org/
C.M
Last edited:
Take the bass/midrange out of the enclosure and inspect it to see if the suspension, especially the spider is supplying a restoring force. Use a 1.5 V dry cell battery across the speaker terminals and observe if the cone shifts forward and back, when the battery is connected and disconnected. Do this test with the battery reversed and again observe the cone motion. If the cone appears to be moving freely and returning to the same position, do an impedance run with this speaker held vertically and do re-runs to verify the consistency of the measurements.
I do have both internal (PCI + 5-1/4" bay) and external (USB) Audigy hardware, though none is currently in-use. That stuff is over 10+ years old, though I'm not sure that matters in this case - so long as its functional.
But does the software run on Linux? I use Mint / Ubuntu.
Not sure that's possible.. these are pretty cheap, I didn't even notice if there are screws in the cabinets (or on the drivers). But I get what you're asking about.
Can you clarify? Most all boxed speakers have the drivers - cones that is - in a vertical orientation. Do you mean the driver should face down towards the table surface, or outward, as in a normal listening situation?
I'm running Windows 7 and it works perfectly, as to Mint/Ubunta ?..... there is no mention of what operating system is applicable ....on the site, so it is a matter of try and see I guess. It is only a 4.6 MByte download.
C.M
Last edited:
legendre, everyone is giving you advice on how to proceed to overcome the apparent fault that your impedance test is showing. Removing a speaker is necessary in order to confirm if it is in working order. If the speaker is fitted via the inside of the box and there isn't a removable panel, then it all becomes a futile exercise. Testing in the vertical plane ,means that the speaker is operating in its normal position, definitely not face down or up.
Simplest would be to try Room Equalization Wizard if you have a java client. You can also do these impedance measurements with a linux program called baudline using the same type of jig you use for other programs, but you will probably have to postprocess using something else...like libre office calc.
You could also try running some of the older programs like speaker workshop or ARTA in Wine.
You could also try running some of the older programs like speaker workshop or ARTA in Wine.
@VaNarn
Hey now.. I'm nothing, if not genuinely appreciative of all the assistance I am receiving here. Very nice forum you've made for yourselves, glad I found my way in.
As for further tests on this specific KLH, this exercise isn't about analyizing or repairing a defective LS, more about working out possible sources of error in my testing methods.. As a pair, the KLH play as expected and serve their intended purpose as "audible test loads" for works in progress rather than risk damaging a finer pair of speaks.
Thanks, that's what I was assuming - but being that vertical is unquestionably the default orientation for an LS, I was confused that you specified it.. (did he mean "horizontal"?). You know, Internet.
Hey now.. I'm nothing, if not genuinely appreciative of all the assistance I am receiving here. Very nice forum you've made for yourselves, glad I found my way in.
As for further tests on this specific KLH, this exercise isn't about analyizing or repairing a defective LS, more about working out possible sources of error in my testing methods.. As a pair, the KLH play as expected and serve their intended purpose as "audible test loads" for works in progress rather than risk damaging a finer pair of speaks.
Thanks, that's what I was assuming - but being that vertical is unquestionably the default orientation for an LS, I was confused that you specified it.. (did he mean "horizontal"?). You know, Internet.
@RonE
Thanks for that. I do have a JVM and will certainly check out that Room Eq software. Not familiar with baudline either, another one to look at.
In case it isn't totally obvious, I don't do much work in software.
@All
ETA: I need to clear up one point. These KLH 9900 have both a removable rear panel and both drivers can be removed from the front of the cab. Sorry about that, if I gave the wrong impression - mostly just that I'd never noticed.
Thanks for that. I do have a JVM and will certainly check out that Room Eq software. Not familiar with baudline either, another one to look at.
In case it isn't totally obvious, I don't do much work in software.
@All
ETA: I need to clear up one point. These KLH 9900 have both a removable rear panel and both drivers can be removed from the front of the cab. Sorry about that, if I gave the wrong impression - mostly just that I'd never noticed.
Last edited:
So what I've gathered from this exercise is that, as in many cases, measurements aren't nearly as simple & straightforward as one might expect (or 'hope' might be a better term). Like most complex physical systems, an LS or even a bare driver doesn't always behave in a predictable way.. some experience and finesse is required to obtain good data.
As for these particular KLH models, my sense is that they never really had any break-in, or if they did, they have spent so much time sitting around that the suspensions have become stiff again. They are only used infrequently, here - maybe a day per month, averaged over a year. I suspect that if they were put through a break-in regimen, they would be much more repeatable, at least over the short-term.
As for these particular KLH models, my sense is that they never really had any break-in, or if they did, they have spent so much time sitting around that the suspensions have become stiff again. They are only used infrequently, here - maybe a day per month, averaged over a year. I suspect that if they were put through a break-in regimen, they would be much more repeatable, at least over the short-term.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.