3-way to active - Hypex FA253 - learning project

Vertical directivity and phase/GD will definitely be different. Most likely LR2 version has a bit higher distortion at some frequencies too. Optimal Fc is often higher for LR2

For a diyer subjectively best sound is more important than optimal measurements/settings! On-axis response don't have to be straight, PIR is closer to subjective "flat/smooth". I tend to prefer LR2 subjectively despite LR4 often measures a bit better, in my projects.
 
  • Like
Reactions: shadowplay62
Today I did another (4th) set of measurements aiming for final setup. Indoors, but using what I think is the best possible speaker position for reducing reflections. I am now more confident with the process and HFD & REW settings.

BTW, I now know why I messed up the midrange sweeps in the church - HFD unmutes everything when you upload a preset to the amp, and I simply missed it.

It worked really well until I reached the step of time aligning the tweeter and the mid.

As in the post 679 above (step one) I managed to to flatten the response of the tweeter and the mid nicely with only two notches per driver and a BFC shelf (the filter suggested by @hifijim worked perfectly).

I then tried to time align the mid and the tweeter and it puzzled me. I tried several delays on tweeter and mid (with one of the drivers inverted) aiming for the deepest null.

With LR4 the two drivers were summing up best (perfectly flat) without any delay. Any delay to any of the two drivers resulted in shallower null. I have tried three lowest possible in HFD delays (11 to 32 microsec) and each increase reduced the null. I did not try longer delays and reverted to XO without any delays on tw or mid.

Should I go through longer delays seeking deeper null?
 
Last edited:
Today I did another (4th) set of measurements aiming for final setup. Indoors, but using what I think is the best possible speaker position for reducing reflections. I am now more confident with the process and HFD & REW settings.

BTW, I now know why I messed up the midrange sweeps in the church - HFD unmutes everything when you upload a preset to the amp, and I simply missed it.

It worked really well until I reached the step of time aligning the tweeter and the mid.

As in the post 679 above (step one) I managed to to flatten the response of the tweeter and the mid nicely with only two notches per driver and a BFC shelf (the filter suggested by @hifijim worked perfectly).

I then tried to time align the mid and the tweeter and it puzzled me. I tried several delays on tweeter and mid (with one of the drivers inverted) aiming for the deepest null.

With LR4 the two drivers were summing up best (perfectly flat) without any delay. Any delay to any of the two drivers resulted in shallower null. I have tried three lowest possible in HFD delays (11 to 32 microsec) and each increase reduced the null. I did not try longer delays and reverted to XO without any delays on tw or mid.

Should I go through longer delays seeking deeper null?
Pictures would help a lot. On axis mid. On axis tweeter, Sum of the two. All on one graph with appropriate dB scales to where it's easy to read.

What is the tweeter x-over point? Are you measuring on the tweeter axis from 1m, or slightly more?
 
I just found my sim of your speaker, and played around with some things. I still would like to see graphs, but what I just tried was changing the delay one inch in each direction. In other words 25.4 mm. I loose track of the decimal changes for metric units. Changes in 6mm increments should put you close to what is needed. That would be roughly 1/4 inch. Try both positive, and negative. ie 6mm, 12mm, 18mm, then -6mm, -12mm, -18mm. The software used may interpret the change differently than you or I think it does.

What x-over frequency are you using?
 
Last edited:
the black trace is LR4, no delays on tw or mid, summing to flat
all other lines are LR8, different delays on tw/mid as noted. Notice how none sum up to flat.
Both LR4 and LR8 XO are at 2,500Hz

aligning tw mid.jpg
 

Attachments

What is that? As a rule of thumb, a null of 12dB is pretty good. Note that that will change depending on where you measure from.

I usually measure from roughly 1m,and on the intended listening axis. So if the tweeter is at ear level, use the tweeter axis.
 
I'm not quite following you. Pick one, and optimize it. There's no need to even look at 8th order in my opinion. Get a good LR4 for a starting point. Then if you want, you can try 8th order. Currently you have too much information on the graph, and it is only creating confusion.
 
My software tells me to use ms for delay. The passive side uses mm. So some confusion on my part.

I think a tweeter delay of .070 ms or less should be close. That would be 70 microseconds.

The .070ms is working well in the sim using your original data, and corresponds to almost one inch. Visually, I would think with your tweeter, it would need virtually no delay.

Make sure only one of the drivers is delayed.
 
Last edited:
  • Like
Reactions: AlmaAtaKZ