Polk RC80i In-Ceiling Speaker Review

[amirm]

This is a review and detailed measurements of the Polk RC80i in-wall/in-ceiling speaker. It was kindly drop shipped by a member and costs US $249 for a pair ($125 each).

Overall look is nice but that 1inch tweeter looks cheesy to me. There is a white metal grill which you had to push out with the mounting screws. Woofer is rated at 8 inches.

Notice the 5000+ reviews on Amazon averaging 4.5 stars.

I measured it using Klippel Near-field Scanner "baffle" mode where it computes the anechoic response of the speaker as if it were in an infinite baffle (getting rid of back reflection and edge diffractions).

POLK RC80i Measurements
Let's start with our usual CEA-2034 measurements even though it doesn't fully apply here:

Story starts good but gradually gets worse until we see massive ups and downs in tweeter range. I am guessing those are interferences from the tweeter to woofer and back. Listening window moderates them so they are worst on-axis. I measured the near-field response of the tweeter (not shown) and it was much cleaner. Given the variability in power response, EQ will be hard to do.

We see similar issues in directivity which is the same vertically as it is horizontal due to center position of the tweeter:

Normally you want a wide sweet spot for an in-ceiling speaker but RC80i delivers the opposite. Off axis is going to have significant colorations depending on angle.

Listening to the sweeps in the lab, I was pleasantly surprised by the power handling at 86 and 91 dBSPL. Only at 96 dBSPL did it get a bit unhappy:

I am working on better analysis of the distortion. In that regard, I computed the difference between 86 and 96 dBSPL. In an ideal world, as SPL would rise, distortion would rise the same amount, keeping the above ratio the same. Naturally we are already into non-linearity which means distortion is rising more than the SPL:

Please excuse the sloppy presentation. Excel doesn't let me choose a log scale for anything other than a scatter plot.

Here are the usual absolute distortion levels:

Impedance is on the high side which is good:

This reduces the impact of long speaker runs.

Step response is quite odd although it may have something to do with baffle measurements:

We see a number of resonances in the waterfall:

Conclusions
Judging in absolutes, the treble response is quite poor both from on-axis and directivity point of view. Then again, distortion is quite low considering what we are dealing with. And price of course, is excellent.

Personally I am enough of a snob to spend more to get more. But if you are on a severe budget, the Polk RC80i may be a decent option.
------------
As always, questions, comments, recommendations, etc. are welcome.

Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

 

[amirm]

Reserved for @RickS to kindly post the specs.

 

[DearSX]

Not bad, gives me some idea about car coaxials

 

[AndreaT]

It seems a good speaker for AM radio broadcasts because they are typically limited to 4,500 Hz. Other than that, considering the time and money required for ceiling mounting, it is a pass. Thank you Amir for a great review of a flawed product.

 

[JSmith]

It was kindly drop shipped by a member and costs US $249 for a pair ($125 each).

Thanks to this member and for your testing Amir. Reasonably priced, but leaning towards the budget category here with the average measurements.

Some further pics for the thread anyway;

JSmith

 

[bachatero]

Is THIS why ceiling speakers setups so often sound funky unless you're in the sweet spot?

 

[kemmler3D]

Please excuse the sloppy presentation. Excel doesn't let me choose a log scale for anything other than a scatter plot

Unrelated to this review but I have found that Claude AI is pretty good for generating graphs with analysis/ presentation as described. I mention it because I normally use Excel/ G sheets for this kind of thing at work but the AI has been pretty effective to do things that take a long time in spreadsheet apps. It usually generates python code to create the graphs.

 

[bachatero]

It usually generates python code to create the graphs.

Matplotlib is just the best!

 

[amirm]

The table has 6000 rows. Do the above tools accept large data sets like this?

 

[bachatero]

The table has 6000 rows. Do the above tools accept large data sets like this?

Matplotlib works great with hundreds of thousands of datapoints so you should be taken care of there! I did something similar when I was putting together the ASR user statistic graphs last year.

 

[kemmler3D]

The table has 6000 rows. Do the above tools accept large data sets like this?

Yes, it should, that's why I ended up using it, I needed to do some semi manual processing on 1000+ rows.

 

[jomo]

Folks installing this home theater ceiling speaker would likely be well off axis and correspondingly use the tweeter's ball and socket to direct the tweeter toward the listening area as much as possible. It would be interesting to see if there is a sweet spot (angle) where the response is best. The previously tested KEF ceiling speaker had dozens of measured angles to get the "perfect" response angle. This poor Polk, not so much.

Also, home theater ceiling speakers will likely use a high pass filter which will reduce the bass distortion. It would be interesting to see the multi-tone distortion improvement by taking 80Hz and lower off the speaker.

 

[welwynnick]

Thanks Amir.
I know its cheap, but I couldn't help being disappointed.
Not terrible, but I do think Kef show how it should be done.

 

[amirm]

Reserved for @RickS to kindly post the specs.

View attachment 460368

Boy, a lot of fantasies there with stated low frequency of 35 Hz and 90 dB efficiency (actual: 85 dB)!

 

[amirm]

Not terrible, but I do think Kef show how it should be done.

Agreed. To wit, I bought my review sample of KEF and another for my Atmos ceiling effects.

 

[kyuu]

Boy, a lot of fantasies there with stated low frequency of 35 Hz and 90 dB efficiency (actual: 85 dB)!

Seems like they are basing the frequency response off the F10, which a lot of speaker companies do.

In any case, comparing this to the Revel C783 you reviewed previously, plainly Revel's configuration is much better for avoiding diffraction or whatever interference is causing those high frequency oscillations and poor off-axis response. Of course that one rather more expensive, but it's likely the lower range models are similarly good. My C363s sound great to me on surround duty.

 

[French Meloman]

Interesting and instructive exam. This ceiling speaker seems to have a good woofer, judging by the fairly correct on-axis response up to 4 kHz, and a frankly mediocre low-cost tweeter. The whole thing could only produce an unfinished and unadvisable product that operates outside of its specifications (efficiency < 5 dB of what is advertised).

 

[Robbo99999]

@amirm , could you go into some detail as to how the new graph you show is calculated and what it's showing? I've thought about it for a while but I don't quite understand what the graph is showing. I want to to be able to visualise how it's created and therefore what it means.

 

[amirm]

@amirm , could you go into some detail as to how the new graph you show is calculated and what it's showing? I've thought about it for a while but I don't quite understand what the graph is showing. I want to to be able to visualise how it's created and therefore what it means.

Sure. The math is simply the THD *percentage* of 86 dBSPL subtracted from that of 91 dBSPL. Since both are normalized (to their amplitude), the difference tells us how more distortion rose at 91 dBSPL vs expected. In other words, if distortion is linear, the difference would be zero. But if we have arrived at the non-linear portion of distortion graph, you get higher than zero value for the specific frequency. It is still work in progress but thought I share it.

 

[Jumu]

Sure. The math is simply the THD *percentage* of 86 dBSPL subtracted from that of 91 dBSPL. Since both are normalized (to their amplitude), the difference tells us how more distortion rose at 91 dBSPL vs expected. In other words, if distortion is linear, the difference would be zero. But if we have arrived at the non-linear portion of distortion graph, you get higher than zero value for the specific frequency. It is still work in progress but thought I share it.

It is a very good indicator. If posibble, please show 96-102dB also, since many likes listening loudly and it is important at home theater usage. Thank you.
Maybe you can also add some explanation notes under graphs.