Polk T15 Bookshelf Speaker Review

[Rick Sykora]

Cannot say I had high expectations, but I thought the T15s would do better than Amir has found...

In case you missed, I shipped a Dayton C-Note ($100/pair DIY speaker) to Amir. So here is my initial measurement on them. This was done at 1m, gated and unsmoothed. Note they are being driven to REW's reference 75 level (pretty loud in a spare bedroom!)

Black trace is on-axis and red trace is 45 degree off-axis.

 

[Billy Budapest]

For about the same amount of money as the T15, a speaker with much better fit and finish and much better sound exists: the NHT SuperZero 2.1.

I mean, how is the Polk logo not even silkscreened correctly on these? Just sloppy. Compare that to the piano black polished finish on the SuperZeros, with beveled edges hiding the seams. It’s amazing how inexpensive they are given their performance and build quality.

 

[tktran303]

I think Amir should measure, reflect, trouble shoot and learn. That is what science is about IMHO.

I appreciate he has limited time, but we all do. Who doesn’t? If the aim is measure as fast as possible and move on the next, like a production line, what is the goal? To get as many data points as possible, build a critical mass of subscribers, and then sell out (get advertising, get bought out?

If not, then the process needs to slow down.
Dropping measurements for public consumption and a thousand replies with low signal to noise ratio; whilst moving on to the next- collectively, what can we learn from that? eg. Kali IN-8
how many hours (collectively) have been typed for all the responses?

Amir,
One cannot simulate it to a high degree of precision without knowing the on-baffle measurements of woofer and tweeter AND the crossover circuit.

This clearly has a problem as the crossover point; invert the tweeter connection/phase will determine if helps or not.

ie. Is it an assembly line error (less likely if there are two measurements from publications showing this 3KHz dip) or a design error (more likely)

The tweeter is the “not the reference axis” is not the correction explanation because the predicted in room response and listening window still shows serious problems.

This is an example where a single on-axis measurement done by a $50 microphone and free software eg. REW is enough to show a serious problem.

but can it be quick fix (unscrew 3 screws and invert tweeter connection), cheap fix (modified crossover needed) or no fix (don’t buy)

 

[amirm]

I think Amir should measure, reflect, trouble shoot and learn. That is what science is about IMHO.

I am not creating science. I am providing critical information consumers lack in making purchase decisions.

I appreciate he has limited time, but we all do. Who doesn’t? If the aim is measure as fast as possible and move on the next, like a production line, what is the goal? To get as many data points as possible, build a critical mass of subscribers, and then sell out (get advertising, get bought out?

What do you mean as fast as possible? You have comprehensive measurements that normally requires multi-million dollar anechoic chamber. Even then, measurements are made at 10 degrees each whereas mine are accurate to 1 degree. I then spend good number of hours analyzing the results, documenting it in graph and review text and then post. Following that, I spend time reading and responding to questions and comments about the review.

This is a $100K measurement system precisely because it provides high throughput. I could sit there and manually spin a speaker 10 degree at a time and blow an entire weekend measuring speakers as JA does in Stereophile. I spent the money not to have any hope of getting that back let alone get rich beyond that, but to be able to make a dent in available number of speakers out there. This model has worked with audio electronics and I think it will be even more effective with speakers because their differences is much larger.

To sit there and keep playing with a budget speaker, redesign its crossover and such is not part of my charter. Danny at GR research does that. Contact him and see what he charges to redesign the crossover in this speaker. I will send him my pair to mess with and then we remeasure. He recently did the change you are asking about: reversing polarity of the tweeter but found that it did not really fix the problems.

Back to your point of selling out, at some point I will either be too old to do the work, or too grumpy to want to deal with all the objections. So there has to be a succession plan. It makes no sense for me to let all of this die with me. When that is, depends on my health and what you all do to me between now and then.

 

[amirm]

Is it an assembly line error or design error.

There is no assembly error as James Larson measured the same thing I did as noted earlier in the thread (except for some reason he chose to not post the spin graph as I did where the problem becomes very clear to see).

Polk is a pure marketing team. They know what it takes to sell speakers. Nothing in engineering can override marketing. So whatever they have here is forced upon them by the need to make money.

 

[tktran303]

Apologies if you feel my post is attacking you.

it is not. English is not my first language.

but you said it yourself:
“This is likely due to tweeter and woofer getting out of phase and causing cancellation on-axis.”

Can the data drop be paused until the problem is found?

 

[infinitesymphony]

What amplifier is used for testing passive speakers? I'm sure this was covered generally elsewhere.

 

[amirm]

What amplifier is used for testing passive speakers? I'm sure this was covered generally elsewhere.

For measurements or listening? Measurements are with a Behringer pro amp. Listening is with Purifi amp.

 

[gr-e]

He recently did the change you are asking about: reversing polarity of the tweeter but found that it did not really fix the problems.

With polk or klipsch rp-600m? It wouldn't help the latter because both drivers are -10dB at the crossover point

 

[pma]

“This is likely due to tweeter and woofer getting out of phase and causing cancellation on-axis.”

Can the data drop be paused until the problem is found?

It looks like crossover design problem. Probably oversimplified crossover circuit. Inverting tweeter phase would probably not fix the issue, but just transform it to another issue.

 

[infinitesymphony]

For measurements or listening? Measurements are with a Behringer pro amp. Listening is with Purifi amp.

What determines if an amplifier is good enough for measurements (i.e. why not a Benchmark AHB2)? Wondering about power output, impedance handling, SINAD, etc.

 

[Robin L]

Found 2 of these at an Amvets for $8 the pair. Decent, but compared to what I finally wound up with [the first iteration of the Infinity "Primus" surround speaker system, 360 fronts, 250s in the back, C25 center, also Amvets, $80 total] they turned out to be superfluous. A friend ended up with them. Kinda thin and nasty sounding, as I recall.

 

[mhardy6647]

1) I'm reminded of an old Peter Aczel quip about the Boston Audio Society... something to the effect that they are nice guys looking for $49.99 audio nirvana. @amirm has to be at least a spiritual member of that BAS crowd, spending time with this sort of thing. I give him credit--he's a hard working guy and no one else is doing it at the low end.

2) I don't know much about the company nowadays, however early Polk was good value. Your photo is a Monitor 10? I've told this story before, but I recall comparing a DQ-10 (Jon Dahlquist/Saul Marantz's Quad look-a-like) with Polk's RTA-12 (not SDA). This was circa 1977-8. DQ-10 was all the rage in the high-end, but I came away distinctly preferring the coherency of the Polk over what I thought was a rather 2-D and harsh sounding Dahlquist. I haven't heard one of the early Polk speakers in decades. But I imagine they would hold up respectfully.

Yes, that was a Monitor 10 in the photo posted above by @Prana Ferox. Looks like a 10B, with the oft-maligned SL2000 "Silver Coil" tweeter. It has an unfortunate spike in the treble (which, at my now-advanced age, I don't find all that objectionable, truth be told). There are 'modern' drop-in replacements that are still available from Polk (AFAIK) which are much (much) better.

FWIW, here's the original morph of the "Monitor Audio Series Model 10", using a (Danish) Peerless silk dome tweeter and uncoated paper passive radiator.

P1020559 by Mark Hardy, on Flickr
GMmusicPolk10primosystem by Mark Hardy, on Flickr

I was never too crazy about the DQ-10 (FWIW). Actually, I wasn't all that crazy about the any of the versions of the Monitor 12/RTA-12, either. The Monitor 7 (and the slightly later, smaller, Monitor 5) were - IMO (and FWIW) - the sweet spots in the early Polk lineup.

AH477etc by Mark Hardy, on Flickr

Aug. 1978 ad from either Audio or Stereo Review via www.americanradiohistory.com

 

[ROOSKIE]

I think Amir should measure, reflect, trouble shoot and learn. That is what science is about IMHO.

I appreciate he has limited time, but we all do. Who doesn’t? If the aim is measure as fast as possible and move on the next, like a production line, what is the goal? To get as many data points as possible, build a critical mass of subscribers, and then sell out (get advertising, get bought out?

If not, then the process needs to slow down.
Dropping measurements for public consumption and a thousand replies with low signal to noise ratio; whilst moving on to the next- collectively, what can we learn from that? eg. Kali IN-8
how many hours (collectively) have been typed for all the responses?

Amir,
One cannot simulate it to a high degree of precision without knowing the on-baffle measurements of woofer and tweeter AND the crossover circuit.

This clearly has a problem as the crossover point; invert the tweeter connection/phase will determine if helps or not.

ie. Is it an assembly line error (less likely if there are two measurements from publications showing this 3KHz dip) or a design error (more likely)

The tweeter is the “not the reference axis” is not the correction explanation because the predicted in room response and listening window still shows serious problems.

This is an example where a single on-axis measurement done by a $50 microphone and free software eg. REW is enough to show a serious problem.

but can it be quick fix (unscrew 3 screws and invert tweeter connection), cheap fix (modified crossover needed) or no fix (don’t buy)

I don't think the point of this is to fix speakers. This projects seems to be about measuring them as is, so when someone buys them they have some measurements to help guide them toward a good speaker. I personally don't want a single extra minute wasted testing any fairly poor speaker. That is time away from making some measurements of another speaker.
Not many people who I trust think only measurements tell the story. As well measurements are great to have. IMHO as someone who is fairly new to this site, these measurements are very appreciated. I just want to see more meat on the speakers tested so we will have to submit some of our gear. Nobody can be expected to buy dozens of $500-2k speakers yet I am sure those are the ones most of us want to see tested.
I am sure some DIY peeps and or moders will eventually make some changes that we might want to see remeasured in some speakers. I know Dennis Murphy had at one point moded the entry Pioneer bookie and it was essentially universal accepted as a budget upgrade worth doing.

 

[AudioTodd]

The only problem with the Panther Scale is that the resolution is fairly low. I am not even sure if it is linear or logarithmic -- so I like to look at the data and commentary and work out the correlation.

Y'all probably know this, but that's how the world got Viagra. Pfizer was testing a drug for cardiovascular indications, but the test subjects in the clinical trials reported some interesting side-effects. Couldn't find a really good synopsis outside of the peer-reviewed literature (that costs $ or requires,e.g., academic library access credentials to read), but here's an example: http://resources.schoolscience.co.uk/pfizer/viagra/viagch5pg1.html

Well... if you got lucky, perhaps you could find a pair of these. (For the record, I am still lookin' for a pair).
View attachment 51951
View attachment 51952
Full disclosure: I am a long-time Polk fanboy and have owned a pair of their early Monitor Series Model 7A ("Monitor 7A") since 1978.

“...intended primarily for...aircraft sound systems...where sound of the highest quality is desired...”

Hmmm... never been in one of those aircraft I guess. Never been in ANY aircraft where this, or any size, speaker would audibly produce “sound of the highest quality!”

 

[infinitesymphony]

I don't think the point of this is to fix speakers. This projects seems to be about measuring them as is, so when someone buys them they have some measurements to help guide them toward a good speaker. I personally don't want a single extra minute wasted testing any fairly poor speaker. That is time away from making some measurements of another speaker.

Exactly, that kind of thinking leads to the belief that we are one corrective EQ curve away from making the JBL LSR 305P MkII sound like a Genelec.

There are modeling microphones and amplifiers in the studio world, and in those and other cases you have to start with something relatively flat and full-range as your block of marble and then chisel away at it from there. Doesn't work as well the other way.

 

[Sancus]

I think Amir should measure, reflect, trouble shoot and learn. That is what science is about IMHO.

I am extremely uninterested in modifications to sub-$100 speakers personally and I don't want Amir wasting his time re-testing such products either. The speaker reviews are product reviews. They're not "product + 10 hours of DIY time" reviews.

There's at least thousands of products out there worth testing. The objective is to find the good, the interesting, the bad, and the ugly. It's not to try and fix the broken ones.

 

[edechamps]

I am extremely uninterested in modifications to sub-$100 speakers personally and I don't want Amir wasting his time re-testing such products either. The speaker reviews are product reviews. They're not "product + 10 hours of DIY time" reviews.

Agreed that going to such extremes is perhaps not the best use of @amirm's time. On the other hand, simpler things like re-processing the measurement data on a different axis when it's extremely obvious the best axis is not the normal reference axis (here, the response is much better at 30°V) could make sense. Especially since, as far as I understand, it doesn't require a lengthy re-measurement process.

Clearly this speaker is capable of significantly better sound than shown on the spinorama and predicted in-room response, as long as you place it below or above ear level. But we can't do advanced processing such as accurate preference score calculation on anything other than the axis @amirm decided to use.

 

[MZKM]

What determines if an amplifier is good enough for measurements (i.e. why not a Benchmark AHB2)? Wondering about power output, impedance handling, SINAD, etc.

The Benchmark is 100W into 8ohm, which may not be powerful enough to do stress testing. For just Spin data, it shouldn’t matter.

I asked this too, and Amir said that the measurements of the A500 meets the requirements from Klippel and he had it on hand for use.

However, @amirm, the A500 was measured to have a -0.7dB response at 20kHz. While the preference scoring doesn’t care about responsive that high, it is very slightly down even at 8kHz. This is the FR without even any speakers attached, if the A500’s output impedance is high, it will alter the FR even more, and not uniformly. It would be a good test to use a more highly linear amp (with really low output impedance) and the A500 and see what differences arise (especially with a speaker with low impedance).

 

[infinitesymphony]

The Benchmark is 100W into 8ohm, which may not be powerful enough to do stress testing.

Benchmark AHB2's bridged mono mode would be perfect for this application because we're only testing one speaker at a time. In that mode, it gets 380W at 8 ohms and it's rated down to 6 ohms. Might be able to go lower:

"The AHB2 is completely stable into 2.2 Ohm loads at full output level...we’ll be driving the AHB2 with a pair of Revel Ultima2 Salons. The Salon2s are nominally 6 Ohm speakers, but their variable impedance drops dips down into the upper 3 ohm range from about 40 Hz – 90 Hz and again at 200 Hz."