Yes, you are talking about two different things.
The circuit he's shown is 'not' the attenuator circuit that was being talked about. In fact, the circuit he's shown has no separate attenuator.
Measuring voltage drop measured across a 27 ohm reference resistor is a perfectly adequate method for capturing driver impedance. No, as you say, it does not maintain constant voltage on the DUT, but that's understood in the calculations.
Dave.
I switched to 0.1 ohms triggered by PURIFI, so i measure the same way thus should get comparable outcomes. Which turnen out to be the case.if you put any resistor in series with the loudspeaker and just measure the voltage over this (= current) plus measure the output voltage just coming out from the power amplifier, the results will have an offset of that resistance (0.1 ohm in this example).
There are definitely ways to compensate for this, but putting such a small resistance just between the output of the amplifier and the feedback loop of the amplifier is much easier.
An error of 0.1ohm is small, but it's one nevertheless
In arta Limp you still need to calibrate, in essence measuring the voltages before and after the resistor.
So even if it introduces an error , first of all no idea how relevant the error is, nor do i believe that the inclusion in feedback is error free.
It is a already very old principle that when observing the working of a system , you influence tbe working of that system.
So in summary the error by Rsense is for me not an issue.
Indeed, it's not an error but you should keep it in mind.
I did "high" voltage impedance measurements of vented speakers to test the influence of port turbulences. In this case the measurement resistor needs to be considered (reducing the voltage at the voice coil). But not for small signal measurements.
I used REW for my stepped level impedance measurements.
You can chose a specific output level for the measurement (dB but it relates to a specific amp output voltage for the complete sweep). The voice coil voltage is reduced by the measurement resistor. Thus I did not measure the voice coil impedance at a specific voice coil voltage but at a specific amp output voltage.
Let's just say that the value of resistor needs to have minimal effect on the driver or measurement. You also have to be able to accurately measure voltages in case they are too small. If you are making a quasi-current source, you need a very high value resistance. If you are measuring current, the resistance needs to be as small as practical. Simple.
Can you sort of correct for things? Sure you can, but you have to make sure they do not affect how the driver operates. Otherwise you will get results but they will not reflect the driver in normal operation.
In testing, designing the experiment (or test) is where skill and knowledge comes into play. That and knowing what will affect your results.
Can you sort of correct for things? Sure you can, but you have to make sure they do not affect how the driver operates. Otherwise you will get results but they will not reflect the driver in normal operation.
In testing, designing the experiment (or test) is where skill and knowledge comes into play. That and knowing what will affect your results.
Regarding the original questions:
I am using a Behringer 8000 calibrated mic with a Steinberg 2-channel interface.
Being new to this, I was hoping I could get away with on-axis measurements. However, doing desktop speakers for super nearfield and sitting higher than the speakers, vertical polars turn out to be crucial.
I am using a Behringer 8000 calibrated mic with a Steinberg 2-channel interface.
Being new to this, I was hoping I could get away with on-axis measurements. However, doing desktop speakers for super nearfield and sitting higher than the speakers, vertical polars turn out to be crucial.
I guess some people work with very special resistors that have no tolerance or error over time.
I guess these people also never heard of error analysis through a bunch of differential equations either.
If you just do a a very simple monte carlo analysis in your favorite SPICE program, you can see how the series resistor attenuator method will lead to a non-constant voltage over DUT.
This behavior does NOT change with a small 0.1ohm reference resistor.
The effect is only a lot smaller.
This is an inherent FACT and a consequence of any resistor divider
For people who don't believe this, go please show us proper evidence with either a monte carlo simulation or with error analysis via the chain method with differential equations.
If you want to investigate the behavior of BL(x) and Kms(x), that is not usable in my opinion.
It's also not compliant to the AES standard that very clearly tells us to measure with a constant voltage source at 0.1V
As I mentioned already earlier, this series resistor attenuator method is fine for just (simple) crossover filters.
btw, ARTA (LIMP) and most other software programs all use the Z(f)= Rref / ((U1(f) / U2(f)) - 1 ) equation to calculate the impedance.
This equation is baked in and cannot be changed.
Luckily we can abuse the "cable compensation" option to compensate the visuals.
However, that still doesn't change the fact that the voltage is not constant over DUT.
The DC voltage method is a nice idea, but this will also lead to errors since it will change Re.
Which is needed to determine BL and Rms. (Zmax - Re)
It's mostly not usable to determine BL(x) en Kms(x) since it already leads to an offset.
The value is not important here, any series resistance (before or after DUT) will have the same effect.
The effect is only a lot less with a smaller resistor.
How good enough totally depends on what people want to do with it.
But I already said that straight from the beginning that a value around 27 ohm (say between 10 and 39 ohm or so) will be fine for most people.
But again, it all depends what you want to investigate.
Talking about expectations and context is important here.
With the much higher resolutions that most audio interfaces have these days, you can probably get away with a much lower value.
That is exactly what I was trying to say!
Also 100% agree with the last part.
Some people use the expression that "measuring is knowing".
I always like to add that it's only true when "you know what you're measuring".
Meaning that results can still be totally useless if you don't understand what they mean.
Which can lead to potentially serious problems that someone is totally unaware of.
All that being said, let it absolutely not withhold people from starting doing measurements themselves!!!
👍Key here is just staying REALLY consistent and knowing the limitations of the measuring setup.
Get a couple of basics right and you can basically measure the vast majority of things almost like a pro.
The rest will come with knowledge and mostly experience.
So I really encourage anyone to start with it!
Hi b_force,
Yes.
Not everyone can know or have a feel for what their error budget is in a test setup. Most would be horrified to see how inaccurate their test equipment is compared to their expectations. You have to look at tolerances in your equipment under the conditions you are making measurements. Also how long since the last calibration (or even if it was ever calibrated - and by what)?
All we can do is raise awareness so that some might even look at the accuracy specs on their equipment used for test. Then maybe, they can look at what they are measuring and their test setup and understand where errors might creep in.
Finally with all that in mind, they can look at the results they get and try to see just how close their results might be to reality. There are times when the errors are so large, the results are useless. Once they can get better measurements, the math will start to work (expected vs measured performance). Then they can design enclosures and systems that turn out as expected rather than make the assumption the system works as designed when it doesn't.
This isn't easy for a beginner, it is tedious for experienced folks.
Yes.
Not everyone can know or have a feel for what their error budget is in a test setup. Most would be horrified to see how inaccurate their test equipment is compared to their expectations. You have to look at tolerances in your equipment under the conditions you are making measurements. Also how long since the last calibration (or even if it was ever calibrated - and by what)?
All we can do is raise awareness so that some might even look at the accuracy specs on their equipment used for test. Then maybe, they can look at what they are measuring and their test setup and understand where errors might creep in.
Finally with all that in mind, they can look at the results they get and try to see just how close their results might be to reality. There are times when the errors are so large, the results are useless. Once they can get better measurements, the math will start to work (expected vs measured performance). Then they can design enclosures and systems that turn out as expected rather than make the assumption the system works as designed when it doesn't.
This isn't easy for a beginner, it is tedious for experienced folks.
Error-budget, never heard of that word (probably because of differences in language/translation), but I like that word!
Usually it's called knowing your order of magnitude you're interested in.
But, error-budget sounds a lot less overwhelming.
But yes, that is absolutely key.
If people would like to know about what it really means, go read about Fermi problems or estimations!
That is correct, and I have done my fair share of those.
After a while it becomes kinda a 2nd nature/feel thing.
Which is good, because solving these freaking equations every time is no fun.
(Although WolframAlpha helps A LOT these days
)
There is something else to just raw error numbers, and that is consistency and trust.
I would much rather have a meter that is less accurate but much more consistent and works all the time.
So in this case it's not always about being able to measure so super accurate, but more about the fact that you can get that accuracy with very little effort.
Meaning that you don't have to think about it when doing measurements and potentially troubleshooting.
By just brushing it off like "nah, it's not important" you run the risk of not knowing where those boundaries are.
Maybe not important for just a simple passive crossover, but when people want to dive deeper, the nuances are a lot different!
Anyway, I think this is going to much offtopic for @hifijim
The reason why I did mention it, is because the fundamentals are important to keep in the back of your mind when actually doing measurements.
To bring that back on topic again, basically it's saying to work with a good step-by-step-guide.
So the very vast majority of things have been known for decades and decades.
There is plenty of literature and people with experience to fall back on.
The only caveat here, are those pesky T/S parameters from datasheets.
Which can be REALLY annoying when designing just a simple 2-way ported system.
And in the end it sounds and measures totally crap.
Especially for beginners that can be a bitter pill to swallow when everything is already so overwhelming.
Don't worry, it's not you!, it's the pesky datasheets showing weird stuff!!!
I would much rather have a meter that is less accurate but much more consistent and works all the time.
So in this case it's not always about being able to measure so super accurate, but more about the fact that you can get that accuracy with very little effort.
Meaning that you don't have to think about it when doing measurements and potentially troubleshooting.
By just brushing it off like "nah, it's not important" you run the risk of not knowing where those boundaries are.
Maybe not important for just a simple passive crossover, but when people want to dive deeper, the nuances are a lot different!
Anyway, I think this is going to much offtopic for @hifijim
The reason why I did mention it, is because the fundamentals are important to keep in the back of your mind when actually doing measurements.
To bring that back on topic again, basically it's saying to work with a good step-by-step-guide.
- do your preliminary research, so you already know what to expect.
- while doing measurements, immediately look at the results and see if you can make sense of them
For example, do they line-up with estimations from certain resonances etc etc - In general do the measurements make sense.
So the very vast majority of things have been known for decades and decades.
There is plenty of literature and people with experience to fall back on.
The only caveat here, are those pesky T/S parameters from datasheets.
Which can be REALLY annoying when designing just a simple 2-way ported system.
And in the end it sounds and measures totally crap.
Especially for beginners that can be a bitter pill to swallow when everything is already so overwhelming.
Don't worry, it's not you!, it's the pesky datasheets showing weird stuff!!!
Last edited:
