Hi!

Last year we built pur own syntripps. We already used them quite a bit, but finally got a round to do some outdoor measurements. And what can i say, it looks like hot garbage, nothing like the measurments Art Welter posted.

Our SyntripPs consist of 2 B&C 10CL51 8 Ohm and one Celestion Cdx14-3050 each. Some few parts including the phase plugs and throat adapter we 3d printed in petg. Our DSP is a DBX driverack pa2 and our amp a cvr d-654. Sound source is the laptop with open sound meter

here is a free field measurment done in opensoundmeter, mid and high driver measured seperately. Mic used is a calibrated minidsp umik-1.

https://imgur.com/a/vsnylb6

So we first measured using the dsp settings suggested in the syntripPs thread: https://www.diyaudio.com/community/threads/syntripp-2-way-2-part-virtual-single-point-source-horn.264485/post-4115408
Hot garbage. There is a clear dip at ~750Hz, and recession in the hf.

We only got around to try and fix the +750Hz dip, and we tried lots of things. For the measurement above, in the dsp crossover, we put the mid driver lp to 1khz and the hf driver hp to 400hz, and as you can see the dip is still there. It appears the box is not really playing the frequency band at all or very quiet. Ther is no eq applied

So we are very much at our limited wisdoms end, and would love to hear some insights from you. if you need any more information, let me know. Would love to hear some insight.

Thanks

Did you apply the 0.4ms delay mentioned in that link?

I’d look at this and also double check polarity is correct.

Ideally use smaart to find the true acoustic delay required.

Did you build the reflex ports or sealed?

You can always reverse the HF or MF and input a sine of X-over freq, adjust both channels so they sound the same level then increase MF delay (from zero) untill you get max cancellation.

Now un-reverse your signal and the x-over point should be as close as it gets without measuring. Now only phase shift will get some cancellation or summation in the x-over overlapping range.

adamb00m wrote: Looks like a time alignment / delay issue possibly. Did you apply the 0.4ms delay mentioned in that link? I’d look at this and also double check polarity is correct. Ideally use smaart to find the true acoustic delay required.

We first used the 0.4ms, yes. Then we adjusted the delay bit by bit, but it only improved the dip marginally. We triple checked polarity as well. We don't have money for smaart sadly, but why would that delay finder be better than the one opensoundmeter uses?

Jan-2T wrote: Did you build the reflex ports or sealed? You can always reverse the HF or MF and input a sine of X-over freq, adjust both channels so they sound the same level then increase MF delay (from zero) untill you get max cancellation. Now un-reverse your signal and the x-over point should be as close as it gets without measuring. Now only phase shift will get some cancellation or summation in the x-over overlapping range.

With reflex ports.

Interesting, will try next time we measure

You have any exact details of the setup? Hight from ground, dustance from DUT etc etc.

I have a picture of the setup here:

https://imgur.com/a/GMOnAbL

Hmmm but it's nowhere near as severe in Art's measurements, or am I mistaken? In our measurements it's a ~12db dip, in art's it's only ~6db

fudge22 wrote: The deep notch in response at 63Hz could be due to your measurement set up. Try moving the microphone, speaker or both to see if the frequency of notch varies. Also, you could measure the the loudspeaker with a ground plane, which I believe is how the original was measured, it should smooth out the low end response.

Yeah those frequencies I just ignored really, we got some subs taking over from 100Hz down

fudge22 wrote: The funky response of the mid drives, at around 1.6KHz, would normally be of little consequence, because it is out of the operating range and filtered out, either by the cross-over filter or a combination of that and an out of band parametric eq to get rid of any residual leftovers. However, that the high frequency driver shows an almost inverse response over the same frequency range is interesting. As a guess, I would suggest that it is due to some reflection or resonance near to the throat of the horn. If you know how to interpret the phase response plot, that may help.

I actually have the theory that we wired up the hf driver incorrectly. Because there definitely is no inverting happening in the dsp.

fudge22 wrote: The first thing that I would try is covering the openings for the mid drivers, in the throat of the horn, and re-measuring the high frequency driver, to see if the high frequency response smooths out.

Hmmmm, would just closing the mid driver horn ports affect the hf measurement so much?

snowflake wrote: when you measured the tweeter did you have the mids either connected to an amp? If not there is not electrical damping. Or you can just short the mid drivers with a wire. Not sure this is the cause but it's easy to try.

Place the syntripp also on the ground, lets say 4m from mic (if possible) otherwise a bit less would also suffice. Place the mic capsule also as close as possible to the ground and try to get a hard and flat surface. Angle the syntripp in a way that the middle points directly at the mic capsule. 

That way you get the best actionable data to verify possible problems.

The response graph in the post that you linked to seems to be post processing. My comments were regarding a comparison with a raw response. Also note that in the Smaart plot the value of the smooth setting is 9. This averages the amplitude at any frequency with adjacent frequencies. To quote the manual, with smooth set to 3, any given data point will represent the value of that point averaged with the next higher and next lower points on the trace. When smoothing is set to 5, each point is averaged with the next two higher and lower points etc.

The below image (from the sameDIY Audio thread) claims to be a raw response.

If you eq the response so that it is flat above 200Hz the notch at 700Hz doesn’t look so bad; especially if the measured graph is smoothed.

The anomaly at about 1.6KHz also shows up in this raw response.

I actually have the theory that we wired up the hf driver incorrectly. Because there definitely is no inverting happening in the dsp.

When used in isolation, the polarity will have no effect on the frequency response. If the high frequency response was measured with the mids muted, the polarity is irrelevant whether due to wiring or dsp setting. Inverting the signal won’t change a peak in response to a notch. The phase/polarity only becomes relevant when there are multiple sound sources covering the same frequency range. That said, a reflected sound wave also acts as a secondary source, which will combine with the direct wave, either constructively or destructively.

Hmmmm, would just closing the mid driver horn ports affect the hf measurement so much?

The ports will have some effect, so conversely, eliminating them should too. Just how much, I don’t know. Apart from the coupling between the compression driver and horn, holes in the walls of the horn would seem to be the only other cause.

Tracking down problems is a case of eliminating what is not causing the problem. Covering the ports with duct tape is quick and easy, and will either confirm of eliminate the ports as the problem.

For what it's worth, I also got some very weird / crappy results when measuring mine, but it was my first time measuring and well, I got a few food pointers that made me think it was better to start the measurement process again rather than try to analyse or correct my wonky results.