Yes, wasted electrical power is usually 99% wasted as heat... In the cold months, that excess heat is not really a waste, it reduces the house furnace load on a joule for joule basis. An electric joule may cost a bit more than a natural gas joule, but in my area, the difference is not that great. If your heating system is electric, there is no cost or efficiency difference at all.
Now in the summer, wow, wasted electrical power is a double whammy. First we pay for the wasted electricity, then we pay again for the air conditioning to remove the excess heat from the house.
But this morning the outside temps are 14 F / -10 C, so not much worry about excess waste heat in the house !
Sorry to geek out that subject, but some of my grad school work was in this area...
Now in the summer, wow, wasted electrical power is a double whammy. First we pay for the wasted electricity, then we pay again for the air conditioning to remove the excess heat from the house.
But this morning the outside temps are 14 F / -10 C, so not much worry about excess waste heat in the house !
Sorry to geek out that subject, but some of my grad school work was in this area...
A joule for joule exchange is not always equal. If using a heat pump system, you can transfer 3-4 joules of heat inside for every joule of electricity compared to resistive waste heat. If everything else is equal, it is superior to use less energy in my opinion. That is why I would take Class D over valve tube equipment in almost any situation.
I just heard a set of speakers yesterday with a small soft dome tweeter - there was so much "sizzle", it was a pain. Different pain to metal dome "sizzle" but still far from natural.
There are some better soft domes, there are some more neutral hard domes. In the middle there are Beryllium domes for me.
(Seems like I'm on repeat in that case ... ;-))
@hifijim.
Thank you for the measurement data. I have fallen hard for the SB26STAC tweeter. I use it in my 4-way active build and subjectively I cannot fault it. My baffle shape has 2" roundovers instead of 45 degree chamfers, but the measurement data should be closely comparable.
https://www.diyaudio.com/community/threads/active-4-way-with-reaction-cancelling-woofers.402645/
I may simply have not found its limits since there is no perfect transducer. I have conducted A/B comparisons with most of the (non-Satori) SB soft dome and aluminium dome tweeters and it is the one I most preferred for doing what I expect of a tweeter for my particular application. It is not shamed in subjective comparison with my Scan-Speak Revelator 2905/970000 at nearly 7x the cost. I have a stash of the 26STAC lined up for future projects.
Attachments
Yes, that was one of the options I looked at.
I used the miniDSP to listen to the SB15NBAC+SB26STAC with a variety of crossover options before I committed to a specific analog passive crossover.
I used Vcad sim to come up with each DSP crossover, and then implemented it in the MiniDSP. With a 2nd order filter LR, 2.6k was just about optimum. It sounded very good. With LR4, the optimum was at 2.1k. I found a slight difference between the LR2 and LR4, even after careful EQ and level matching, but I did not have a preference.
From a simulation standpoint, the directivity performance of the LR2 and LR4 were similar, and probably within the accuracy capability of the simulation. I would have been happy with either.
When I designed a passive crossover for both the LR4 and LR2, I found it difficult to get a 2nd order slope on the drivers. The mid driver would need an electrical 1st order filter (i.e. a single inductor) combined with fairly sophisticated notch filtering above 4k in order to achieve an (almost) LR2 acoustical response. The tweeter was even more difficult. It has a shelved up +3 dB response from 800-2.5k, and a natural 3rd order rolloff below 800 Hz. I struggled to get something approaching a 2nd order acoustical slope at 2.6k using passive components. Only with a lot of complexity could I get something close.
It seems like these drivers just naturally "want" to have 4th order acoustical slopes. With a 2nd order electrical filter on the mid and a 3rd order electrical filter on the tweeter, things just fell into place. I am not surprised. This is a very common filter topology for 2-way passive speakers using small woofers and dome tweeters.
Of course with a fully active design, it is possible to implement any sort of crossover slope, as you well know. With my other designs (all active), I have found that I tend to prefer a 2nd order filter between woofer and mid, which is usually in the 160 - 400 Hz range. I tend to prefer a 4th order filter between the mid and tweeter, which is usually in the 1.6k - 3k range.
In my experience, too much sizzle, too much sparkle... this is usually caused by the frequency response of the speaker, more so than the particular driver technology. The painful speakers you heard, it is possible they could be tamed quite a bit with some careful EQ. Thoughts?
There are some tweeters that have a frequency response that is difficult to work with using a passive crossover, no doubt about that.
I agree with Jim regarding the sizzle. Sometimes it can sound a little bit confused.
Sometimes changing, EQ is not quite enough to rectify the situation. Sometimes you have to change the directivity. This can be done by you change how the baffle near the HF device loads or guides the HF element, or changing the HF device.
Sometimes changing, EQ is not quite enough to rectify the situation. Sometimes you have to change the directivity. This can be done by you change how the baffle near the HF device loads or guides the HF element, or changing the HF device.
I'm sure it was intended to have that amount of "sparcle". Sounds catchy in the shop and the owner loves it. You can also put a protection grill on which dampened it a lot (maybe even to much, feels like double thick fabric).
But my problem was that it also didn't sound "real". I know how loud hf sounds but it was "more" as that. And it was a pretty small dome - I was surprised about that to be honest.
Sits in the living room of a colleague, he really likes it that way ... we will never know
It was just one more confirmation of my bias against soft domes ... looking forward to find some to break that
Here is an issue I have been pondering for the past couple of weeks.
When I made frequency scans of the completed system, I noticed a dropping off in response above 7k compared to the simulation. By 16k, the loss was about 3 dB. I repeated the measurements several times and confirmed the effect is real.
Since this is an active speaker that uses DSP EQ, correcting this is no problem at all, I just add a bit of shelving to bring the high frequency response back to what I want it to be. But the underlying cause has been a mystery to me.
My simulation is based on driver measurements made in this cabinet, under the same conditions as the final system measurements. There is no reason the sim should be different than the measured reality, and in the past, I have gotten very good agreement between sim and system measurements.
If I change the simulation by adding 39 uH (0.039 mH) of inductance between the amp and the tweeter crossover, I can simulate the effect almost exactly. This does not necessarily mean that 39 uH of parasitic inductance is the cause of the difference, but it is a plausible explanation.
At this point, I would like to ask for some opinions.
First, is it reasonably possible that the wiring and the crossover could introduce 39 uH of inductance between the amp and the tweeter?
Second, has anyone else ever experienced a loss of high frequency response like this?
When I made frequency scans of the completed system, I noticed a dropping off in response above 7k compared to the simulation. By 16k, the loss was about 3 dB. I repeated the measurements several times and confirmed the effect is real.
Since this is an active speaker that uses DSP EQ, correcting this is no problem at all, I just add a bit of shelving to bring the high frequency response back to what I want it to be. But the underlying cause has been a mystery to me.
My simulation is based on driver measurements made in this cabinet, under the same conditions as the final system measurements. There is no reason the sim should be different than the measured reality, and in the past, I have gotten very good agreement between sim and system measurements.
If I change the simulation by adding 39 uH (0.039 mH) of inductance between the amp and the tweeter crossover, I can simulate the effect almost exactly. This does not necessarily mean that 39 uH of parasitic inductance is the cause of the difference, but it is a plausible explanation.
At this point, I would like to ask for some opinions.
First, is it reasonably possible that the wiring and the crossover could introduce 39 uH of inductance between the amp and the tweeter?
Second, has anyone else ever experienced a loss of high frequency response like this?
Is that notch for the woofer breakup? I've found those to be extremely sensitive to exact value. Maybe the breakup is higher than you expected and interfering with the tweeter response? I would expect more of a localized effect, not broadband like you have, but just a thought.
Could it be something in your measurement setup that changed? Like cable length or something?
Could it be something in your measurement setup that changed? Like cable length or something?
Is the difference, the simulation of the passive components combined with measurements of the raw drivers in the cabinet, to a measurement of the actual passive filters with the drivers in the cabinet?
If that is the case there could well be a discrepancy between the values of the parts you have and the values that were simulated.
One way to check is to compare the simulated frequency response of the filters parts on their own, to an electrical frequency measurement of the actual filter parts, to see if that is where the difference is happening.
Notch filters are also hyper sensitive to inductor location and orientation, so much that the wrong location can completely negate its use.
I would look at the change in the frequency response of the amplifier when the amplifier is operating on the passive filter you calculated+twitter. If the amplifier has feedback before the output LC filter, then the output frequency characteristic is amplified at high frequencies, depending on the load impedance. There is also a decrease in the frequency response in the signal processor if upsampling is used there with a subsequent filter, that is, the signal processor itself can distort the frequency response at high frequencies.
Last edited:
Yes, it controls the 5" midwoofer breakup. In this particular case, I can change the notch filter values by +/- 10% without affecting the response too much.
I measured the woofer response with the crossover, and it agrees closely with the sim. When I measured the tweeter response with the crossover, I see the difference... a rolloff starting at about 7k and growing to be about 3 dB at 16k.
Good thought...The wiring length definitely changed. For the testing of the drivers, I was using short leads from the amp to the drivers. The actual finished system uses longer wiring. As you can see, I used twisted pair wiring for a neater installation. There is probably about 30" of twisted pair wire between the amp and the tweeter... Surely twisted pair wire would not add this much inductance? or maybe yes?
yes that is correct.
I agree that is a concern in any analog circuit. In this particular case, I can not simulate this response with any reasonable variation in the tweeter crossover values.
Interesting... I need to think on that. I need to think about how the inductors were arranged in the cabinet.
Thank you for all the comments...... j.