So I built this boombox a little while ago and I've been trying to implement a minidsp to get it to sound better. The thing is, whenever I try to eq it I make it sound like crap. For whatever reason, I haven't had this problem with the speakers inside my house, but for this one, since it is portable, I've been taking it outside to measure and eq it. I sit it on a desk and push it to the edge so the desk doesn't reflect, and make sure it's far away from anything outside that can interfere with my measurements. I use a umik 1 mic with a proper stand now and set it up four feet or so behind the right channel of the speaker. I have the mic aimed right at the tweeter where my ear would be while I'm listening. (This is a typical woofer + tweeter arrangement so the tweeter is on top just like a bookshelf speaker) I have the left channel disconnected so it doesn't interfere with the measurements. I can eq it so that it reads within +- 2 db on a graph but it sounds awful, it doesn't sound faithful at all. When I do a frequency sweep manually and listen for how flat the speaker is it's very obvious to me that it's far from flat even though that's what the graph shows. My handheld spl meter confirms this. I'm not sure what exactly I'm doing wrong, I've done this several times and I'm on the verge of giving up and not using any eq at all but I hate leaving performance on the table, especially when I already have the minidsp. I will add that it might be a factor that my enclosures for the drivers are a little strange, the enclosure for the tweeter is separated from the woofer. I'm not sure how much of a difference it makes. So what am I doing wrong? Is there a book somewhere telling me how to properly do something like this?
Are you doing RTA or impulse measurements? Being the application for spectral balance, I think the RTA might be more useful.
I recently built a bluetooth box, and I know it has a slight bump in the bass before it rolls off a bit earlier than most of my designs do. This adds the heft to the bass and makes it sound substantially larger. The rolloff is amplifier FR induced with an integrated highpass to both keep the drivers happy and not tax power requirements.
If we could see your measurements via RTA (SynRTA is free), then we might be able to help your progress.
Later,
Wolf
I recently built a bluetooth box, and I know it has a slight bump in the bass before it rolls off a bit earlier than most of my designs do. This adds the heft to the bass and makes it sound substantially larger. The rolloff is amplifier FR induced with an integrated highpass to both keep the drivers happy and not tax power requirements.
If we could see your measurements via RTA (SynRTA is free), then we might be able to help your progress.
Later,
Wolf
Hi! I'm was using rew, I'm not entirely sure how to get measurements off of synrta, here's my rta graph though. This is without any DSP.
If you are xovering at 3kHz, you need to revise your xover. There is a huge null there that could be misaligned or out of phase. You also should tilt your woofer down a bit with a larger lowpass coil value and increase the tweeter padding resistance.
You also need to tune higher your bass alignment to fill that in which might require less tilt on the woofer xover.
Once you try these things, you should be closer to where you might not need the EQ as badly.
Later,
Wolf
You also need to tune higher your bass alignment to fill that in which might require less tilt on the woofer xover.
Once you try these things, you should be closer to where you might not need the EQ as badly.
Later,
Wolf
I'm crossing at around 1600 hz actually. What do you mean by tilting the woofer down? I'm actually running it bi amped so turning down the gain on the tweeter's amp should have the same effect as padding right? Also I'm not certain as to how accurate the graph is. That's my main issue, and I don't know why it's not accurate. Even when I eq it to +- 2 db, it sounds completely wrong, I have no idea why. I'm pretty sure I have the umik 1 and the speaker set up right, but a nearly flat measurement sounds completely wrong. Is there anything that can be causing this?
Actually when I eq it, it looks closer to that curve than truly flat. I really feel like the problem is the accuracy of my measurements. For example, I can boost the 40-60 hertz range by 12 db and it would show up as only a 3 db boost on the graph. Obviously this is inaccurate because I can hear the 12 db boost with my ears, far from 3 db. My spl meter also confirms the boost. I've had the same issue with boosting for example 3.5 khz. If I boost it by 10 db it only shows a 2 db gain, when in reality it sounds definitely much more like 10 because I can hear the peak. So I have no idea why my measurements are so far off.
If you are biamped- then the gain on the tweeter can be lowered. Same difference, just different application.
I'm seeing a HUGE SUCKOUT at 3kHz. That's why I thought maybe 3k was the xover, and maybe the tweeter polarity needed reversed. Have you tried to reverse the tweeter polarity yet and remeasured?
Another possibility is that the woofer breakup is at 3kHz, and either not suppressed well, or creating a suckout in the tweeter response due to it's higher magnitude. This type of result can be quite common if not compensated.
Tilt is referred to as the flatness of your curve. To me, the woofer looks tilted up in the RTA you have shown, or maybe that the tuning of the woofer is too low causing the low peak and midbass suckout. This could also be floor-bounce, without knowing your measurement methods.
Have you measured any ground-plane measurements? This might work to your advantage as it gives you a gain in the bass, and you have to tune the low-end down to flatten it out.
Just throwing some thoughts out there,
Wolf
I'm seeing a HUGE SUCKOUT at 3kHz. That's why I thought maybe 3k was the xover, and maybe the tweeter polarity needed reversed. Have you tried to reverse the tweeter polarity yet and remeasured?
Another possibility is that the woofer breakup is at 3kHz, and either not suppressed well, or creating a suckout in the tweeter response due to it's higher magnitude. This type of result can be quite common if not compensated.
Tilt is referred to as the flatness of your curve. To me, the woofer looks tilted up in the RTA you have shown, or maybe that the tuning of the woofer is too low causing the low peak and midbass suckout. This could also be floor-bounce, without knowing your measurement methods.
Have you measured any ground-plane measurements? This might work to your advantage as it gives you a gain in the bass, and you have to tune the low-end down to flatten it out.
Just throwing some thoughts out there,
Wolf
Thank you, I explained in more detail how I'm measuring in the op, but basically I'm putting it on a desk, taking it outside, setting the mic up to ear level a few feet away and measuring. I haven't tried reversing the polarity, so maybe I'll give that a shot. I'm rolling the woofer off at 1600 hz at 48 db/octave, so I think break up shouldn't be audible? Regardless I'm not sure why it looks like that at 3 khz. It doesn't sound like it really drops off that much... I haven't tried ground plane measurements actually, but for a portable player won't that be a worse idea because of the changing environment? The woofer is also not tilted up, however, I'm measuring the speakers at tweeter level, so the microphone is placed above the woofer. Is this counterproductive?
The measurements are outside actually but my enclosure is strange, so I don't know if that would have a similar effect has room modes. The tweeter enclosure is "levitated" one inch above the woofer enclosure, so there is a 1 inch gap. I'm not sure if this would have an effect though.
I do not mean the speaker or woofer is tilted up, I mean the 'frequency response' of the woofer in your measurement is tilted up spectrally when compared to flat average response. It does not look like a level ruled line as an average, but one with a rising line from left to right.
I would not measure in any table if outside, but either on the ground or free-space as full reinforcement or no reinforcement.
Later,
Wolf
I would not measure in any table if outside, but either on the ground or free-space as full reinforcement or no reinforcement.
Later,
Wolf
I've used equalization for some time and recently did some head phones with good results.
Equalizer settings for Sennheiser hd599 headphones.
Equalization can be very tricky. Not everything can be equalized either. Neither measurements or hearing are completely reliable.
My advice is to find the best measurements you can of the drivers. Then start by flattening any obvious peaks, and listen. Then adjust the the overall contour and balance of treble and bass, essentially baffle step.
Equalizer settings for Sennheiser hd599 headphones.
Equalization can be very tricky. Not everything can be equalized either. Neither measurements or hearing are completely reliable.
My advice is to find the best measurements you can of the drivers. Then start by flattening any obvious peaks, and listen. Then adjust the the overall contour and balance of treble and bass, essentially baffle step.
Yes they do have enough headroom, I can definitely hear the 12 db boost even though it doesn't show it on the graph unless I aim the umik directly at the woofer. Then it does show the peak.
Thanks, I'm thinking of just doing this instead. Besides, I've heard that as little eq as possible is better anyways.
Its easy to make a mess unless you know where the peaks and dips are. You want to try to find off axis information to. If the drivers are decently flat then you can adjust for speaker or room, but a lot of things like early reflections can't be fixed with eq.
Being outside, on a table, gets rid of room modes but does not mean anechoic conditions by any means.
Old style measurements often relied on 1 of 2 tricks.
1) mounted speaker on a pole, at least 30 feet high, so floor reflections went down in frequency and lower in intensity so as not to bother much.
2) dug a hole in the middle of a large area, say a parking lot, away from buildings, pointed speaker upwards and mounted microphone on a boom above it.
This effectively turns outside space into an anechoic chamberbut results are really valid only for halfspace propagation, by definition.
Modern home measurement gets rid of anechoic chamber by splitting measurements:
* gated preamp for mids/highs where direct signal reaches microphone but reflected one not (yet)
* real close mesurement for low frequencies so unavoidable floor bounce is way lower than direct pickup, where microphone can be just a few inches from cone.
Software combines both.
Not very Kosher but good enough.
Old style measurements often relied on 1 of 2 tricks.
1) mounted speaker on a pole, at least 30 feet high, so floor reflections went down in frequency and lower in intensity so as not to bother much.
2) dug a hole in the middle of a large area, say a parking lot, away from buildings, pointed speaker upwards and mounted microphone on a boom above it.
This effectively turns outside space into an anechoic chamberbut results are really valid only for halfspace propagation, by definition.
Modern home measurement gets rid of anechoic chamber by splitting measurements:
* gated preamp for mids/highs where direct signal reaches microphone but reflected one not (yet)
* real close mesurement for low frequencies so unavoidable floor bounce is way lower than direct pickup, where microphone can be just a few inches from cone.
Software combines both.
Not very Kosher but good enough.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.