First FR with SW

[Vikash]

The response doesn't look particularly good, but I don't know how these Speaker Workshop scare stories came about. Wasn't too difficult getting it to spit out a frequency response at all (thanks Claudio!) albeit a little laborous

The measured speaker is the TB-W3871S monitor. The top end looks very unconvincing even with the notch applied (help! is this a gating issue possibly? and how do you apply the baffle step into the equation to sort that bottom end out?

All in all it's a straighter squiggle than I was expecting for a first effort.

 

[Vikash]

A bit more playing around with the view and the top end doesn't look too out of place all of a sudden.

Using the same scale as Zaph, and superimposing his FR a little higher up, it's actually comparable - so something's right somewhere . There seems to be alot more data in his mid and low end response, what's have I missed here?

And how do I sort that blasted bass hump. I'm still guessing that accouting for the baffle step will smooth it out?

 

[ShinOBIWAN]

A bit more playing around with the view and the top end doesn't look too out of place all of a sudden.

Using the same scale as Zaph, and superimposing his FR a little higher up, it's actually comparable - so something's right somewhere . There seems to be alot more data in his mid and low end response, what's have I missed here?

And how do I sort that blasted bass hump. I'm still guessing that accouting for the baffle step will smooth it out?

Hi Vikash,

I had some similar problems with SW. Its great for measuring T/S specs etc. though.

I'll mail you and try to help out.

 

[lufbramatt]

hi
the last few days ive been playing around with SW, and id agree that it was nowhere near as hard as i had been led to believe, but ive still got to do a bit of fine tuning as my graphs still seem a little wierd- im getting a huge max impedance at the Fs of my scanspeak 8555 woofers ive been testing, although if i smooth the graph out a bit the estimated TS params are very close to the published ones. Ive been using a hastily badged together claudio negro cable loom jig, which seems to work fine, but im going to build a slightly more permanent version this weekend.

I have had a little play with the FR function using the mic and preamp kit i got from vikash- cheers it went together really well - so will be interested to hear peoples comments here.
My main problem at the mo is the high ambient noise level here in uni halls- think im going to have to stay up really late to wait for all the drunkards to fall asleep before i can do some decent testing!

 

[soongsc]

A bit more playing around with the view and the top end doesn't look too out of place all of a sudden.

Using the same scale as Zaph, and superimposing his FR a little higher up, it's actually comparable - so something's right somewhere . There seems to be alot more data in his mid and low end response, what's have I missed here?

And how do I sort that blasted bass hump. I'm still guessing that accouting for the baffle step will smooth it out?

Looks like baffle diffraction (do not confuse with baffle step). Depending on how far the measurement was made, the low2 end might be okay. But with this kind of data, I wonder what you hear when organs play. What notch values are you using?

 

[Vikash]

lufbramatt, the wallin jig is defintely worth considering It's just so easy with it.

Zaph measured his response using TrueRTA which he reports didn't have gating at that time. My pulse response shows the first pulse at 4ms and the second at 6ms! So not much data to play with in the gating. I'm not sure to what extent the effects of the above are, but I'm sure there is some.

A little excerpt from from John's write-up:

The W3-871 is a little different from other smaller drivers in that it doesn't need much baffle step compensation at all due to the natural response curve. The filter is a simple 3 component notch. This notch serves two purposes, it not only reduces a small bump in the speaker's natural response, but the main baffle step hump is in the same vacinity.

Ok, so the low end rise is ok and the nearfield measuring technique doesn't show the effects of baffle step (sanity check please)? which would otherwise show a flatter low end response...

soongsc, Baffle edge diffraction refers to the rough mid/top end right? 5.75" wide box with 6.8ohm-3.3mfd-.7mh notch

Measurements were taken in my converted attic with sloped shallow wall/ceilings. Farfield measurements were taken from 1m, and spliced with nearfield at 750Hz.

 

[Nemophyle]

hello vikash

i think that what zaph is saying here is that this driver, due to the drooping response that we see on your graph, dont need Baffle step compensation , so if you want the expected response youd better do farfield measurements all the way ...

 

[soongsc]

lufbramatt, the wallin jig is defintely worth considering It's just so easy with it.

Zaph measured his response using TrueRTA which he reports didn't have gating at that time. My pulse response shows the first pulse at 4ms and the second at 6ms! So not much data to play with in the gating. I'm not sure to what extent the effects of the above are, but I'm sure there is some.

A little excerpt from from John's write-up:

Ok, so the low end rise is ok and the nearfield measuring technique doesn't show the effects of baffle step (sanity check please)? which would otherwise show a flatter low end response...

soongsc, Baffle edge diffraction refers to the rough mid/top end right? 5.75" wide box with 6.8ohm-3.3mfd-.7mh notch

Measurements were taken in my converted attic with sloped shallow wall/ceilings. Farfield measurements were taken from 1m, and spliced with nearfield at 750Hz.

Not knowing what room dimensions you have, but you could try adjusting the measurement setup around 45 degrees to all walls, and let the speaker backside face the down side of the ceiling slope. This will possibly delay the first reflection a bit more. If you haven't smoothed the curve, it looks pretty clean. I'm amazed.

Yes, the two humps starting from the mid to the right. The extreme highs is possibly the driver itself.

 

[Vikash]

I'll move things around see if I can delay that first reflection a bit. Only the nearfield below 750Hz has been smoothed to 1/8 octave as per Claudio's webpage. The rest is as-is.

I went through John's article again and picked up things that I tended to ignore in my pre measurement days

Namely he mentions the baffle diffraction dip at ~3Khz, and a dip and peak at 6Khz and 12Khz respectively related to flush mounting. I wonder if those socket head bolts are having an effect too?

 

[soongsc]

I'ts probably also good to look at the manufacturers FR plot. If some of the trends match, then it's most probably the driver itself. There's also a dip at the 1KHz.

 

[navin]

I'ts probably also good to look at the manufacturers FR plot. If some of the trends match, then it's most probably the driver itself. There's also a dip at the 1KHz.

me thinks the dip from 200-1k is baffle step. baffle difraction can be reduced by rounding edges and applying felt cloth to them. felt might also help with te counter sink. right?

 

[rabw]

me thinks the dip from 200-1k is baffle step. baffle difraction can be reduced by rounding edges and applying felt cloth to them. felt might also help with te counter sink. right?

hi navin
baffle step is caused by the driver changing from radiating into 4pi space at high frequencies (ie front and back) to 2pi space (just in front) as the wavelength of the sound wave get longer at lower frequencies. So it needs more energy to maintain the same sound level as the frequency decreases, so the sound gets "quieter" in front of the driver for the same power level. Therefore a baffle step "dip" gives a slope that is the other way around- dB slopes down from high to low frequencies, not as seen here.

im sure the baffle step effect is a result of the size of the smallest baffle length- so rounds and felt only serve to smooth the transition- having a really really wide baffle would be the only way to get rid of it

regards, matto.



edit- rich stop leaving ureself signed in my my computer! post really made by lufbramatt

 

[lufbramatt]

hi navin
baffle step is caused by the driver changing from radiating into 4pi space at high frequencies (ie front and back) to 2pi space (just in front) as the wavelength of the sound wave get longer at lower frequencies. So it needs more energy to maintain the same sound level as the frequency decreases, so the sound gets "quieter" in front of the driver for the same power level. Therefore a baffle step "dip" gives a slope that is the other way around- dB slopes down from high to low frequencies, not as seen here.

im sure the baffle step effect is a result of the size of the smallest baffle length- so rounds and felt only serve to smooth the transition- having a really really wide baffle would be the only way to get rid of it

regards, matto.



edit- rich stop leaving ureself signed in my my computer! post really made by lufbramatt

ignore me, i missed some posts and kinda missed the point too, sorry

 

[soongsc]

me thinks the dip from 200-1k is baffle step. baffle difraction can be reduced by rounding edges and applying felt cloth to them. felt might also help with te counter sink. right?

Some people have measured differences using some sort of foam or thick cloth. I recall a site where a square cloth or somthing of the sort was used with better results than the round ones then surrounding a speaker. I think you might be able to hear differences, but the measured effects would be less than the proper rounded edges. Many people indicate that round edges need to be at a radius at least 1/4 of the wave length of the frequency in question. Smaller radiuses are more for cosmetic reasons.

There should be a few threads that talk about this.

 

[claudio]

originally posted by Vikash:

I'll move things around see if I can delay that first reflection a bit. Only the nearfield below 750Hz has been smoothed to 1/8 octave as per Claudio's webpage. The rest is as-is.

Well, I think I wrote to smooth to 1/8 octave the complete response, NF + gated FF.

In my opinion, the 2 response cannot be compared since the measurement method is different: Zaph used a calibrated mic, what about your? Zaph don't say at which distance the mic was, and since his curve is unsmoothed and ungated, I believe he used a very large room. Look at your curve high end: looks like you have a flatter response, that recalls the driver free air response, visible in here: http://www.zaphaudio.com/minitest/fr.html

About your wonder if those socket head bolts are having an effect too, I did some test once, and their influence was close to zero.

Regards,
Claudio

 

[navin]

hi navin
baffle step is caused by the driver changing from radiating into 4pi space at high frequencies (ie front and back) to 2pi space (just in front) as the wavelength of the sound wave get longer at lower frequencies.

yup you are right it should have been a rise instead of a drop. my mistake.

 

[planet10]

baffle step is caused by the driver changing from radiating into 4pi space at high frequencies (ie front and back) to 2pi space (just in front) as the wavelength of the sound wave get longer at lower frequencies.

baffle step is caused by the driver changing from radiating into 2pi space at high frequencies to 4pi space as the wavelength of the sound wave get longer at lower frequencies.

dave

 

[Vikash]

In my opinion, the 2 response cannot be compared since the measurement method is different: Zaph used a calibrated mic, what about your? Zaph don't say at which distance the mic was, and since his curve is unsmoothed and ungated, I believe he used a very large room. Look at your curve high end: looks like you have a flatter response, that recalls the driver free air response, visible in here: http://www.zaphaudio.com/minitest/fr.html

About your wonder if those socket head bolts are having an effect too, I did some test once, and their influence was close to zero.

The mic isn't accurately calibrated per se, just with the generic WM61A calibration file.

Even though the driver is countersunk the dip/peak is still there at 6/12khz. This is why I wonder how much affect the bolts have.

 

[soongsc]

The mic isn't accurately calibrated per se, just with the generic WM61A calibration file.

Even though the driver is countersunk the dip/peak is still there at 6/12khz. This is why I wonder how much affect the bolts have.

Do you get similar results if you use the other speaker? I assume you have at least two. What about after taking out the cap in the BSC? Best do all far field at lease 1M distance.

 

[wintermute]

Hi Vikash,

Dickason claims that nearfield is only accurate (and usable) up to about 200Hz. After that point the measured SPL gets progressively lower than the actual SPL. I've noticed this on my own measurements, and it usually really does start dropping off at about 200Hz. the effect in your measumrements look a bit more pronounced than mine, but I suspect it is the scale

Unfortunately getting good gated data down to 200Hz is nigh impossible unless you have an auditorium at your disposal

As already mentioned try changing the angle to walls, but also try measurements at different heights off the ground. I found that this made the most difference, you can almost completely get rid of the floor bounce if you have the angle right, and use something like eggcrate foam on the floor between the speaker and mic

One other suggestion I got from a pro was that the "stand" that you put the speaker on is very important. If you stick it for instance on something with a solid front, then it will act as an extended baffle, and effect the measurements. ideally I guess you want a pole with a stand on top no bigger than the base of the speaker (I have been using a table..... not the best.....

and just on the comments about difficulty of speaker workshop, I think it depends on what documentation you have at your disposal before I knew about jay's manual or Claudios site I had very mixed results (couldn't get consitency) using just the docco that came with SW Oh and I think it also makes a big difference what sound card you have!

Welcome to hours and hours of testing

Oh and one final suggestion. Get a log book, and write down everything about the various tests. height of the speaker (center) off the ground, distance to the mic, power level used (Voltage on the driver) , what the VU readings were in SW, (and if you are really keen temp and humidity..... I haven't resorted to that). When you can't get the same measurements at a later date, you can go back see what you did, and hopefully reproduce

Tony.