Post by Owen Y on Sept 28, 2018 16:35:36 GMT 12
These loudspeakers were built over 10 years ago & modification, tweaking has been done in-room largely by ear. My ears tell me that they sound 'OK' but there are a number of areas that I would like to optimise, eg: - maximise bass output - check for & try to iron out any in-room peak & troughs in response. - improve HF balance (it's hard to find a HF driver to match the high sensitivity of a mid-horn). - improve sound-staging (ie stereo effect). Some background: I had studied various speaker-testing freeware-shareware over recent years (eg REW, HOLM Impulse...) but ideally you needed a calibrated quality microphone. I had gone as far as building a mic preamp + one of those Panasonic capsule mics popular with DIYers. But, this sort of thing was soon superseded by self/phantom-powered USB mics. Then there was the problem of a suitable computer, ideally a mobile laptop with a decent soundcard. (Without borrowing the kids' macbooks ) However, technology moves faster than me & now, you can test your speakers using just a smartphone loaded with an inexpensive smartphone app. (As seen on the Show your system page, the loudspeakers currently look like this..)
|
Post by Owen Y on Sept 28, 2018 16:44:05 GMT 12
The design concept was like this... The idea was to generate bass & midrange 'full-range' sound using only one driver. And high sensitivity, high SPLs, enabling the use of low powered triode amplification. (Front view) (Floor plan view below, against front wall...)
|
Post by colinf on Sept 29, 2018 7:09:01 GMT 12
“- improve HF balance (it's hard to find a HF driver to match the high sensitivity of a mid-horn). - improve sound-staging (ie stereo effect).” Here’s where impulse response plays a part. I’ve found that when the time alignment and level of the two drivers are matched the soundstage becomes transparent. I sometimes use a short square half wave pulse put through a small dc coupled amp, and watch the speaker's response from a microphone connected to an oscilloscope. It’s reasonably easy to see when the phases and level are aligned. Nice cabinetry by the way!
AMR-iFi R&D
|
Post by jon on Sept 29, 2018 10:41:40 GMT 12
What does Hornresp show for your cabinet?
|
Post by Owen Y on Sept 29, 2018 13:14:59 GMT 12
Thanks colinf - that may come in handy soon. jon - good Q - firstly I simulated the front midrange horn, then the basshorn which is a lot trickier, because of the variables involved.....
|
Post by Owen Y on Sept 29, 2018 13:55:02 GMT 12
The midrange horn was relatively straightforward: 1) Tractrix flare profile as derived & patented by Paul Voigt in 1927! (Based on the idea that sound wavefronts within a horn must be spherical & maintain the same radius from throat to horn-mouth.) The sonic efficacy of this profile was endorsed by Dinsdale in '74 & later by Bruce Edgar (who is largely responsible for modern advancement of horn speaker design & whom I met in LA in the 90s). 2) The design cut-off frequency (Fc) was 140Hz - so you can expect output down to within 1/2 to 1/3 of an octave , ie 180-200Hz 3) I added my own wide, gentle edge roll, for more gradual mouth pressure transition. 4) The driver used at the time was a Lowther PM2C (derived from Voigt's original 1930s light paper-coned, double voice-coil wound, high-strength magnet, horn drivers). You can read more about the design & construction here.
|
Post by Owen Y on Sept 29, 2018 14:44:02 GMT 12
|
Post by Owen Y on Sept 29, 2018 16:03:44 GMT 12
(The mid-horns are shown sitting atop Mauhorn cabinets, but the above plots were done sans bass cabinet connected.) Freq response back then was not real flash: - Broad lift in the 5k-12kHz upper mid-lower treble, upper 'brightness' region (a well known Lowther trait). - Pretty wide midrange output down to around 200Hz up to 13k-15kHz or so - not bad for a single driver. - A mysterious 1kHz dip - probably cancellation between driver rear output & front horn
|
Post by Owen Y on Sept 29, 2018 16:32:43 GMT 12
The bass-horn gets more complex. The design ideas were: - A full -range 'compound' front mid + rear bass horn using a single driver (no crossover). - Good LF output down to say 30Hz or lower. - Utilise 'corner-loading' so that the bass horn-mouth can be reduced in size by a factor of 8. - A shallow depth cabinet so that the bass horn-mouths could end up in close coupling to the room corners. - At the same time, the front mid-horn should end up located in a reasonably 'conventional' stereo spacing, away from the corners. - Internal horn profile as smooth as possible (without distortion-generating sharp bends) - A driver chamber that would allow for chamber-volume fine-tuning (enabling what Bruce Edgar called 'throat-annulling', for best bass response). - And of course practical to hand-build (eg in 2 parts). (At the time, I was sharing ideas with a pal in Sth Africa, 'Winding Man', who also built his own version of a 'compound' mid + bass horn, below )
|
Post by Owen Y on Sept 29, 2018 21:01:39 GMT 12
In around 2012, I switched from Lowther drivers (too costly & high maintenance), to SEAS FA22RCZ drivers (1/7 the price & excellent build quality) - both high sensitivity, paper-coned drivers with 'whizzer' cones. Lowther PM4: SEAS FA22RCZ (on the right): The drivers on the left are 'Old Brown Dog' branded Field Coil (powered electromagnet) drivers made by 'Winding Man'. These have handmade stiff paper cones & 30 Watt wire coils which require a 145Vdc PSU to energise the electromagnet. (The diaphragm cone frames/baskets are re-purposed Lowther frames.)
|
Post by Owen Y on Sept 30, 2018 16:18:19 GMT 12
Around this time, I decided that: - Running a driver full range tends to expose the cone-break-up region & so.... - I inserted a simple low-pass bass-coil to roll off the top end + added a tweeter via a high-pass cap. (With Solen Split, the selected crossover point is at -6dB (not -3dB) and with the tweeter connected in reverse phase, the response at crossover point is almost flat, with only a slight -1.2dB dip. [Ref. Douglas Self: The Design of Active Crossovers] ) Tweeter - it is difficult to find a (affordable) good tweeter that can match the sensitivity of the mid/bass horns. The following tweeters were tried: - Fostex T90A - OK but not sensitive enough - Beyma CP12/N - OK but not as sweet as the AlNiCo Fostexes.
|
Post by Owen Y on Sept 30, 2018 19:03:11 GMT 12
Freq response was looking better: (Although lumpy bass response.) As said, I had ditched the Lowthers & tried the inexpensive SEAS FR22RCZ drivers. Fast- forward to 2017 (with favourable NZD exch rates), I sprung for Fostex T925A tweeters (bigger magnet, more sensitive versions of the T90A). And Freq Response: I wouldn't read too much into comparisons between the above 2 plots. They were taken 7 years apart & I can't recall the positioning of the SPL meter in to top plot. Also the driver chamber volume has never been optimised properly & using more accurate, repeatable measuring techniques more readily available now. That brings us up to date: I will make better measurements now with a better measuring set-up - software & hardware bits now more inexpensively available.
|
Post by Owen Y on Oct 1, 2018 15:57:33 GMT 12
I have enjoyed the SEAS drivers for the past few years (rather than the OBD Field Coil drivers) because of their smooth, warm-toned response & fault-less overall tonal & dynamic performance. However, I realise now that their parameters do not seem to ideally suit basshorn loading & thus basshorn output with them is poor.
|
Post by jon on Oct 1, 2018 17:33:07 GMT 12
The Seas drivers are good in a sealed or vented application, which is why they have been good in the build you have.
However that dual ability limits them. Their EBP is around 70 indicating their nature.
How big is your room, it's difficult to judge? That may be a limiting factor too.
|
Post by jon on Oct 1, 2018 17:56:14 GMT 12
Efficiency Bandwidth Product...
EBP = Fs / QeS
<50 closed boxes >50 & < 100 closed and vented boxes >100 vented boxes
Owen's Seas drivers, at ~72, are a jack of all trades.
The Lowther is very firmly in the vented box only group.
|
Post by Owen Y on Oct 3, 2018 9:36:19 GMT 12
I'm not familiar with the concept of EBP, but jon is referring to the ratio of driver resonant freq (Fs) to electrical 'Q' of the driver (low Q is very damped, high Q less damped). Both the SEAS & Lowther have powerful magnets, but the Lowther has famously huge magnet flux + ultra narrow magnet gap. The OBD Field Coil drivers have similar physical specifs to the Lowther, but of course a powered electromagnet motor. Over the coming days (weeks?) I'll be comparing these drivers: - SEAS FA22RCZ (presently installed) - 'OBD' Field Coil drivers (in storage) - 'OBD' Field Coil drivers with new design diaphragms (cones) coming from Winding Man in Sth Africa. (SEAS FA22RCZ) (Existing 'OBD Field Coil drivers) (New 'OBD' diaphragm units, to be fitted)
|
Post by Owen Y on Oct 8, 2018 19:50:22 GMT 12
That brings us up to date: I will make better measurements now with a better measuring set-up - software & hardware bits now more inexpensively available. An essential bit has arrived from China for the new measuring set-up....
|
Post by Owen Y on Oct 10, 2018 16:38:16 GMT 12
Another piece of the puzzle arrives...
|
Post by colinf on Oct 10, 2018 19:14:05 GMT 12
|
Post by Owen Y on Oct 26, 2018 14:57:53 GMT 12
Looks like things will be on hold for a bit as I will be too busy with other things for the next fortnight. However, I downloaded a Pink Noise app offered on the Dynaudio webpage for a preliminary test of the test set-up. This free ios/android app appears to be a stripped-down version of the Studio Six RTA app ($7.49) that I was intending to use. The Studio Six RTA has mic compensation for ios device in-built mics & I'm not sure whether the free Dynaudio version does, however I ran some quick & dirty freq plots with the free app anyway on my lspkrs; An iPad hooked up playing Pink Noise into my line preamp, with long cable & iPad mic placed at 1m on tweeter axis: Pretty lumpy, roll-off at both ends, big bumps & troughs & a 100Hz peak (hum coming through my horns).
|
Post by Owen Y on Oct 26, 2018 15:11:59 GMT 12
Moving the ipad mic out to 2.6m on tweeter axis (as far as my TRS-to-RCA cable would allow), things start looking smoother in the main midband. Still rolled off, a 100Hz hum tone & a dip at ~160Hz. But overall it begins to indicate how horns don't start sounding OK until you get some distance away - then the tweeter / midhorn / basshorn start to integrate.
|
Post by Owen Y on Oct 26, 2018 15:15:58 GMT 12
Just for reference, I took a freq plot in an adjacent 'quiet' bedroom (9pm at night), just to see what the residual background noise looks like: You can see that there is 20-30db of low level, mainly LF 'noise' at our place
|
Post by Owen Y on Oct 26, 2018 15:28:00 GMT 12
This test set-up was dead-easy & user-friendly, so next I'll be doing the following: - D/loading the Studio Six RTA app (or Onyx RTA) onto the iPhone. - Running comparison tests between the in-built iph mic & the Dayton mic (with mic calibration files uploaded to the app I hope). Then optimising the horns by: - Varying the back-chamber volume of the midhorn SEAS driver to optimise bass-horn output (a la Bruce Edgar 'throat annulling'). - Installing the OBD Field Coil drivers horns again - with new cone units which have arrived from S Africa.
|
Post by colinf on Oct 28, 2018 3:18:31 GMT 12
“But overall it begins to indicate how horns don't start sounding OK until you get some distance away - then the tweeter / midhorn / basshorn start to integrate.” Sounds like the impulse response could be checked. I wonder how much the room noise affects the measurements. Years ago I used ETF software that took long averages of the pink noise response to better get the result from the speaker while reducing room noise. The pink noise was generated by the software itself at the same time as analysing the microphone output so it could compare what was being sent to the speaker and what was being measured quite accurately.
AMR-iFi R&D
|
Post by Owen Y on Feb 1, 2019 10:01:07 GMT 12
Time to progress this one this year.... I used to think that I'd never get involved in loudspeaker building, because it looked to involve lots of complex measuring & costly equipment. Then around 15 years ago when I started messing with DIY loudspeakers, measuring methodology involved a costly calibrated microphone or building one of those Panasonic WM-61A electret capsule mics that DIY people were building (which I did) & a mic preamp (which I also built), as below. However, technology moves fast. Almost before I could get around to using such DIY rigmarole, USB 'phantom-powered' mics became available - but good calibrated mics were still fairly costly. But, over the recent few years, you can now use your smartphone to measure loudspeakers & do room-tuning, using downloadable apps. (Pros even use these for venue sound-checking, etc.) Studio Six Digital, for example offers ios apps that can be used on all ios devices up to ios 12. They pre-calibrate their software & compensate for Apple's high-pass filer by turning off the LF roll-off built into iphones & ipads. (They say that the in-built Apple mics are very consistent.)
|
Post by Owen Y on Feb 1, 2019 11:25:54 GMT 12
As mentioned above, the Studio Six RTA app is downloadable for NZ$8.99. Octave RTA is another, with nice display options (but does not have ability to apply mic calibration files in their standalone app, NZ$8.99): There are also a few Free apps, but with limitations of course.
|
Post by Owen Y on Apr 5, 2020 14:55:26 GMT 12
Recent improvements (TT, phono stage PSU, etc) have however, highlighted an existing shortcoming in these loudspeakers. Something I've been aware of & well-known & common for single-drivers with a whizzer - some peakiness, break-up, intermodulation maybe, at the upper-mid end of the tonal range. Now more obvious to me. (At some musical peaks, woodwinds, brass, upper voice. etc. SEAS recommend a small R paralleled with the lo-pass coil network, to flatten the upper response - I will try implementing that first-up.
|
Post by Owen Y on May 7, 2020 21:40:32 GMT 12
This lockdown period has certainly been quieter around here (central Akld) & possibly where you live too. As Citroen mentioned, music listening environment may have improved for you. Except, of course, those of you who can enjoy living away from background (urban) residual noise - aside from the odd cicada horde or Tui birds This below is a RTA reading (on an iphone) of residual noise in a quiet bedroom here at 9:30pm, Thurs night : Compare with the above plot taken last October at 9:00pm (on an ipad mini) - 5-10dB more residual background noise at 100Hz & below. (Studio Six say that they compensate for iphone inherent LF rolloff in the various Apple devices.)
|
Post by Owen Y on May 8, 2020 15:30:52 GMT 12
For comparison, another residual noise plot taken with same iphone, but with the Dayton IMM-6 mic: Slightly different (smoother) overall response and/or LF sensitivity.
|
Post by Citroen on May 8, 2020 15:34:02 GMT 12
I need to get you round to my place for a measuring session!
|