Probably a sizable difference in ambient noise floor too. You don't usually notice that, but it's almost never constant through the day, except in anechoic chambers.
There could be a few dozen things contributing to the difference. Strengthening the panel is one, as you pointed out, noise floor is another. Are you sure you did a fairly major surgury on the sub and put the sub and mic back in EXACTLY the same spot? Are you suggesting that you measurements (including the environment you measured in) are infallible?
This is one data point, not comprehensive proof that a small triangle actually does anything other than what Hornresp predicts.
Not necessarily. We know or at least have a fair idea of what the internals of the TH115 and TH118 look like. Can we say the same about the others? And there's also the question of *when* the designs were created - perhaps he discovered something in-between designs, so we may see features in the newer designs that don't show up in the older ones.
There used to be wire diagrams of all the designs on his site. None of them had any "cone correction". I don't know the dates of any of the designs, but if you don't have any proof that even a single one of DSL's designs has "cone correction" (including the 115/118 as a conscious choice rather than a layout that resembles "cone correction", it's a bit absurd to say that this is DSL's secret ingredient.
Nope, and in any case the measurement process I used is actually pretty impervious to background ambient noise. When I was measuring POC4 for example, it was in the club while I was doing so and there were salsa dancing lessons going on nearby. The measurements I took during that time were quite repeatable, quite likely because it's a close-miked measurement.
Yes 🙂. The room is tiled, and it was pretty simple to line up the front of the sub with a specific tile so it was in exactly the same spot for both measurements. The mic was left in place during the "surgery", so it didn't move either. And all of the electronics were left on and running for some time before the first measurement and until the second one was completed.
In any case, I'm just going to treat it as a curiosity for the moment. I don't have another TH around to check to see if the decrease in distortion happens with it as well when cone compensation is added at S2.
I'd agree with the other poster about the data in post 10
to use it form a conclusion is IMO "touch-n-go"
what is measurement uncertainty of the test up?
The Delany engineer even says "theory cant take you all the way", you maybe on to some of their special sauce tho, doubtful you get to crown a name it yet.
to use it form a conclusion is IMO "touch-n-go"
what is measurement uncertainty of the test up?
The Delany engineer even says "theory cant take you all the way", you maybe on to some of their special sauce tho, doubtful you get to crown a name it yet.
The only thing that changed during the tests was (1) the inclusion of the "cone compensation" and (2) the temperature. I did have to unscrew the driver to insert the cone compensation, so maybe it's screwed in slightly differently too. Everything else was basically identical for both tests.
I understand the minimal changes to DUT but was asking about the accuracy and repeatability of the measurements itself.
in otherwords maybe some more experimental data of the same thing another day by another knowledgeable tester may show a opposite trend due to some unaccounted for variances of instruments and personnel.
in otherwords maybe some more experimental data of the same thing another day by another knowledgeable tester may show a opposite trend due to some unaccounted for variances of instruments and personnel.
Last edited:
I did the second measurement a few times (HolmImpulse allows the comparison of three measurements at a time). There was little or no noticeable difference. The method I use for FR measurements usually results in very repeatable results, once the position of the DUT does not change.
Yes, even screwing the driver in differently (maybe a bit tighter) can cause these changes, along with a few dozen other things. If you didn't measure the noise floor you can't rule that out by just saying it didn't change.
Listen, this is a good experiment, but realistically it's only a single data point. If you can repeat these results a couple dozen times then maybe there's something to this. But I seriously doubt it, not distortion reduction at 30 hz.
There's already pages and pages of "cone correction" stuff posted by people that have never even tried it, and a couple of subjective opinions that don't make a whole lot of practical sense from a couple of people that have tried it. I'd hate to see people latching on to this single measurement of a single experiment and proclaiming that "cone correction" increases power handling, reduces distortion through the whole bandwidth, and sounds better, all claims that have been made based on subjective opinion, and in some cases wild guessing from people that haven't even tried it.
Furthermore, taking this single measurement and extrapolating that "cone correction" is DSL's secret ingredient is just going way too far, when none of DSL's cabs have it. If you can find one case other than the dubious 115/118 case then post it.
DSL doesn't have a secret ingredient other than marketing catchphrases leading people to believe there is some secret. We've got lots of examples to study. They use low compression ratios, good drivers, large, fairly undamped cabs. No magic, nothing that people can't replicate or design themselves using similar guidelines.
There's a lot to be said for this kind of experiment, most people don't even bother. But let's hold off on making judgments just yet, this is only a single measurement of a single experiment so far.
Listen, this is a good experiment, but realistically it's only a single data point. If you can repeat these results a couple dozen times then maybe there's something to this. But I seriously doubt it, not distortion reduction at 30 hz.
There's already pages and pages of "cone correction" stuff posted by people that have never even tried it, and a couple of subjective opinions that don't make a whole lot of practical sense from a couple of people that have tried it. I'd hate to see people latching on to this single measurement of a single experiment and proclaiming that "cone correction" increases power handling, reduces distortion through the whole bandwidth, and sounds better, all claims that have been made based on subjective opinion, and in some cases wild guessing from people that haven't even tried it.
Furthermore, taking this single measurement and extrapolating that "cone correction" is DSL's secret ingredient is just going way too far, when none of DSL's cabs have it. If you can find one case other than the dubious 115/118 case then post it.
DSL doesn't have a secret ingredient other than marketing catchphrases leading people to believe there is some secret. We've got lots of examples to study. They use low compression ratios, good drivers, large, fairly undamped cabs. No magic, nothing that people can't replicate or design themselves using similar guidelines.
There's a lot to be said for this kind of experiment, most people don't even bother. But let's hold off on making judgments just yet, this is only a single measurement of a single experiment so far.
inventing and science is hard work, marketing them is even harder but then the "science" gets buried even further under. Brian maybe on to something but more experiments needed IDK who gets the naming or "branding" credits anyway?
Last edited:
I can safely rule it out because the method I use minimizes its effects. The "noise floor" thing is a non-issue IMO.
That it is, and it should be treated accordingly. OTOH, one measurement is better than no measurements at all, and this one at least demonstrates that the impact of cone compensation can approximated somewhat by HornResp.
The impedance curve for the TH118 actually suggests that it is quite damped. I have not examined the impedance curves for the others in any great detail to draw any conclusions about their damping.
That requires measurement, not opinion. I won't mention this again, as noise floor is only one of dozens of things that could mess up the results.
We already knew this, accurate sims match measurements, sloppy sims don't.
You theorized there was stuffing in there, that would account for a damped response.
Anyway, you don't need to analyze impedance curves to see that DSL cabs are larger and have a spikier (more undamped) response than most diy'er designs, which are usually as small as possible with the flattest (unspiky) response possible.
LOL - I'm not trying to "name" anything. Cone compensation has been discussed several times here before. All I've tried to do with this test is to (1) confirm that it actually makes a difference, and (2) whether or not that difference can be approximated in a HornResp sim. Everything else is just "gravy", and some data, like the distortion decrease observed in my test, might of interest to investigate further. It's curious that it happened, but I'm certainly not going to draw any conclusions over it.
There have been a bunch of other TH builds featured here that did not include cone compensation. Maybe someone with one of those THs could repeat the same experiment that I did, to see if they get similar results.
I should note that I did include "cone compensation" in my POC4, but it was a rough estimate of what might be required. Even then, the measured FR seems to be a better match for the sim I did with cone compensation included, rather than the one without, so maybe we could call this 1.5 tests 🙂.
Brian
so you are basically improving the model by pinching one parameter and then measuring all results mainly increased upper bandwidth, but have discovered other improvement as a side issue? the small increments expected made me comment about measurements.
so you are basically improving the model by pinching one parameter and then measuring all results mainly increased upper bandwidth, but have discovered other improvement as a side issue? the small increments expected made me comment about measurements.
For you to draw a conclusion that the "noise floor" could affect my measurements, you'd have to know the details about how I performed those measurements. As you don't have this information, I think I'll take my own counsel on whether or not the "noise floor" had any significance ;-).
FWIW, here are the relevant details:
1. HolmImpulse configured to use a swept sine wave as the test signal.
2. Mic placed 18" from the sub (basically the width of one tile).
3. Amp's DSP engaged to remove any significant response peaks from 40 Hz to 200 Hz.
3. Volume increased on the sub until HolmImpulse indicated clipping during measurements, then volume was backed off a few dB.
With that setup, you could be banging away on drums in the same room and it wouldn't register in the measurements. It certainly wouldn't produce a broadband decrease in measured distortion between tests.
You're referring to "damping" due to the chosen alignment, not damping due to the addition of stuffing, the latter of which WILL show up in the measured impedance curve, if you know what you're looking for.
Oh, and what's "spiky" about the response of TD's TH412 or TH812? And those are just two of the four or five TH subs I just checked on TD's site.
I'd say I've confirmed HornResp can approximate cone compensation, and I may have discovered a "possible improvement" as a side issue. Can't draw that sort of latter conclusion based on just one test.
I think I'm going to try a full-on "pinch" at S2 for my next test, i.e. perhaps double-up on the cone compensation. Unfortunately the results of that can't be predicted by HornResp.
Im not seeing this confirmation in post 8, I need more magnification on my monitor
you said the new improved model predicts slightly less output at the low end and 2-3 dB more at the upper end ~120-150 Hz
but your test results shows improvement at both ends? +/-1 or 2 dB is tough territory on accuracy of most any measurements esp acoustics
Last edited:
Im not seeing this confirmation in post 8, I need more magnification on my monitor
LOL - it's the scale that's the issue. The improvement is around 2~3dB around 120-150 Hz, and a significantly greater impact around 280 Hz, both of which show up in the HornResp approximation. I've highlighted the areas in question in the attached image.
There's a minor change at the lower end. I wouldn't jump the conclusion yet that it's an improvement. It might be as a result of screwing in the driver more tightly, or even a change in temperature, for example 🙂
Attachments
Whatever. Maybe noise floor has nothing to do with anything, but as I mentioned, there's a few dozen other considerations that could.
I think you know very well what I'm talking about. Most diy'ers shoot for a response like this:
An externally hosted image should be here but it was not working when we last tested it.
That's severely overdamped compared to DSL's tapped horns, no bump at Fb, no bump before the big spike, no undamped bumps anywhere really (except the problematic tapped horn spikes that are very hard to get rid of), because it's overdamped. Very high compression ratio and small horn = response as shown there, the small size and flat(ish) response are very appealing to many diy'ers and MOST sims I've seen are more like this one than DSL's.
- Status
- Not open for further replies.