The absolute magnitude of the impedance plot isn't vital either here. I mean say you get a minimum at 10 ohms when the driver actually hits its minimum at 6ohms. This will still let you see any anomalies created by resonances within the drivers construction.
Obviously if you're wanting to use the impedance data for passive crossover simulation you'll need to tighten things up, but for something that is only designed to show the trend of the impedance rather then its absolute magnitude you'll be fine.
Obviously if you're wanting to use the impedance data for passive crossover simulation you'll need to tighten things up, but for something that is only designed to show the trend of the impedance rather then its absolute magnitude you'll be fine.
I just get 6.1 ohm linear until a very slow rise starting at ~1.5k, plus random inconsistent stuff in the 50-100hz range - nothing like you'd expect to see for bass impedance spike(s). Maybe the output of my USB preamp/soundcard thing isn't going to cut it for this?
Just copy the impedance plot and paste it on the forums so we can have a look. It sounds fine. Impedance measurements don't tend to need doing at huge volumes. To see any irregularities you might need to expand the vertical axis, this will push the impedance peak at fs off the screen, but will allow us to see the 500-10k range in more detail for example.
If this uses MLS signal for the measurement, the probably the window is too long.
Can you show the impulse?
Can you show the impulse?
Also are you using averaging with Limp? If there's noise in the system it can show up as spikes such as those, also you might be clipping the signal somewhere a little. Try it at different volume and see if it helps to clean up the low end.
Okay nevermind that first one - I had misread an important part of the instructions. Now I have a new question... why does the bass impedance look upside down? (This one is averaged PN for speed - I can get cleaner results with the stepped measurement, which I will post as soon as I am getting something that looks right).
Attachments
Well the only area where the second graph is superior to the first is that it succeeded in measuring the lower frequencies. Everything else about it is 'off'. The first graph was far better, in that the overall magnitude of the impedance was correct, as was the overall shape, the only problem was what's going on at low frequencies.
This way of measuring the impedance is very simple and hence there isn't an awful lot that can go wrong with it, but you do have to make sure that a few things are correct.
1) Are the left and the right input channels to the sound card connected up correctly?
2) As you're using the sound card to drive the loudspeaker is the output of the sound card clipping?
3) Is the input to the sound card clipping in any way?
The first is an easy check and it's easy to try it both ways and see what happens.
The third point is probably not a problem in your case as you're using the sound card to drive the loudspeaker. Most sound cards can accept the full scale output back into the input without any problem, but just make sure that the inputs aren't clipping.
The second point is a little harder unless you have a scope, but what I'd suggest you do is try the measurements at different volume settings and see if it improves things at all.
You should be able to get very clean results from the averaged pink noise, this is how I do my measurements. The averaging is necessary to help remove noise.
The last thing to perhaps watch out for is any processing that your sound card might be doing somewhere as this could easily screw the measurements up. Such as any EQ or sample frequency re-sampling.
This way of measuring the impedance is very simple and hence there isn't an awful lot that can go wrong with it, but you do have to make sure that a few things are correct.
1) Are the left and the right input channels to the sound card connected up correctly?
2) As you're using the sound card to drive the loudspeaker is the output of the sound card clipping?
3) Is the input to the sound card clipping in any way?
The first is an easy check and it's easy to try it both ways and see what happens.
The third point is probably not a problem in your case as you're using the sound card to drive the loudspeaker. Most sound cards can accept the full scale output back into the input without any problem, but just make sure that the inputs aren't clipping.
The second point is a little harder unless you have a scope, but what I'd suggest you do is try the measurements at different volume settings and see if it improves things at all.
You should be able to get very clean results from the averaged pink noise, this is how I do my measurements. The averaging is necessary to help remove noise.
The last thing to perhaps watch out for is any processing that your sound card might be doing somewhere as this could easily screw the measurements up. Such as any EQ or sample frequency re-sampling.
1) The only thing I'm not clear on is the output - I used the positive from the right headphone channel to go into the cable that has the resistor and divides things up like the diagram above.
2) it's possible
3) No. It can clip itself, but it has level indicators.
I could probably get better results as far as cleanliness from the averaged PN if I messed with the settings - I did not try since the stepped method gives very clean results and the overall curve is the same as above.
2) it's possible
3) No. It can clip itself, but it has level indicators.
I could probably get better results as far as cleanliness from the averaged PN if I messed with the settings - I did not try since the stepped method gives very clean results and the overall curve is the same as above.
Last edited:
I am assuming that when you say above you mean the first graph.
How clean is clean? Neither of your posted measurements are clean enough yet to reveal small resonances. Unless of course all those wiggles between 700hz and 5khz are all little resonances, in which case 😱
Okay, so I really can't figure out why my magnitude is so far off. I even tried connecting 40Ω of resistor instead of a driver and running limp, then adjusting the levels until I got exactly 40Ω on the graph. Still got pretty much the same ~10Ω nominal in the driver results.
However, when I had the resistor replacing the driver, I did notice the influence of 60Hz noise coming and going in the averaged readout, so I took things upstairs (to the desktop PC and away from a lot of electricity). Bass makes a little more sense now, at least.
However, when I had the resistor replacing the driver, I did notice the influence of 60Hz noise coming and going in the averaged readout, so I took things upstairs (to the desktop PC and away from a lot of electricity). Bass makes a little more sense now, at least.
Attachments
Last edited:
if you are driving the speaker directly from the soundcard output, there probably isn't a lot of low frequency energy getting to the speaker. The output caps on the soundcard are probably about 10uf, and have about 100ohms resistors in series, so into an 8 ohm load, that's gonna start tapering off up at around 150 Hz. For measuring impedance that way, a number of things have to be right. Both channels have to be matched, sufficient energy has to get to the driver at frequencies of interest, there has to be low distortion in the amp driving the load (mls method is very fragile when used with distortion, and a line level amp trying to drive a speaker will likely distort), and the amplifier driving the speaker should be ground referenced (not bridged like with many class-c amps). Simplest way is to use a wt3, wt2, or woofer tester.
Last edited:
I'm using the headphone output, not the line level. While I understand I probably wouldn't be having these problems with a woofer tester, isn't this pretty much the same system that I'm trying to put together here?
The output of the sound card doesn't see an 8 ohm load, it sees the 8 ohm load in series with the test resistor. Some sound cards have output coupling caps, some don't. Mine doesn't, but then again I also use an amplifier to drive the loudspeaker under test and not the sound card.
Either way, the results above look okay to a certain extent.
I am colour blind so I am having a very hard time in telling the traces apart and then telling which is which... blue and yellow would be easier for me to discern. Either way.
One of the drivers from post 76 looks like your typical sub, lowish sealed alignment with a high voice coil inductance. This obviously causes the very steep rise in the impedance plot. It is also smooth showing no real blips or bump, hence the earlier bumps seem driver related.
The other driver and the graph from post 73 are a bit of a mystery. They do show the tuning frequency and the associated hump in the impedance, but what troubles me is what happens at higher frequencies, something isn't quite right here as the inductive rise seems almost non existent.
At a guess there's a problem with the measurement set-up somewhere and the fact the sub driver shows rising impedance is probably because its inductive rise is so large that it swamps the issue with the measurement set-up.
I use exactly the same measurement method to what your doing except that I put an amplifier between the sound card and the loudspeaker. This might be what's causing the issue? You could try the same yourself but you have to be really really careful. You must keep the signal level low otherwise you will fry the sound cards input. In limps measurement setup you will see that it mentions protecting the input to the sound card. This isn't necessary providing you are careful! I don't use any kind of protection for my sound card and I've never had an issue with LIMP, then again I know that if I do blow the input its only a 50p opamp that needs replacing. (I've blown it but not when using LIMP!). The safest bet is to use an integrated amplifier with the volume down low, this will prevent any large signals from reaching the sound card. And if the amplifier has any turn on thump, do make sure to turn it on before connecting the output to the input of the sound card.
You mentioned earlier, that when measuring the value of a resistor that the value LIMP measured was affected by the output level, this should not happen. LIMP measures the difference between two points which is done as a ratio of the output voltage. This is set-up by the fixed resistor and the component under test, the ratio should remain exactly the same regardless of the output voltage, so if this was having some effect then something is wrong.
I take it you have gone into the LIMP set-up and made sure everything is set correctly?
Either way, the results above look okay to a certain extent.
I am colour blind so I am having a very hard time in telling the traces apart and then telling which is which... blue and yellow would be easier for me to discern. Either way.
One of the drivers from post 76 looks like your typical sub, lowish sealed alignment with a high voice coil inductance. This obviously causes the very steep rise in the impedance plot. It is also smooth showing no real blips or bump, hence the earlier bumps seem driver related.
The other driver and the graph from post 73 are a bit of a mystery. They do show the tuning frequency and the associated hump in the impedance, but what troubles me is what happens at higher frequencies, something isn't quite right here as the inductive rise seems almost non existent.
At a guess there's a problem with the measurement set-up somewhere and the fact the sub driver shows rising impedance is probably because its inductive rise is so large that it swamps the issue with the measurement set-up.
I use exactly the same measurement method to what your doing except that I put an amplifier between the sound card and the loudspeaker. This might be what's causing the issue? You could try the same yourself but you have to be really really careful. You must keep the signal level low otherwise you will fry the sound cards input. In limps measurement setup you will see that it mentions protecting the input to the sound card. This isn't necessary providing you are careful! I don't use any kind of protection for my sound card and I've never had an issue with LIMP, then again I know that if I do blow the input its only a 50p opamp that needs replacing. (I've blown it but not when using LIMP!). The safest bet is to use an integrated amplifier with the volume down low, this will prevent any large signals from reaching the sound card. And if the amplifier has any turn on thump, do make sure to turn it on before connecting the output to the input of the sound card.
You mentioned earlier, that when measuring the value of a resistor that the value LIMP measured was affected by the output level, this should not happen. LIMP measures the difference between two points which is done as a ratio of the output voltage. This is set-up by the fixed resistor and the component under test, the ratio should remain exactly the same regardless of the output voltage, so if this was having some effect then something is wrong.
I take it you have gone into the LIMP set-up and made sure everything is set correctly?
Correct, that is the sub, and both the nominal magnitude and the location of the spike are right about where they should be, which makes me extra-confused about why the magnitude seems way off for the TD15M, which is only 2.9 DCR. There are some small bumps in the mfg. measurement posted (far) above in the same range. There is also some 1kHz noise with harmonics in the system, which is appearing in some of these and not others.
The Bozak driver also has a realistic magnitude and the Fs spike looks right, but I agree. It is worth noting that both woofers are supposed to have very small inductive rise, but not this small. The TD15 should rise from something like 4Ω to 10Ω, not 9Ω to 10Ω...
No, not the output level. I can attenuate each input individually, effectively throwing off the channel balance and thus changing the magnitude coming out of Limp. Obviously, these passive components could be messing up my results, but since the magnitude of these other two drivers looks like the right neighborhood.. 😕
I will see about trying it with an amp.
One thing is clear - this does NOT work at all when the furnace is running, which makes me think that it's too sensitive to acoustic noise in general.
edit: just for kicks, here's another one: a pair of ADS L810 (measured one at a time, they are supposed to be 6Ω nom., 4Ω min.). 0.4Ω at crossover? I sure hope not.
edit: just for kicks, here's another one: a pair of ADS L810 (measured one at a time, they are supposed to be 6Ω nom., 4Ω min.). 0.4Ω at crossover? I sure hope not.
Attachments
Last edited:
Good grief I think that helps confirm that something is definitely a miss with your measurement setup. No the ADSL810 does not dip to 0.4 ohms, the impedance plot is seemingly inverted as it was in post 69, the confusing part is the so called 'rising' part of the impedance at higher frequencies, but having seen post 79 I think its safe to say that the small rise at high frequencies that all your plots have been showing could in fact be system related rather then driver related? Either way something is really off.
So really that giant dip at 700hz or whatever is in fact a giant peak, which probably corresponds to a crossover frequencies somewhere.
If the measurements are being affected to such a degree that they are made impractical when the furnace is running then it really makes me wonder what's going on as that's quite bizarre.
So really that giant dip at 700hz or whatever is in fact a giant peak, which probably corresponds to a crossover frequencies somewhere.
If the measurements are being affected to such a degree that they are made impractical when the furnace is running then it really makes me wonder what's going on as that's quite bizarre.
- Status
- Not open for further replies.