Quick test results, driver changes with heat

[avaserfi]

A more in depth comparison

I just spent some time taking a more in depth set of measurements. Using a calibrated WT3 and a Dayton RS150 I took 5 measurements with a 'cold' driver and 5 measurements with a 'warm' driver each after 30 minutes of a 100Hz sine wave. I averaged the information and found total percent difference as well as modeling the two different averages in WinISD Pro.

The compiled data:

The modeling [Blue is warm - Yellow is cold]:

Upper end frequency response is not shown due to inaccuracy of modeling. It is highly unlikely that there is any audible differences between 'warm' and 'cold' drivers even if there are minor measurable differences.

Please note even this set of data is far from statistically complete, but it is a more sufficient example than that offered by the OP due to larger sample size.

 

[Swerd]

Well that settles that. Thanks. You did a 30 minute warm up just to be sure they were warm, but what is the least time needed for that?

 

[no. 5]

I realize the OP was talking about changes to a driver under 'normal' listening conditions, but the thread is "driver changes with heat", what if that is continued to higher listening levels; wouldn't it be useful to know how how a driver changes (especially frequency response) as its voice coil really heats up?

 

[avaserfi]

Well that settles that. Thanks. You did a 30 minute warm up just to be sure they were warm, but what is the least time needed for that?

This would depend on a few factors such as driver, SPL, source material, etc... I used 30 minutes simply because the OP did.

I realize the OP was talking about changes to a driver under 'normal' listening conditions, but what if that is continued to higher listening levels; wouldn't it be useful to know how how a driver changes (especially frequency response) as its voice coil really heats up?

With a properly designed cone driver, at high SPLs, there is the possibility for minor power compression. Otherwise at high SPLs it is likely that audible distortion will be introduced before T/S specifications are sufficiently changed to effect frequency response. That being said it is unlikely, if proper drivers for the specific application are chosen, that this will be an issue. Of course if someone sends too much constant power to a driver far greater issues will arise, such as driver destruction.

 

[TLS Guy]

This would depend on a few factors such as driver, SPL, source material, etc... I used 30 minutes simply because the OP did.



With a properly designed cone driver, at high SPLs, there is the possibility for minor power compression. Otherwise at high SPLs it is likely that audible distortion will be introduced before T/S specifications are sufficiently changed to effect frequency response. That being said it is unlikely, if proper drivers for the specific application are chosen, that this will be an issue. Of course if someone sends too much constant power to a driver far greater issues will arise, such as driver destruction.

You have hit it on the nail. The only significant parameter is the increase in resistance of the coil with heating, which causes thermal dynamic range compression. That would not show up on the condition of test with a steady state signal.

That is the reason for large motor systems in higher priced drivers, especially those used in high power systems. In big peaks the coil can heat quickly, and with it the resistance, and hence the thermal dynamic range compression. This affects all moving coil drivers to some degree. There are quite large instantaneous voice coil temp changes with the dynamic range of the music. This effect is minimized with large motor systems and vented pole pieces.

 

[Spkr_Bldr]

You guys are nit-picking way too much here, that wasn't the point at all. The point is params DO change and should be tested under operating conditions. I'm going to dig out some bigger woofers to test like some of my AV15's. I did this before/after to them about 6 months ago, and it was a dramatic change! I chatted with Keith Larson (who made the Woofer Tester) about this, this is what he said in regards to the importance of testing in a way that can reveal variances - his WT Pro has much more capability in this regard, and I'm planning on getting one:

DC offset testing (you need a DC servo amp) reveals mechanical voice coil and suspension offsets, but does not show how a driver performs with an AC signal. If you look at the graphs posted on our web page you will see that Fs dropped by an astonishing 35% (this is not uncommon).
As the spider and roll 'heat' from mechanical excursion, Cms decreases.
'Heat' is however a poor term since chemical bonding is also apparently involved. Never the less, the time lag has the characteristics of heating, thermal capacity and the like. What I am pointing out here is that this is NOT revealed with DC offset testing.

 

[avaserfi]

You guys are nit-picking way too much here, that wasn't the point at all. The point is params DO change and should be tested under operating conditions. I'm going to dig out some bigger woofers to test like some of my AV15's. I did this before/after to them about 6 months ago, and it was a dramatic change! I chatted with Keith Larson (who made the Woofer Tester) about this, this is what he said in regards to the importance of testing in a way that can reveal variances - his WT Pro has much more capability in this regard, and I'm planning on getting one:

Parameters might change, but the changes are typically not drastic enough to alter cabinet design such that effects are audible. This was the point of my example in post 21.

As far as members being critical of your results this is most likely because many of the forum members require credible evidence to make judgments*. A single measurement of one driver is anything but credible evidence for any scenario.

*Please note I cannot speak for the forum as a whole, but rather am making this statement based on my observations.

 

[Spkr_Bldr]

Parameters might change, but the changes are typically not drastic enough to alter cabinet design such that effects are audible. This was the point of my example in post 21.

I don't remember exactly what changes I saw when I did this with my AV15 - I should be able to dig them out and test again on Monday - but the 20% Qts change seems about right from what I remember.

[quoteAs far as members being critical of your results this is most likely because many of the forum members require credible evidence to make judgments*. A single measurement of one driver is anything but credible evidence for any scenario.[/QUOTE]

I never asked anybody to make judgements. Maybe if you have a bigger driver handy, try the same with it. All of my sub drivers are buried in a basement closet right now.

I'm going to order a Woofer Tester Pro soon since Keith has offered the best communication and support of all the options (Clio, Praxis, WT Pro). And when I get it we can investigate this with more detail. Keith has convinced me that testing at a drive level that's more representative of what you can expect with normal use is very important.

 

[Swerd]

You guys are nit-picking way too much here, that wasn't the point at all. The point is params DO change and should be tested under operating conditions...

Please don't take this wrong. This isn't nit-picking. We are not doubting that measured T/S parameters may change depending on the temperature and possibly other conditions encountered during use. What we are saying is that you haven't yet clearly established that. Your original post was fine until your very last sentence.

When you post simple observations, think about how you word things so that you don't make (or seem to make) broad conclusions that can mislead other readers. I think you clearly understand what you meant to say in your original post. But I think your words didn't exactly say what you meant.

I never asked anybody to make judgements.

When you post it on a public forum like AH, that is just what you asking others to do.

 

[TLS Guy]

I don't remember exactly what changes I saw when I did this with my AV15 - I should be able to dig them out and test again on Monday - but the 20% Qts change seems about right from what I remember.

[quoteAs far as members being critical of your results this is most likely because many of the forum members require credible evidence to make judgments*. A single measurement of one driver is anything but credible evidence for any scenario.

I never asked anybody to make judgements. Maybe if you have a bigger driver handy, try the same with it. All of my sub drivers are buried in a basement closet right now.

I'm going to order a Woofer Tester Pro soon since Keith has offered the best communication and support of all the options (Clio, Praxis, WT Pro). And when I get it we can investigate this with more detail. Keith has convinced me that testing at a drive level that's more representative of what you can expect with normal use is very important.[/QUOTE]

This whole situation is mute. Even if there are changes what are you going to design for? Will it be Vaughn Williams Symphony No. 6, largely quiet music, or a rousing performance of the Beethoven No. 9? I don't think this a fruitful line of endeavor and is only going to be of academic interest.

My point is the suspension changes are unlikely to be significant. The current through the voice coil varies enormously from milli watts to a hundred watts or more. You know how fast a light bulb heats up with a hundred watts. Unless you know how to stop wire heating with a current going through it, other than cooling it close to absolute zero, all you can do is choose good quality drivers that get heat away from the voice coil as fast as possible.

 

[Spkr_Bldr]

This whole situation is mute. Even if there are changes what are you going to design for? Will it be Vaughn Williams Symphony No. 6, largely quiet music, or a rousing performance of the Beethoven No. 9? I don't think this a fruitful line of endeavor and is only going to be of academic interest.

My point is the suspension changes are unlikely to be significant. The current through the voice coil varies enormously from milli watts to a hundred watts or more. You know how fast a light bulb heats up with a hundred watts. Unless you know how to stop wire heating with a current going through it, other than cooling it close to absolute zero, all you can do is choose good quality drivers that get heat away from the voice coil as fast as possible.

I disagree, the driver will change - in some cases dramatically - under operating conditions. Mids and tweeters it's not likely to matter nearly as much if at all compared to the bass end of a system, and in sealed systems it's not likely to significantly matter either. But in any form of a vented system the box tuning reacts greatly to driver parameter changes. Like I said earlier, a 20% change in Qts will make a 3db difference at the corner frequency ... and if Keith Larson is telling me he routinely sees 35% changes, then I'm going to try and find ways of tuning my box to perform how I want under realistic drive levels. These changes are due to a lot more than just heating of the coil and materials, chemical bonding and changes in compliance of the suspension are likely more a factor as hinted at by Keith.

 

[avaserfi]

I disagree, the driver will change - in some cases dramatically - under operating conditions. Mids and tweeters it's not likely to matter nearly as much if at all compared to the bass end of a system, and in sealed systems it's not likely to significantly matter either. But in any form of a vented system the box tuning reacts greatly to driver parameter changes. Like I said earlier, a 20% change in Qts will make a 3db difference at the corner frequency ... and if Keith Larson is telling me he routinely sees 35% changes, then I'm going to try and find ways of tuning my box to perform how I want under realistic drive levels. These changes are due to a lot more than just heating of the coil and materials, chemical bonding and changes in compliance of the suspension are likely more a factor as hinted at by Keith.

One is welcome to hold their beliefs, but in a case where data can be quantified and the experiment can be repeated [such as this] it is typically hard to convince others without actual credible evidence. Thus far only conjecture and personal belief have been offered as evidence along side measured data that does agree with the original hypothesis [both yours and mine].

 

[annunaki]

True woofer analysis has always been an area of debate to some. I would suggest that if one is after thorough analysis of any said driver they should have it Dumax tested or run through a Klippel analyzer. That way one gets a look at the dynamic properties of said driver. Couple that with doing said testing of T/S parameters with the driver at temperature so there is an accurate model.

As avaserfi has clearly shown, it appears that performance differences, in nearly all cases, be inaudible. I believe that proper non-resonance cabinet construction and a well executed crossover will play a much more integral part than a very slight deviation of T/S parameters.

 

[Spkr_Bldr]

Here you go, a HiVi D10.8 was on top of the stack in the closet so I pulled it out. Feel free to change your mind

Pre-warming

Re         =    6.5588 ohms
Fs         =   28.1953 Hz
Zmax       =  116.4173 ohms
Qes        =    0.5102
Qms        =    8.5458
Qts        =    0.4815
Le         =    1.3217 mH  (at 1 kHz)
Diam       =  235.0715 mm  (  9.2548 in  )
Sd         =43399.9971 mm^2( 67.2701 in^2)
Vas        =  116.0020 L 
BL         =   13.2375 N/A
Mms        =   73.4965 g
Cms        =  433.5322 uM/N
Kms        = 2306.6338 N/M
Rms        =    1.5236 R mechanical
Efficiency =    0.4788 % 
Sensitivity=   88.8200 dB @1W/1m
Sensitivity=   89.6826 dB @2.83Vrms/1m

Post-warming (I must have stopped the sweep early, hence Le = 0)

Re         =    6.7228 ohms
Fs         =   25.1953 Hz
Zmax       =  117.1426 ohms
Qes        =    0.4471
Qms        =    6.8442
Qts        =    0.4105
Le         =    0.0000 mH  (at 1 kHz)
Diam       =  235.0715 mm  (  9.2548 in  )
Sd         =43399.9971 mm^2( 67.2701 in^2)
Vas        =  136.4520 L 
BL         =   12.9297 N/A
Mms        =   73.2422 g
Cms        =  447.5695 uM/N
Kms        = 2234.2900 N/M
Rms        =    1.5140 R mechanical
Efficiency =    0.4698 % 
Sensitivity=   88.7376 dB @1W/1m
Sensitivity=   89.4930 dB @2.83Vrms/1m

 

[TLS Guy]

Here you go, a HiVi D10.8 was on top of the stack in the closet so I pulled it out. Feel free to change your mind

Pre-warming

Re         =    6.5588 ohms
Fs         =   28.1953 Hz
Zmax       =  116.4173 ohms
Qes        =    0.5102
Qms        =    8.5458
Qts        =    0.4815
Le         =    1.3217 mH  (at 1 kHz)
Diam       =  235.0715 mm  (  9.2548 in  )
Sd         =43399.9971 mm^2( 67.2701 in^2)
Vas        =  116.0020 L 
BL         =   13.2375 N/A
Mms        =   73.4965 g
Cms        =  433.5322 uM/N
Kms        = 2306.6338 N/M
Rms        =    1.5236 R mechanical
Efficiency =    0.4788 % 
Sensitivity=   88.8200 dB @1W/1m
Sensitivity=   89.6826 dB @2.83Vrms/1m

Post-warming (I must have stopped the sweep early, hence Le = 0)

Re         =    6.7228 ohms
Fs         =   25.1953 Hz
Zmax       =  117.1426 ohms
Qes        =    0.4471
Qms        =    6.8442
Qts        =    0.4105
Le         =    0.0000 mH  (at 1 kHz)
Diam       =  235.0715 mm  (  9.2548 in  )
Sd         =43399.9971 mm^2( 67.2701 in^2)
Vas        =  136.4520 L 
BL         =   12.9297 N/A
Mms        =   73.2422 g
Cms        =  447.5695 uM/N
Kms        = 2234.2900 N/M
Rms        =    1.5140 R mechanical
Efficiency =    0.4698 % 
Sensitivity=   88.7376 dB @1W/1m
Sensitivity=   89.4930 dB @2.83Vrms/1m

I haven't changed my mind. Those drivers do not have a robust motor system like a Dynaudio, Morel, or SEAS Exel range. What values you choose for your design parameters is still going to arbitrary in the end. Peoples listening habits vary from individual to individual, and individuals vary their listening habits all the time. Anyhow I still want to know which symphony you are going to design for, the VW No. 6 or the Beethoven 9. I can tell you there will be a chasm in the thermal effects between the two.

I agree with Annunaki, that there are more fruitful lines of endeavor, such as crossovers and I would suggest active powered speakers.

 

[avaserfi]

Again using your non-generalizable sample [due to small n] the results would likely only be audible if one was listening in a completely reflection free environment. Please note, in no way am I implying this data is usable, in fact it is most likely not, a far more reasonable approach would be doing exactly what I did in post 21 to minimize variation due to extraneous variables. For the sake of furthering this discussion I will still analyze this data. Also, if I have a chance will measure a larger driver I own this weekend in a more scientifically useful fashion. Please note the driver I plan on measuring has been shown to be extraordinarily linear and well designed it also happens to be far more robust than the HiVi D10.8.

Sealed with both cabinets modeled to the same exact size [blue is warm]:

Note the f3 difference is only 5Hz

Ported both cabinets are modeled exactly the same, including tuning frequency [blue is warm]:

Note the f3 difference is only 4Hz

In both cases response difference is minimal the only difference being a slightly extended response. At the frequencies being discussed the room will have a far greater effect on response rather than this slight extension in f3. Please note the OPs hypothesis seems to have been revised [in post 31] such that smaller drivers are not actually effected by this supposed phenomenon making issues at higher frequencies moot.

 

[Spkr_Bldr]

I haven't changed my mind. Those drivers do not have a robust motor system like a Dynaudio, Morel, or SEAS Exel range. What values you choose for your design parameters is still going to arbitrary in the end. Peoples listening habits vary from individual to individual, and individuals vary their listening habits all the time. Anyhow I still want to know which symphony you are going to design for, the VW No. 6 or the Beethoven 9. I can tell you there will be a chasm in the thermal effects between the two.

I agree with Annunaki, that there are more fruitful lines of endeavor, such as crossovers and I would suggest active powered speakers.

So you're saying that proper box tuning doesn't matter to you?

It's pretty simple what I'm saying, even if you refuse to acknowledge it - get the driver up to operating temperature and conditions, then measure it. No matter how you define 'operating temperature', the results you get are going to be more indicitave of how the driver will perform under real-world use. Of course crossover design is more important than getting the box tuning spot on, but taking proper measurements of simple T/S parameters should still be done and in no-way precludes any of the other things you 'should' be doing.

I'm guessing you've never seen one of the HiVi D woofers, they're essentially Dynaudio/Morel 4" voice coil clones. They have extremely robust motor assemblies.

 

[Spkr_Bldr]

In both cases response difference is minimal the only difference being a slightly extended response. At the frequencies being discussed the room will have a far greater effect on response rather than this slight extension in f3. Please note the OPs hypothesis seems to have been revised [in post 31] such that smaller drivers are not actually effected by this supposed phenomenon making issues at higher frequencies moot.

Why are you guys being so stubborn about this? Clearly driver parameters will change, so what's the big deal with striving to get the best data you can?

So a 4db and 5db change in F3 is minimal to you? I hardly think so. Of course other things can have larger impact, but what does that have to do with this issue?

My findings, not a hypothesis, haven't changed. The simple fact is T/S params just don't matter much except when considering box tuning, so of course changes evidenced in mids/tweeters would have little if any impact. Combined with changes being greater in larger drivers, the impacts on box tuning is worth considering.

 

[TLS Guy]

Why are you guys being so stubborn about this? Clearly driver parameters will change, so what's the big deal with striving to get the best data you can?

So a 4db and 5db change in F3 is minimal to you? I hardly think so. Of course other things can have larger impact, but what does that have to do with this issue?

My findings, not a hypothesis, haven't changed. The simple fact is T/S params just don't matter much except when considering box tuning, so of course changes evidenced in mids/tweeters would have little if any impact. Combined with changes being greater in larger drivers, the impacts on box tuning is worth considering.

I guess the point is that it is better to be more nearly correct than precisely wrong. The dynamic range of music is 130 db commonly 120db. So just look at the power ranges, and as a consequence the thermal effect will have a huge statistical scatter. I don't know what operating temperature is and nor do you.

Bottom line, your endeavors are of academic interest only. Their practical significance is minimal to none and the onus is on you to prove otherwise. Your data is not new, or some starting revolution in loudspeaker design. It is of interest only, without a practical application, even if it is audible. Because of the inevitable huge scatter over time this data has no application.

If you think it does then answer my question, which of those two symphonies are you going to design for, and it can't possibly be both.

Without acknowledging this point the argument is ridiculous.

 

[mtrycrafts]

.. What are the temperatures of the voice coil, and have they reached a steady-state?

Well, that would be fine and dandy for a test bench condition where you put on ear plugs and run it at a constant power, but, with music and movies, well, it may cool down between cannon shots of T-Rex stomps What then?