Hi,
Playing with REW to measure different fullrangers, and probably made some configuration mistake or something, because on the last measurement getting +20dB on pretty neutral speaker such as Dayton PS95-8
Measurement setup:
0,8m measurement distance,
dbx measurement mic, only ~1month old
Presonus interface with 48V phantom power, as sound source also
Topping MX3 amp (I don' trust it!!!)
~10Liters MLTL with ~20cm wide baffle (Yes I know, that is an oversized box for this driver, but this was done for MLTL experiment - the question is about high frequencies anyway)
And the worst part... REW is not on pretty much default settings, it is a bit tuned according to https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_REW.pdf
Measurement levels are not matched, so there is no 120dB levels, everything was done on around ~95dB I think. I made no level matching this time - it is my fault.
There is a ~20 and even ~30dB rising response, from 2.5kHz to 15kHz at 0 angle, 15 and 30deg angles, which seems odd. When compensating it with EQ down this response to similar levels - 10-15dB on 4-15kHz - it sounds off. If I EQ it a little less, from 5-10dB - everything sounds more or less correct...
Measurements on 1/6 oct smoothing, as picture below, also as attachment.
When looking at various PS95-8 tests, benchmarks and so on - it is pretty much neutral speaker, there are no measurements like mine.
The problem is that I have pretty good ears, not fully "golden" but pretty close to it. I can hear up to 14-15kHz, and definitely from 5-10kHz, done many hearing tests online with pretty good results.
So I somehow started to doubt my setup. Tomorrow will measure all the available speakers and other amps at hand, now it is too late. Have TC9 at hand, and measured it before - there were no such rising response as I remember.
The question is:
My ears assumed, that this massive 10-20dB rising response is "nothing extraordinary", and I should trust measurements?
Setup is wrong? REW?
Sound level is too high and amp adds some kind of protection?
Mic is bad?
Level matching is ok, and that little speaker generated those 120+ dB sound levels????
What are your ideas?
Playing with REW to measure different fullrangers, and probably made some configuration mistake or something, because on the last measurement getting +20dB on pretty neutral speaker such as Dayton PS95-8
Measurement setup:
0,8m measurement distance,
dbx measurement mic, only ~1month old
Presonus interface with 48V phantom power, as sound source also
Topping MX3 amp (I don' trust it!!!)
~10Liters MLTL with ~20cm wide baffle (Yes I know, that is an oversized box for this driver, but this was done for MLTL experiment - the question is about high frequencies anyway)
And the worst part... REW is not on pretty much default settings, it is a bit tuned according to https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_REW.pdf
Measurement levels are not matched, so there is no 120dB levels, everything was done on around ~95dB I think. I made no level matching this time - it is my fault.
There is a ~20 and even ~30dB rising response, from 2.5kHz to 15kHz at 0 angle, 15 and 30deg angles, which seems odd. When compensating it with EQ down this response to similar levels - 10-15dB on 4-15kHz - it sounds off. If I EQ it a little less, from 5-10dB - everything sounds more or less correct...
Measurements on 1/6 oct smoothing, as picture below, also as attachment.
When looking at various PS95-8 tests, benchmarks and so on - it is pretty much neutral speaker, there are no measurements like mine.
The problem is that I have pretty good ears, not fully "golden" but pretty close to it. I can hear up to 14-15kHz, and definitely from 5-10kHz, done many hearing tests online with pretty good results.
So I somehow started to doubt my setup. Tomorrow will measure all the available speakers and other amps at hand, now it is too late. Have TC9 at hand, and measured it before - there were no such rising response as I remember.
The question is:
My ears assumed, that this massive 10-20dB rising response is "nothing extraordinary", and I should trust measurements?
Setup is wrong? REW?
Sound level is too high and amp adds some kind of protection?
Mic is bad?
Level matching is ok, and that little speaker generated those 120+ dB sound levels????
What are your ideas?
Last edited:
Member
Joined 2003
Impulse is just a noisy mess, check your equipment and set up before proceeding.
Primary peak of the mess is at t = -2.3ms, but also peak at t=0, and t=-4.74ms, as well as very low SNR
For comparison, here's one of my own measurements of a small fullrange driver. First reflection is at 6.5ms.
Primary peak of the mess is at t = -2.3ms, but also peak at t=0, and t=-4.74ms, as well as very low SNR
For comparison, here's one of my own measurements of a small fullrange driver. First reflection is at 6.5ms.
Not at all! You can even see it in the spec. The impedance and the response.
But it can sound very good. I got a few as first drivers 🙂 I was a complete noob (check this thread for some measurements by others, not mine!) and I had issues with:
- Dubious quality amp
- Getting distortion for low frequency without high pass
- Maybe bad box
- Not knowing how to EQ
Mostly because consistent xrk971's praise for the driver and after learning how to use tools I gave them a shot, and:
- Great for open baffle (with a woofer to help)
- Great as full-range drivers with a sub
In both cases with good EQ. Less is better. I use EasyEffects -- REW now lets you listen/measure with the EQ settings you're creating.
Last time I paired them with two small experimental PPD subs for the PC and I got great sound, I could not believe it.
That first thread about the driver shows up in Google often and I have been planning to update because it's a bit unfair. I have the front pieces for better boxes but I haven't set aside the time to finish the build.
Attachments
Yes. Probably yesterday was a bad day for measurements - in such a simple task so many facepalms:
Loopback cable was in wrong socket... Also some other stuff I missed.
Current measurements:
Yes, it is a VERY neutral speaker. It needs only BSC circuit or EQ from 5 to 13-14kHz, only babies and pets can hear above.
I have many at hand, even so much praised TC9 - this one is more to my taste. Especially strong and pleasant is the voice zone.
And it is best at ~30 degrees
Have read your post. Amps have very very little to the overall distortion levels. Even the cheap class D ones are pretty good overall. Yes, it is not good driver for bass. Box is irrelevant, only the size and stuffing of it will define low end response, and the baffle size - the boost/loss at some frequencies. EQ is the last resort, I am using EQ APO on windows.
And yes, subwoofer for 120Hz and below is very recommended.
What calibration file are you using with the mic? A rising high frequency response is what is typically seen with an uncalibrated mic.
...I am using mic, which has no calibration file.
Full name and model:
DBXRTA-M
dbx RTA Measurement Microphone
Commercial calibration costs are typically more than a cheap measurement mic. If you have a tweeter or two to hand then one can often DIY a reasonable correction for the high frequency bump from averaging the difference between the measured and published curves. If the corrections are similar with different tweeters that builds confidence. The high frequency end of a wideband driver is likely to vary a bit too much from unit to unit to be particularly reliable but it may tell you something.
Another thought. If you point the mic at 90 degrees it is likely to measure flatter though it will roll-off at some point. Still best to check against something reasonably known like the response of tweeters though.
You got me really really worried.
Is there a place in EU which calibrates microphones?
I have couple tweeters, but I have trust in them even less than in that microphone. Rising response is the thing of PS95, but the question is - how much?
Probably yes. I need to find the manual of this mic, maybe it is 90deg type mic.
Lots and lots of thought. Right now the idea to buy another mic came to mind.
I already have it on Amazon Germany shopping cart. More expensive than in US.
Ok, another questions:
1. Can pretty reputable manufacturer such as dbx (Harman group) sell mic, which they say is without cal file, but is described as "RTA-M Reference microphone"? Can it be trusted????!!!
2. All those mic are said to be omnidirectional, maybe on some of them measurements should be measured on 90deg angle?
Now this is not a measurements questions, but of a trust.
There's nothing to be worried about since all measurement microphones require calibration in order to provide accurate measurements and so there are plenty of places that provide a service. A recent calibration by an accreditted service is usually a requirement for sound measurements to be legally acceptable. There are plenty of places around but for speaker DIY using a generic Asian mic costing £20-100 depending on branding it is common to get by with an estimated one or to accept the calibration curve that sometimes come with a cheap microphone (the one with mine was obvious nonsense but some are likely to be in the right ballpark). Borrowing a calibrated mic and comparing against it is a simple approach and is pretty much what calibration services do. If you shop around you may find a service for £50 but most will cost more like £150. The cheaper ones are unlikely to be accreditted though.
Mainstream driver manufacturers know how to take accurate measurements although they can do odd things with their setup so one needs to find out how the measurements were performed to interpret their curves. A small decent quality tweeter will have a largely pistonic (hard) or controlled bending (soft) in the 5-15kHz region and is likely to be reasonably reproducible (unlike the strongly resonant high frequency response of larger wideband drivers). If the bumps and dips line up in frequency with your measurements that builds confidence. If several different tweeters produces pretty much the same difference curve then that also builds confidence. It won't be as accurate as a calibration curve from an accredited service but for DIY it is likely to be fine. I didn't consider the cost worthwhile to calibrate my cheap microphone after the calibration curve that came with it proved to be wrong. A microphone's calibration curve changes with time and so the ones used professionally tend to be recalibrated every 1-2 years. Measurements also change with atmospheric conditions which will be corrected for professionally, with angle of incidence at high frequencies, etc...
They are being honest in telling you it is not calibrated. Calibrating properly involves significant time dedicated to each microphone and so may not be commercially worthwhile for cheap microphones. I don't know how the "calibration curves" that sometimes come with cheap microphones are created (anyone?) but they are unlikely to be particularly reliable.
The orientation of the mic depends on the type of soundfield, what you want to measure and which orientations are calibrated (it is common to calibrate only a few orientations and sometimes only one). It is common for uncalibrated microphones to have a signficantly smaller deviation from flat at 90 degree compared to 0 degrees. However, if you want accurate measurements then both need calibrating.
There is no way I would trust some mainstream tweeter than some mic. I do not have some state of the art like Bliesma or Seas at hand, so at this moment I would trust mic more. Even if this is not calibrated mic. There are no such peaks on other full range speaker, I will retest them and post here.
I think I decided to buy Dayton mic. Even if it is (probably) lower class, but has some calibration. I hope that this calibration will be somewhat close to real life.
Also, a comparison between cheaper+calibrated and average+uncalibrated would be interresting.
I think I decided to buy Dayton mic. Even if it is (probably) lower class, but has some calibration. I hope that this calibration will be somewhat close to real life.
Also, a comparison between cheaper+calibrated and average+uncalibrated would be interresting.
This should be a good (not too expensive) place to start:
https://shop.hifi-selbstbau.de/produkt/mikrofonkalibrierung/
Maybe you need a translator.
Fair enough but your reasoning makes little sense to me. BTW my microphone with an incorrect calibration curve was a Dayton. To be fair when I contacted Dayton they offered to pay for a proper calibration but only at a place in the US they presumably had a business relationship with. A reasonable response but having dealt with shipping to and from the US in the past the time, effort and likely customs costs and hassle just didn't seem worth it for a low cost microphone I could DIY calibrate.
It is the cheapest I have seen but note the shipping issues. Might be worth a phone call though.
I know ~20-50 words in German, Not enough for any practical conversation.
Anyway, the service pricing is reasonable, probably an option is to buy another calibrated mic.
Anyway, the service pricing is reasonable, probably an option is to buy another calibrated mic.
It makes all the sense in the world!
I will ask in other words: is there a tweeter, in reasonable price range - lets say similar to the measurement mic, that is up to 60EUR/unit, 120 EUR/pair, which has in real life freq curve the same, as manufacturer specified freq curve, displayed with minimal smoothing, measured in conditions, which I can replicate? And I am talking not about how good it sounds, and so on - the only requirement is the willingness of manufacturer to provide accurate and good resolution data in specification docs.
My opinion is that I will trust more that uncalibrated, yet not very cheap, mic than some tweeter. Probably the one and only exception can be overall raise/drop in very broad range, but I will see this measuring any cheap-ish 3-4 fullrange drivers, I have many different models of them.
Member
Joined 2003
I use Line Audio Omni1. It is "good enough" without requiring a calibration file, and one of the best value performers for taking distortion measurements. Made in Sweden.
http://www.lineaudio.se/Omni1.html
Dbx provides no accuracy specification for the RTA mic, but some basic chart is provided here:
https://dbxpro.com/en/product_documents/rta-m_cutsheet-pdf
I would assume, that it's intended use is for room EQ, where the frequencies of interest are <2kHz, so nonlinearity >5kHz is not incredibly important.
http://www.lineaudio.se/Omni1.html
Dbx provides no accuracy specification for the RTA mic, but some basic chart is provided here:
https://dbxpro.com/en/product_documents/rta-m_cutsheet-pdf
I would assume, that it's intended use is for room EQ, where the frequencies of interest are <2kHz, so nonlinearity >5kHz is not incredibly important.