advertisement


Troubleshooting level-dependant frequency response in loudspeakers

ToTo Man

ToTo Man

the band not the dog
I've just brought a pair of speakers out from storage, they haven't been played for around nine or ten years. They sounded great when last played but now they don't. The upper third of the frequency response is lumpy and the sound is coloured. This is the how one channel measured after just a few minutes warm-up:

first-measurement-after-storage.jpg


Note the spikes at 7kHz, 9kHz and 13kHz, these begin to smooth out as turn the volume up and get spikier again as I turn the volume back down. Any ideas what would cause this level-dependent behaviour? Tired crossover components? Misbehaving MF and HF level adjustment switches? Driver voice coils sticking in the gap? (None of the drivers use ferrofluid so at least I can rule that out!).

N.B. this is the very first time I've measured these speakers (I didn't have the gear 10 years ago!) so I cannot completely rule out that they exhibited this behaviour before going into storage but I'd like to think my less-well-trained ears would've noticed it...)

response-at-different-spls-3d-B-increments.jpg
 
Should be OK in a room with CH and decent ventilation. Surrounds do stiffen with age, depending on type, not usually a tweeter issue though. What model are they?
 
Speakers take time to warm up after storage IME. I suspect it is for a whole host of reasons, e.g. acclimatisation if they’ve gone from colder storage to another warmer room, capacitors re-forming, driver surrounds and spiders loosening-up, re-centering etc. Very much like guitars, pianos etc, which also like to be kept in use. Give them a couple of weeks use before doing anything drastic, there’s a very good chance they’ll just wake up IME. Especially after such a long time dormant. I’d actually be amazed if they didn’t.
 
OK I'll give them another few days to see if things improve.

On a more positive note, these speakers are the only ones I've measured thus far that have been truly "full-range" in this room, extending all the way down to 17Hz before rolling off which is, by coincidence, exactly how low my XXLS400 subs go before rolling off so I suspect 17Hz is the lowest frequency this room will support.

48108539156_f5b241e1f3_o.jpg
 
Face them together and loop a good full-range mono album through them out of phase for a couple of days (a lot of the sound output will cancel). I’d use dynamic music rather than test-tones as you really don’t want to overheat the voice-coils, but a good 48 hours or so of that at about 65-70db should help reform any tired old electrolytic caps and wake up the drivers.
 
I measured the other speaker (left) this morning and it doesn't exhibit the level-dependent FR as the right speaker, apart from some variation at 5kHz. The FR in the top end is different between the two speakers but I can't really conclude much from this as the left speaker was measured in a different position. and even if they were measured in identical spots, the mirror-image arrangement of the drive units makes obtaining perfectly comparable FR measurements impossible. Having said that, given my struggles to make two perfectly matched pairs of HF2000s for my Dittons from a pool of twenty units it wouldn't surprise me if the FR of the supertweeters in the IMFs weren't matched!

other-speakers-response-at-different-SPLs-3d-B-increments.jpg
 
You still have those small peaks about 5 and 7khz as before, but the 9 and 13kHz ones are a broader continuous peak of 9-12kHz. Still wonder how critical measuring position is as could be a bit of room interaction, or even comb filtering.
 
a.palfreyman said:
You still have those small peaks about 5 and 7khz as before, but the 9 and 13kHz ones are a broader continuous peak of 9-12kHz. Still wonder how critical measuring position is as could be a bit of room interaction, or even comb filtering.
Click to expand...
Comb-filtering and room reflections will certainly be contributing to the measured response but I've never before seen these vary with SPL, with the exception of one 'lazy' HF2000 I had that would suddenly 'wake-up' at a certain voltage level and get 5dB louder from a 1dB increase in input level. The HF2000 in the IMF crosses over at 13kHz though, it's an HF1300 that handles the 3kHz-13kHz range and IME these are robust and don't tend to develop problems. I suppose if the caps drift far enough off-spec the crossover to the HF2000 could be lower than 13kHz but how much lower?

EDIT - Measurements of said 'lazy' HF2000 supertweeter when I tested it in my Ditton 66 last year:

51319277693_5f81dca9e9_o.jpg


51319766424_7f262d9c23_o.jpg
 
OK that's most interesting. Who knows what effects a few lazy components in the crossover could do to the overall phase and how it sums... Any elcos in either of these? Suppose you could measure the drivers directly but that would be SOME undertaking...
 
I’d just let them do their thing for a week or two. I bet they’ll wake up. Which IMFs are they? I’ve just googled TLS80s and they have about 5 electrolytic caps a side, plus a load of film caps. It is a pretty large and visually complex crossover, inevitable given the number of drivers I guess. It will take time for an ancient cap not used for a decade or so to come back, especially with the typically low AC signal of a loudspeaker under normal use. That’s assuming it ever does come back. There is obviously an argument for re-capping, but I’d still just use them for a couple of weeks as every vintage speaker I have ever used, and that is a lot, never sounds at all good straight from a rest period. They really do wake up! It may not just be the caps either.

What drivers do your IMFs have? I’ve certainly seen some with things that look kind of like B110s, but aren’t as they have surrounds that perish and crack. Obviously that will have a huge impact on the mid and crossover region.

PS You can obviously measure caps for capacitance, leakage, ESR etc if you have the kit, but as I say I’d give them a chance to reform in use first as they likely will. My view on old capacitors has softened somewhat since getting the huge vintage valve Tektronix oscilloscope and finding after some time sitting on a dim-bulb tester its capacitors were mainly fine with very low ESR.
 
Tony L said:
I’d just let them do their thing for a week or two. I bet they’ll wake up. Which IMFs are they? I’ve just googled TLS80s and they have about 5 electrolytic caps a side, plus a load of film caps. It is a pretty large and visually complex crossover, inevitable given the number of drivers I guess. It will take time for an ancient cap not used for a decade or so to come back, especially with the typically low AC signal of a loudspeaker under normal use. That’s assuming it ever does come back. There is obviously an argument for re-capping, but I’d still just use them for a couple of weeks as every vintage speaker I have ever used, and that is a lot, never sounds at all good straight from a rest period. They really do wake up! It may not just be the caps either.

What drivers do your IMFs have? I’ve certainly seen some with things that look kind of like B110s, but aren’t as they have surrounds that perish and crack. Obviously that will have a huge impact on the mid and crossover region.

PS You can obviously measure caps for capacitance, leakage, ESR etc if you have the kit, but as I say I’d give them a chance to reform in use first as they likely will. My view on old capacitors has softened somewhat since getting the huge vintage valve Tektronix oscilloscope and finding after some time sitting on a dim-bulb tester its capacitors were mainly fine with very low ESR.
Click to expand...

The model is Professional Monitor mkIII, the big formica-covered one that uses the HF2000, HF1300, B110 and B139 6171 complement.

Facing the speakers into each other and wiring one out of phase does a good job cancelling overall output but causes an inexplicable resonance. I noticed it straight away when I played a track I know inside out and thought, "what on earth is that annoying, ringing whistle?!!". A quick REW sweep sweep confirmed it, see the huge spike at 2.75kHz? If I remove the smoothing it peaks at almost 90dB! :eek: I have no idea what's causing it but I notice the frequency shifts if I pull the speakers apart slightly.:confused:

speakers-facing-each-other-one-wired-out-of-phase.jpg
 
That's in the xover from the b110 to the hf1300 then. How well aligned are the cabs and therefore the drivers? (mirror pairs I assume) I reckon if they are not aligned, you'll get a lateral (sideways) transmission peak, a bit like the 15degree downward angle you get with 90 degree phase on 6dB covers, but that's guesswork TBH. Sha99ed cap on one of these drivers?
 
a.palfreyman said:
That's in the xover from the b110 to the hf1300 then. How well aligned are the cabs and therefore the drivers? (mirror pairs I assume) I reckon if they are not aligned, you'll get a lateral (sideways) transmission peak, a bit like the 15degree downward angle you get with 90 degree phase on 6dB covers, but that's guesswork TBH. Sha99ed cap on one of these drivers?
Click to expand...
Yes, mirror-imaged pairs. I've aligned the edges of the cabinets as precisely as I can but the baffles are pre-CNC so perhaps the drivers aren't in exactly the same position on each one. Maybe I need to deliberately offset one of the cabs?

EDIT - How close should the cabs be, literally nose to nose? I've currently got a small (probably 8mm) gap between the two.
 
Never done this (except A level physics years ago to show peaks and trough along a standing wave) but close should be good enough so that you get a small volume of air that literally moves back and forth between the out of phase drivers. The imbalance resulting in output has to be due to either phase or dB differences, or both. Chuck a blanket over the gap ;)
 
I’d certainly not allow any driver to run the risk of touching the other. I’ve never actually done this, I’m fairly patient so just let stuff warm up in use, maybe run a new component in the TV rig for a week or two or whatever. I always assumed a cm or two was fine and then just cover with a blanket to block any treble spill.
 
No risk of the drivers touching as the captive metal frames that hold the grilles in place each add an extra cm distance to baffles, I was just concerned whether not having enough of a gap would restricting the airflow too much and potentially damage the drivers? I'm no doubt over-thinking this!

FWIW - Pulling the speakers a few cm's further apart lowers the frequency of the resonance to 1.5kHz.

EDIT - I've just discovered that the resonance is still there (but at a lower volume) when only one speaker is playing. This would suggest that it's being caused by the output from one speaker reflecting/diffracting off the enclosure that's facing it?
 
You must log in or register to reply here.


advertisement


Back
Top