HiFi Subjectivist ¦ Audio Forum ¦ NVA User Group

Unfortunately you have only answered a part of my two questions. What did you learn from the measurements you did that you did not know before from just listening to the mods you carried out.

When I did Doc mods to my JPW's and when I made my Cube 1 clones, I just used my ears to ascertain the amount of doping to the drivers and the value of the protection capacitors - big green Russian paper in oil thingy's. The size of hand grenades they are. I played around (a lot I must be said) until I had a presentation I liked.

If you produce a frequency response curve for a speaker, having first built it to be subjectively satisfactory, you will have a reference against which you can evaluate other speakers.

That assumes that the other speakers were measured in the same way using a similar microphone and measuring software, otherwise there will be differences that you can't account for.

I suspect however that the main result of your measurement will be to demonstrate just how wrong the idea that speakers require a flat frequency response is in practice.

I measure when I'm not happy with something so I can look for a reson why. If for example there is a hole in the response that I can hear then I need to have an idea of what frequency range it is in so that crossover component values can be altered to alleviate the problem.
If treble is spikey the plot will show where the spikey bits are and a notch filter could be added
Just examples, but you get the idea. I'm not inclined to measure for the sake of it, my earoles tell me wether I need to measure in the first place.
The big 15" sealed boxes haven't really needed measuring, they have no crossover at all, they have been voiced by ear

Daniel Quinn wrote: ↑Tue Oct 08, 2019 5:41 pm The frequency response is a must measurement for speaker builders, to augment your ears

I agree with Daniel on this one. Whilst one's ears are are the final arbiter, it's helpful to know why something is sounding as it is. For example, honky vocals might point to peaks in the 800-1500hz region. A quick bit of measuring could well get one closer to a decent result more quickly than trial and error changes done by ear alone. Similarly, piercing or painful vocals point to the 4khz region for attention and this is usually pretty obvious in any measurements.

OTOH, being a slave to measurements and EQing can give some undesirable results, depending on the impact of room modes, reflections, reinforcements etc. For example, I tried to EQ out some measured nulls in the measured response and the results were horrible.

Therefore, I use REW to give me initial ball park measures and then refine by ear from there. I've used a Samson USB Go Mic but I have no idea as to how well it's actually calibrated. I have a Behringer measurement mic too. I can import REW EQ settings directly into my miniDSP and then tinker from there.

I feel a bake-off coming on...

With reference by various speaker measurements in another thread. I'm pretty sure frequency response is the only one you validly need.

Earlier in the year it occurred to me that it would be possible to use MS Excel, an on-line sine wave generator, and the dB app on my phone to entertain myself.

No idea how accurate are the sine wave generator, the microphone in my phone or the dB app software, but hey-ho, fun was had.

Specifically, I was interested in how my room effects the sound, and so came up with various graphs, including these 2.



The top one is data recorded with the iPhone placed where my head is when I'm in my favourite listening seat (i.e. where room effects are well in play).
The bottom one is data recorded with the iPhone close to the speaker (about 80cm in front of the left speaker, about 50cm off the ground) i.e. where, due to proximity to the speaker, the measured dB level has somewhat more to do with the speaker directly, and somewhat less to do with the room effect.
(Frequency on x-axis, dB reading on y-axis.)

I'm not too interested in any precise dB values differences between the 2 graphs (as in both cases, I just set the volume to achieve around 80dB), but rather in the difference in the overall shapes of the graphs.

Over the years, all my work tweaking the speakers and my room has been done (by ear) sitting in my preferred listening position (i.e. from where the data for the top graph was taken). So, I'm not too surprised (but relieved) that from 90Hz the response curve on the top graph (I think) appears to be within narrower limits than when it's measured closer to the speaker.

I've no intention whatsoever of doing anything meaningful with this; I'm happy with my existing process of judging by ear. Just a bit of fun that I want to share.

Also, the close measurements (bottom graph) at the low end vaguely resemble the original curve for the 18P supplied by Goodmans back in the 70s.



An initial slope up to 100Hz, and then an additional incline from about 400Hz.
The fall away on my bottom graph from 600Hz is probably the crossover kicking in (not present in the Goodmans graph of course).

("Specialist applications in fairgrounds". I just love that. )

Here they are btw.