I have been learning about measurement of T/S parameters and I have run across a site that recommends to do a voltage calibration to take into account the change in Vs that an amp will give as you sweep though frequencies. The site that talks about it is here:
http://www.mh-audio.nl/downloads/read-measuring-tsp.pdf
I really feel like this could be beneficial to getting better measurements. What I can't quite grasp about it is the method that they suggest to accomplish this. This site suggests to take measurements across the in series resistor as your sweep frequencies and then use that as the voltage calibration file. THis makes no sense to me. I'm thinking that they actually meant to take a measurement across the speaker terminals to get the actual Vs so that you can see how it changes as frequency is changed and then adjust accordingly.
I also thought that perhaps it would be better to simply hook up another multimeter to the speaker terminals and just adjust voltage to be the same for all frequencies as you do your sweep. Can you see a downside to that?
I have also noticed that the methods for measuring Vs seem to be a bit ambiguous in many of the post that I have read. To me it would make sense to set up the test rig with the resistor and the driver in series. Once it is set up play a frequency that is likely within the linear region and measure and set the voltage across the speaker terminals to get Vs. Many of the posts seem to be measuring this value without the testing rig connected. Is there a reason for this because this seems wrong to me. I would love any pointers with regard to this subject.
If you are curious the main sites that I have been looking at are the above one and the following:
The Subwoofer DIY Page - Measurements
Measuring Loudspeaker Driver Parameters
http://www.mh-audio.nl/downloads/read-measuring-tsp.pdf
I really feel like this could be beneficial to getting better measurements. What I can't quite grasp about it is the method that they suggest to accomplish this. This site suggests to take measurements across the in series resistor as your sweep frequencies and then use that as the voltage calibration file. THis makes no sense to me. I'm thinking that they actually meant to take a measurement across the speaker terminals to get the actual Vs so that you can see how it changes as frequency is changed and then adjust accordingly.
I also thought that perhaps it would be better to simply hook up another multimeter to the speaker terminals and just adjust voltage to be the same for all frequencies as you do your sweep. Can you see a downside to that?
I have also noticed that the methods for measuring Vs seem to be a bit ambiguous in many of the post that I have read. To me it would make sense to set up the test rig with the resistor and the driver in series. Once it is set up play a frequency that is likely within the linear region and measure and set the voltage across the speaker terminals to get Vs. Many of the posts seem to be measuring this value without the testing rig connected. Is there a reason for this because this seems wrong to me. I would love any pointers with regard to this subject.
If you are curious the main sites that I have been looking at are the above one and the following:
The Subwoofer DIY Page - Measurements
Measuring Loudspeaker Driver Parameters
Just a little bump and an added thought.
I have been reading in speaker design cookbook, and there is a secession (8.31) on page 196 that talks about methods for doing impedance measurements. One of the methods that is talked about uses a potentiometer in the place of the driver to find the impedance at a specific frequency. In other words, you have the driver hooked up with a 0.1 ohm series resistor and you play a frequency and record the voltage drop across the resistor. Then you replace the driver with the potentiometer and change it's value until you get the same reading across the resistor. Then finally you can remove the potentiometer and measure it's resistance value to get the impedance at that frequency. That seems like an unnecessarily long process to me.
He also mentions a method in the same section where you have virtually the same setup, but instead of replacing the driver, you measure the drop across the driver as well, and then make two curves: a curve called the current curve (although it is measured in voltage) from the measurement across the series resistor, and a voltage curve measured across the driver terminals. This method makes a lot of sense to me until he started describing what to do with this data to get the true impedance curve. According to the book the two curves should be divided by each other while keeping them in volts, then using a log to linear conversion convert it from V to VdB, then again using a log to linear to go from VdB to ohms. I don;t quite understand why and what he is doing here. If it were me I would take the voltage measurements across the series resistor and divide them by the resistor value to get current, then I would take the voltage across the driver and divide it by the current I just calculate for every test point. Would this not give me the impedance curve that I am looking for, or am I thinking in DC here? I would love to hear your advice on this one.
I have been reading in speaker design cookbook, and there is a secession (8.31) on page 196 that talks about methods for doing impedance measurements. One of the methods that is talked about uses a potentiometer in the place of the driver to find the impedance at a specific frequency. In other words, you have the driver hooked up with a 0.1 ohm series resistor and you play a frequency and record the voltage drop across the resistor. Then you replace the driver with the potentiometer and change it's value until you get the same reading across the resistor. Then finally you can remove the potentiometer and measure it's resistance value to get the impedance at that frequency. That seems like an unnecessarily long process to me.
He also mentions a method in the same section where you have virtually the same setup, but instead of replacing the driver, you measure the drop across the driver as well, and then make two curves: a curve called the current curve (although it is measured in voltage) from the measurement across the series resistor, and a voltage curve measured across the driver terminals. This method makes a lot of sense to me until he started describing what to do with this data to get the true impedance curve. According to the book the two curves should be divided by each other while keeping them in volts, then using a log to linear conversion convert it from V to VdB, then again using a log to linear to go from VdB to ohms. I don;t quite understand why and what he is doing here. If it were me I would take the voltage measurements across the series resistor and divide them by the resistor value to get current, then I would take the voltage across the driver and divide it by the current I just calculate for every test point. Would this not give me the impedance curve that I am looking for, or am I thinking in DC here? I would love to hear your advice on this one.
It's actually my work and MH Audio downloaded an old PDF from my site.
The calibration is not looking at the series resistor and there is no speaker connected. Instead of a speaker (whose impedance is unknown and will change with frequency, hence be useless to calibrate against) a 10 ohm resistor is used. The voltage across this resistor is measured.
The calibration is not looking at the series resistor and there is no speaker connected. Instead of a speaker (whose impedance is unknown and will change with frequency, hence be useless to calibrate against) a 10 ohm resistor is used. The voltage across this resistor is measured.
By speaker terminals I assumed you meant the terminals of the drive unit. If you mean the amplifier terminals then yes I suppose you can see how the output changes with frequency by looking there directly with a meter, but then you would have to scale the readings to suit your measurements. The method I propose was the easiest way to keep everything relative.
You could do that and using some maths it would give you a result. But it's more complicated than you need it to be. You have to adjust voltage and frequency so there becomes more room for error. And the time to adjust the voltage as well.
I suspect you haven't done any actual measurements yet? Have a go following my guide and maybe you will see why it's done that way. Sometimes just reading stuff you don't get the full picture.
You could do that and using some maths it would give you a result. But it's more complicated than you need it to be. You have to adjust voltage and frequency so there becomes more room for error. And the time to adjust the voltage as well.
I suspect you haven't done any actual measurements yet? Have a go following my guide and maybe you will see why it's done that way. Sometimes just reading stuff you don't get the full picture.
Last edited:
- Status
- Not open for further replies.