More Metronome Musings

Dedicated to those large boxes at one end of the room
Post Reply
Cressy Snr
Amstrad Tower of Power
Posts: 10552
Joined: Wed May 30, 2007 12:25 am
Location: South Yorks.

#1 More Metronome Musings

Post by Cressy Snr »

I thought I would start this topic to gather together all the stuff scattered all over the place and try to bring some order to the chaos.
I suppose, if I had kept it all together as Mark has done with his epic "Something Stirs in the Undergrowth" thread, then there would be no need for this. However these speakers are entering a completely different phase of development, so here we go.

So the situation now is that our Ant came round this morning and produced a spectrum analyser app on his Android mobile phone. OK I know these apps are alright as far as they go but, a mobile phone mic is not exactly Bruel and Kjaer stuff and this should be borne in mind when interpreting results from the app. That caveat out of the way, we set about measuring the speakers at 1M away on axis.

I downloaded a super little signal generator app for the Mac Mini and gave the analyser a 10 second frequency sweep from 100Hz to 20KHz. This showed up a 6dB peak at 3.4kHz and a 10dB narrowband peak at 7.7kHz.....Jeez :shock: OK we were measuring in room, but I was not impressed to put it mildly.We did a bit of tweaking with the parametric eq before he left and took out most of the problem but I needed a second checkup myself, just to make sure. So I downloaded my own iOS "pocket RTA" jobbie and had a go myself.

Yep...same shocking results from this one too on frequency swept sinewaves from the Mac Mini app.
White noise from the Mac signal generator corroborated the sinewave analysis evidence showing clear and nasty peaking, which on the white noise I could also hear as two standout effects above the nominal level. Despite this, the encouraging thing was that everywhere else on the spectrograph looked quite decent.

These (admittedly crude) measured results seem to correlate with my subjective experience of things being mostly OK except on records that set off these peaks sending me into wince mode.

So I dealt with the upper peak, which is very near to the 8KHz first order crossover point, by reversing the phase of the ribbon tweeters so that I was able to get a partial cancellation. This was then supplemented by a 2dB deep low Q notch filter using the JRiver parametric eq. Another sweep and burst of white noise confirmed I had gotten rid of the HF peak and had a nice smooth, flat response out to 20KHz.

Reinstating the 3.65KHz notch filter dealt with the upper mid nasty, giving a nice flat in-room response from 100Hz out to 20K. There were a few wiggles, but nothing on anywhere near the scale of those peaks.

Measuring out in the far field, at the listening position gave excellent results, nice and flat; within the limits of these crude measurements of course.

Auditioning the system after these adjustments was a pleasant experience. the difference with and without the eq was not subtle on recordings picked deliberately to set off the nasties. On recordings that did not set them off, the difference was more subtle. This explains why I was perfectly happy some days and totally pissed off on others.

The major fly in the ointment however is that replicating these filters in the analogue domain without introducing more problems is going to be nigh on impossibe, as the 3.5KHz notch is 7dB deep, with a Q of 6.

So it seems that Scott's suggestion of rolling off the FF225WK and then going in with a conventional tweeter is the one that has the best chance of long term success.
Last edited by Cressy Snr on Sun Mar 08, 2015 7:24 pm, edited 1 time in total.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
simon
Thermionic Monk Status
Posts: 5600
Joined: Thu May 24, 2007 11:22 am
Location: People's Republic of South Yorkshire

#2

Post by simon »

Interesting stuff Steve. What was the Android app, might have to try it.

Smart phone technology is scary when you're of a generation that thinks digital watches are a pretty neat idea (been a while since we've had a Hitchhikers reference!). I saw a documentary some months ago which included an American research doctor who detected he had a rare (cancer?) illness because he was hooked up to his iPhone which was running numerous blood analyses. He was using himself as a research guinea pig and ordinarily the illness wouldn't have been detected. Scary.
Cressy Snr
Amstrad Tower of Power
Posts: 10552
Joined: Wed May 30, 2007 12:25 am
Location: South Yorks.

#3

Post by Cressy Snr »

Hi Simon,
The app is called "Spectrum Analyze" He got it off the Google Play store.

Yes these apps are incredible.
20 years ago, a spectrum analyzer would probably have cost a small fortune.
Amazing what they can do these days.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#4 Re: More Metronome Musings

Post by IslandPink »

Very interesting, Steve .
SteveTheShadow wrote:I suppose, if I had kept it all together as Mark has done with his epic "Something Stirs in the Undergrowth" thread, then there would be no need for this.
Yes, it's really easy to find everything on my thread :roll:
SteveTheShadow wrote: .... and a 10dB narrowband peak at 7.7kHz.....Jeez :shock:
Oooh ! - bit like a Clearaudio cartridge !

Anyway - I feel your pain ... - but it's good to have an 'engineering' explanation for your listening experiences - you don't have to blame it on your hearing .
"Once you find out ... the Circumstances ; then you can go out"
chris661
Shed dweller
Posts: 2556
Joined: Mon Feb 22, 2010 7:29 am
Location: Sheffield

#5

Post by chris661 »

Hmmm...
Low-Q notches affect a very wide frequency band, Steve.

I'd keep the ribbons with their original polarity and try a high-Q notch. Higher the better.

If you could post up some before/after screenshots, I'd be interested to see.

Chris
Cressy Snr
Amstrad Tower of Power
Posts: 10552
Joined: Wed May 30, 2007 12:25 am
Location: South Yorks.

#6

Post by Cressy Snr »

Yep... low Q notches do affect a wide range, as you know. The reason for the lower Q notch was that reversing tweeter polarity had pulled the massive HF peak down from +10dB to +3dB, which put it within a broader hump of frequencies from 5 - 13kHz. The low Q notch flattened that hump nicely, whilst the higher Q notch at 3.65KHz removed the objectionable peak lower down as it was well below the crossover to the ribbon, so needed a bit more strongarm tactics to deal with it as there was no help from the tweeter inter-relationship.

The tweeter, as it turned out was actually the wrong way round anyway as I had failed to allow for the cap in series with it when I connected it up :roll:

Unfortunately Chris, I did not save the plots, and I don't want to be applying any more heat to the tweeter terminals to re-reverse it and do the sweeps and white noise again. I'm not much of a scientist.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
simon
Thermionic Monk Status
Posts: 5600
Joined: Thu May 24, 2007 11:22 am
Location: People's Republic of South Yorkshire

#7

Post by simon »

With apologies to Steve for thread jacking I couldn't resist having a play with the spectrum analyser app (there are loads available for free). Below are two frequency sweeps - neither are meant as scientific experiments and I suspect are entirely limited by the freebie frequency sweep generator (which increases frequency in small steps), the laptop sound card and the smartphone microphone. The sweeps were run from 100Hz to 20kHz.

The first plot is both Fostex 208e sigmas with T90A supertwatters speakers, from about 5m away, with GM70 amp

Image

The second is both Quasars with Alpair 12Ps, Supravox GM285s, and Monacor pseudo ribbons above 12k, with 2A3 amp, from about 3m away

Image

Not sure any of it really means anything, but I'm easily impressed by technology :-D. Interestingly when I tried it with my tablet the response fell off a cliff at 5k5Hz.
SimonC
Old Hand
Posts: 545
Joined: Mon Sep 24, 2007 11:34 am
Location: West Yorkshire

#8

Post by SimonC »

simon wrote:Not sure any of it really means anything, but I'm easily impressed by technology :-D. Interestingly when I tried it with my tablet the response fell off a cliff at 5k5Hz.
I found the opposite, my phone mike drops off a cliff at 5kHz, but the tablet goes out to around 10kHz...but I'm not convinced that either of them are any more than general indicators of reality. I don't rely on them for anything beyond a quick visual check, and drag my calibrated mic setup out for making measurements beyond that.

Simon C
chris661
Shed dweller
Posts: 2556
Joined: Mon Feb 22, 2010 7:29 am
Location: Sheffield

#9

Post by chris661 »

Any chance of setting a logarithmic scale there?

I use a similar app to find feedback frequencies in live sound applications - it usually happens at a few specific frequencies, so taking those out with a high-Q notch filter leaves the rest of the range intact.

Chris
simon
Thermionic Monk Status
Posts: 5600
Joined: Thu May 24, 2007 11:22 am
Location: People's Republic of South Yorkshire

#10

Post by simon »

Only if I pay my 99 cents :-). Looks like you have to pay for Pro versions if you want all the features.
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#11

Post by IslandPink »

I'm struggling to see anything in the first plot that I recognise from tests on the FE208Ez's using Nick's kit with Audiotester.de ; like for instance the peak at 3.5 to 4Khz that also appears in Fostex's datasheets. I would therefore not have much confidence in anything shown in either of those plots !
"Once you find out ... the Circumstances ; then you can go out"
simon
Thermionic Monk Status
Posts: 5600
Joined: Thu May 24, 2007 11:22 am
Location: People's Republic of South Yorkshire

#12

Post by simon »

And that's pretty much why I'm keeping my 66p in my pocket :-).

Still, looks cool dunnit!
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8863
Joined: Mon May 21, 2007 9:42 pm

#13

Post by Paul Barker »

It's a foreign language to me.
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
User avatar
cressy
Shed dweller
Posts: 2906
Joined: Fri Aug 03, 2007 7:07 pm
Location: the great white space
Contact:

#14

Post by cressy »

SimonC wrote:
simon wrote:. I don't rely on them for anything beyond a quick visual check, and drag my calibrated mic setup out for making measurements beyond that.

Simon C
yep that was the point of using it, so we could at least get an idea of what was going on and a starting point to work from rather than blindly adjusting the dsp hoping for the best

the large peaks were immediately obvious and the scale of the peaks was initially puzzling until it dawned that the ribbons being in the wrong phase was the probable culprit.

so as an accurate measurement tool the app is probably little better than a chocolate fireguard, but as an initial visual representation of what the response looked like it was rather useful and set us off on a path to be able to adjust the dsp in the right manner
simon
Thermionic Monk Status
Posts: 5600
Joined: Thu May 24, 2007 11:22 am
Location: People's Republic of South Yorkshire

#15

Post by simon »

And all for a Yorkshireman's favourite price :-)
Post Reply