I'm having trouble with crossing this ribbon (GRS RT1.R) with this planar mid (GRS PT6825-8). I can cross them anywhere from 2k-8k and there ends up being a dip in response immediately once you go off axis even a few degrees. How can this be? I've modified my XO to be nearly all 4th order filters to try to keep them in phase. If I go any less, even a 3rd order, they will be in phase at the XO and then a few thousand hz later they will be out of phase and cancelling out. The same thing seems to happen to the mid and woofer crossover. The woofer is just a cone woofer. It is far less pronounced with the mid to woofer than with the mid to ribbon XO. I am not understanding how this can be.
Is this just the nature of ribbons and planar?
I managed to get the woofer and planar XO halfway decent.
This is designed to be listened to at a desk so it isn't that horrible of an issue but I would like to understand it.
The drivers are not vertically aligned. Here is what the unit looks like. I have entered the offset data into the XO. Even if I removed the X axis offset from the data. Essentially making them vertically aligned, it makes little difference.
Here is the current XO. You can see how big of dip there is at even just the 15 degree mark
Then here are the responses of the tweeter, mid, and woofer respectively. Notice how crazy these responses on. I had to hammer these things into shape to get anything like an accurate response.
Woofer
Mid
Tweeter
I took these measurements with a jig so that the microphone was positioned exactly at the front of each driver when taking the measurement. It was aligned with the face of the unit so that the Z offset would be taken with the data.
I'm sure there is just something I do not understand here. If anyone knows, please let me know. Possibly its just hard using these types of drivers. That I can accept as I didn't think this would be easy. Odd parts cause odd issues.
Lastly, someone mentioned to me to hit ctrl+alt+E to fix phase issues in REW. Does anyone know exactly what this does? It seems to affect the delay. Wouldn't this throw off my Z offset data that would be inherent to the measurement?
As always, thanks for reading. Any and all responses will be read and appreciated.
Is this just the nature of ribbons and planar?
I managed to get the woofer and planar XO halfway decent.
This is designed to be listened to at a desk so it isn't that horrible of an issue but I would like to understand it.
The drivers are not vertically aligned. Here is what the unit looks like. I have entered the offset data into the XO. Even if I removed the X axis offset from the data. Essentially making them vertically aligned, it makes little difference.
Here is the current XO. You can see how big of dip there is at even just the 15 degree mark
Then here are the responses of the tweeter, mid, and woofer respectively. Notice how crazy these responses on. I had to hammer these things into shape to get anything like an accurate response.
Woofer
Mid
Tweeter
I took these measurements with a jig so that the microphone was positioned exactly at the front of each driver when taking the measurement. It was aligned with the face of the unit so that the Z offset would be taken with the data.
I'm sure there is just something I do not understand here. If anyone knows, please let me know. Possibly its just hard using these types of drivers. That I can accept as I didn't think this would be easy. Odd parts cause odd issues.
Lastly, someone mentioned to me to hit ctrl+alt+E to fix phase issues in REW. Does anyone know exactly what this does? It seems to affect the delay. Wouldn't this throw off my Z offset data that would be inherent to the measurement?
As always, thanks for reading. Any and all responses will be read and appreciated.
Last edited:
I don't think phase is meaningful when one dimension of the driver is large relative to the wavelength.
Ed
Ed
I think your array has some inherent challenges because of the arrangement. If you look at most vertical dispersion data in reviews with normal driver layout (drivers arranged in a vertical line) you will see much poorer vertical off-axis frequency response behavior than in the horizontal (the range of vertical angles measured is typically much more limited also). This difference is due to vertical driver spacing vs horizontal driver alignment. Your layout has driver spacing in both axes, so you've essentially distributed poorer dispersion behavior to both horizontal and vertical.
And while 24 dB/octave is better than shallower crossovers in this regard, it's still not the extreme steepness you see in some tweakier designs.
Additionally, you have large vertical size and a high cross point on the planar currently, which will contribute to the chaotic behavior vertically at higher frequencies. You are off axis vertically for the planar given your microphone aiming.
If you want to try to get to the bottom of it all, you could do some off axis measurements of the drivers in isolation and see how they look. You could also rearrange the driver layout to something more typical in your model and see how much each thing is contributing to what you are seeing.
And while 24 dB/octave is better than shallower crossovers in this regard, it's still not the extreme steepness you see in some tweakier designs.
Additionally, you have large vertical size and a high cross point on the planar currently, which will contribute to the chaotic behavior vertically at higher frequencies. You are off axis vertically for the planar given your microphone aiming.
If you want to try to get to the bottom of it all, you could do some off axis measurements of the drivers in isolation and see how they look. You could also rearrange the driver layout to something more typical in your model and see how much each thing is contributing to what you are seeing.
Last edited:
This I very much understand. I knew this going in that there would be design challenges and possibly some inherit flaws. I was limited to 350mm cubed to fit the unit where I wanted it to go. I can accept certain flaws due to the design limitations. I am only trying to see if there is a fix, or if this is just one of those flaws I will have to live with.
Please expand because I am not fully understanding this. I understand that the planar is very long and I have taken the measurement at the center of it. Some parts of the planar are right next to the ribbon while others are 8 inches below. Is this the reason for them going out of phase so suddenly? This I could understand. Where should I have placed the microphone on this driver?
I am crossing high so that the XO isn't in the bandwidth of vocals. Also, according to the distortion graph of the tweeter, it likes to play above 5khz and goes to very high distortion below 2khz.
This is the dip I am trying to fix. I removed everything except 0 and 15 degrees off axis to better show what is happening.
Not sure that's phase related in the directivity graph. More like comb filtering at the mid/tweet crossover. Phase graph shows reasonable phase alignment between the two. You displace the mid and tweeter in both x and y axis - so the usual directivity dips will occur in both the horizontal and vertical axis (as @mattstat has mentioned above). At least this is how I see this. I think this is a compromise you will need to live with to have this arrangement of drivers/baffle/cabinet.
Just wanted to say that I think you've built a really cool desktop speaker 👍
Hope you get it all "to work" to your satisfactory
There is a slight gap between the mid and the front baffle. I'm not at the shop so I cannot take a picture.
Essentially the front face piece is offset from the face of the mid planar by 1.5-2mm. Could this be causing a reflection inside which then causes the comb filtering effect?
I can seal up that area with gasket material so it goes straight up from the mid planar to the "waveguide" instead of it getting trapped in there.
I didn't both because I noticed that most front loaded horns and waveguides are offset a bit form one another.
Do you think this is worth investigating?
Long winded thought. I have made x-overs that didn't have a reverse null. So they were not tracking phase correctly like LR2 or LR4 do. (I did not use these filters, they just happened while trying to come up with a x-over. ) Odd order x-overs are what...??? 90 degrees from being in phase, or something. If drivers are 90 degrees out of phase, and the distance changes, they may become 70 degrees out of phase, and the sum may measure close to the same. So, I guess what I'm suggesting is an odd order x-over with as small an overlap as possible. Maybe 5th order, and accept a small dip, or peak, at the x-over. It's hard for me to accurately put this into words. It would be a WAG, but maybe a mixed 5th, and 3rd order.
I think this is what was happening at the mid to woofer XO. A combination of odd and even order XO cleaned it up and got them somewhat back into phase.
I just cannot manage it at the high frequency XO for whatever reason
I don't think so. Seems too insignificant. I think it's just the two drivers playing together.
If you want to run an experiment... Consider getting some f26 industrial felt and cut out "felt frames" to go around the mid and the tweeter. Kind of like felt wave-guides in a way. I'd make them at least 1cm wide or wider, and have the openings tight to the planar openinig (attach with double sided tape to the drivers). It's thick felt and it is ugly - so not a solution - but you should be able to measure the effect on diffraction and possibly comb filtering and at least you can be certain if that is what you need to work on.
Odds are, it's better to one side than the other. So, you may need the tweeter to the outside of the mid.
I can try it. Around the tweeter it wouldn't be so ugly. I have a laser machine so I can laser out a cool pattern and it will look nice and neat
I did consider this. Space limitations and all again. This is only showing off axis to the right currently. With me siting in the middle the off axis to the left doesn't matter as it is, essentially, impossible with me sitting inbetween them. The left side will be a mirror of the right side.
OK. Why don't you just make it a 2-way since the PT6825-8 has a response of 500-20kHz?
I was just speaking in generalities about the inherent complications of sizes, cross points, and driver spacing. Without being there while you measure things, I'm not sure exactly how things were done for every graph, so my interpretation could be off.
Your image seems to show the microphone aligned with the ribbon.
So let's take a 6.5 inch diameter driver (about the height of the planar, I think) and look at its off-axis response, assuming perfect pistonic behavior:
5°
15°
These effects are just path length difference issues related to the driver's size and getting off axis.
Given how you intend to use the speaker, on the listening axis for the whole array makes sense. You may also want to measure at the intended listening distance, if you aren't already. This is all a reminder that the higher your cross point, the larger your drivers, the closer your listening position, etc., the smaller your listening sweet spot can become due to path length issues between the drivers or between different areas of a large driver. If you're crossing in region where a few degrees makes a significant difference in frequency response, things are inherently dicier. Doesn't mean it can't be done, just that things tend to be fussy and you may need more tuning by ear in the end to get things to sound right.
This matters as well. You should measure and model based on the listening angle you intend. With spaced drivers you can typically steer the primary lobe of dispersion with crossover adjustments. You'll sacrifice frequency response smoothness in other non-listening axes, but it can be the right approach to broaden the listening window a bit.