Review and Measurements of Denon RCD-N9 (Ceol) Mini Stereo System

[dominikz]

Hello all,

New member alert! After several months of reading the forum, I thought it might be a good time to join and introduce myself with a review of my own - perhaps someone else will find the information useful. It also seems like a nice way to document my work and exchange thoughts with the many knowledgeable members here.

The device in question is the Denon RCD-N9 (Ceol) mini stereo system:

Back of the unit (with measurement equipment connected):

We've selected this mini system to power our living room system a few years back due to its small size, convenience and support of various inputs and formats including streaming, internet radio, AM/FM tuner, CD optical and analogue inputs etc. Basically it seemed like an all-in-one streamer, receiver, DAC and 2x65W (@4Ohm) amplifier solution. This was originally powering a set of cheap passive speakers and all was well

However a month or so ago we decided to go for a better set of speakers and decided on a set of Revel M16s (ASR definitely had a hand in our pick of the top contenders ). Turned up slightly I started noticing some looseness or muddiness in the ~100kHz area that I had not heard auditioning the speakers. Naturally, my first thought was that this was probably caused by a room mode, maybe even facilitated by the M16's bass hump. This also motivated me to do some basic measurements on the Denon.
Note: Latest version of Denon Ceol is RCD-N11DAB, but based on the info I found online it seems that the basic amplifier section hasn't changed much since at least the N7 version and N10 seems to use the same amp chip too, so some parts of this review may be relevant to other devices in the series.

One of the first things I did was to measure the frequency response with the amp loaded with resistive and complex (speaker) loads. The result is:

As you can see, the variation in response is >1dB, wide-Q, with peak and dips around the speaker's (Revel M16) impedance resonance frequency @~85Hz. For comparison, you can see sweeps with 4 and 8 Ohm pure resistive loads, and the soundcard baseline loopback response (which accounts for the low frequency roll-off).
Notice that the amp's high frequency response >2kHz is modulated by the load - likely caused by the output impedance/damping of the amp.

After some Internet investigation, it seems that this series of amps uses the TAS5142 digital amplifier (powered DAC). Looking at the chip evaluation version user's guide we can find the following:

This implies output impedance of ~0,667 Ohms @ 1kHz and is of course terrible. Same document shows high frequency behaviour in line to what was measured above:

Seeing this, I decided to measure the output impedance vs frequency to get the idea of the damping factor vs frequency. To measure the output impedance I measured the output of the amp unloaded and with a known resistive load, e.g. as described here. The result is:

As you can see, this is only slightly better to the info from the user guide, resulting in <2kHz output impedance of ~0,54 Ohm (but rising all the way to ~2,32 Ohm @ 20kHz!).
Note: Similar output impedance performance is independently measured here in a relatively similar device - Denon PMA30.

The following tests were done via Denon's optical input, with the device volume set to 45 (out of 60). Level was then controlled in SW with the REQ generator. This is because there is no analogue volume control and at 0dBFS output Denon starts to show signs of clipping with volume set around 37-40. I didn't bother to test the analogue input as it is digitized (I believe at 48kHz sample rate).

Moving on, I measured the SINAD @ 5W power output into 4 Ohms, one channel driven (as I have only one 4 Ohm dummy load ):

With one channel driven into 4 Ohm resistive load we get to about 65W output just before clipping onset, with 1% THD+N at around 70W output. As you can see, THD+N becomes distortion-driven already above ~0,2W output:

Equivalent diagram with SINAD:

Compared to the data in the amplifier chip datasheet and the user guide referenced before:

It seems Denon may have made some improvements in their implementation with regard to distortion performance - maybe due to use of lower voltage on the power supply and therefore slightly lower power output? RCD-N7 service manual seems to suggest 29V is used instead of 32V from the above datasheet example.

THD performance vs frequency @ 5W, sweeped sine:

I only did a basic power test with 8,2 Ohm resistive load, one and both channels driven so sadly no pretty graphs here :
- 39,6W @ 8,2Ohm: -62 SINAD, one channel driven (Note: this is also the point where the noise floor shoots up.)
- 42W @ 8,2Ohm: -41 SINAD, one channel driven
- 42,6W @ 8,2Ohm: -38 SINAD, one channel driven
- 39,6W @ 8,2Ohm: -57 SINAD, both channels driven

So we get about 40W of power into 8 Ohms, with the sum of distortion and noise getting worse by about 5 dB if we drive both channels:

I will not qualify overall device performance based on these measurements, but I will say that I definitely hope to test my speakers with an amp with higher damping factor - in hope to quantify influence of damping vs room response on my perception of the in-room bass response.

The measurements were done using the following basic equipment:
- REW for measurement and Google Sheets to create additional graphs
- RME Babyface silver edition sound card (best-case loopback SINAD ~100 dB)
- UNI-T UT60E true RMS digital multimeter
- DIY soundcard stepped attenuator/limiter based on a slightly modified Akitika schematic to allow higher output voltage range and additional connectivity
- DIY dummy load (1x 4,1 Ohm @ 100W or 2x 8,2 Ohm @ 50W)

Whoah, now this is a long first post Anyway, hope some will find it useful!

[EDIT 2022-05-24] Revisited measurements added to post #3.

 

[dominikz]

I will not qualify overall device performance based on these measurements, but I will say that I definitely hope to test my speakers with an amp with higher damping factor - in hope to quantify influence of damping vs room response on my perception of the in-room bass response.

Just wanted to follow this up after doing a lot more investigation.
Though I haven't had a chance to test with a different amplifier (yet), I did a lot of other tests since this post, e.g.:

• Created an EQ correction that fixes the amplifier FR variability introduced by speaker (M16) impedance and compared to no EQ
  • Difference was very subtle and IMHO completely dwarfed by room modes and loudspeaker response. The test was sighted so I can't rule-out confirmation bias.
• Ran and recorded a song directly on the speaker terminals to capture the effect of the amp/speaker FR variability but without room and speaker effects, and compared that to the same song with manually created (but equivalent) EQ curve and the unprocessed file
  • The difference was audible to me between unprocessed file and one with embedded amp FR variability, but pretty minor. Manual EQ correction IMHO makes the difference pretty much go away. The test was again sighted so I can't rule-out confirmation bias.
• Did a lot of testing with DRC / room EQ
  • This of course makes a huge positive difference and for me really puts things into perspective.

All in all, as many others before, I also decided that the FR variability caused by high output impedance of the power amplifier is a very marginal issue even in pretty severe cases such as this one. It is also relatively easily corrected by EQ, though one needs to measure the amp frequency response on speaker terminals to be able to correct for it - which I wouldn't recommend doing for most people, due to risk of injury and damage.
However, if one applies room EQ you anyway linearize the response at the listening position, so IMHO there's no need to fix the amp response specifically.

In short, I'm still using the N9, just added some room correction EQ up-stream (via miniDSP OpenDRC-DI) and I'm pretty happy with it.

That being said, if I were buying a new integrated amp of some sort, I'd give much more relevance to features (especially integrated room EQ), ergonomics/form-factor, price and power capability than to output impedance or even absolute SINAD (though all else equal, I'd still prefer to have better figures, of course).

 

[dominikz]

Almost 2y later, I thought it might be interesting to revisit these measurements (and add a few new ones) - now that I've gained much more measuring experience, and improved my process and instrumentation. So here we go (again)

The tools used for these measurements:

• RME Babyface Silver Edition as digital signal generator via optical (SPDIF) output
• Topping E50 DAC as analog signal generator (USB input, RCA outputs), 192kHz sample rate and 32bit depth
• E1DA Cosmos ADC (2,5mm TRRS Aux input), 192kHz sample rate and 32bit depth
• Asio4All drivers (48kHz and 192kHz sample rate, depending on test)
• REW software (by @JohnPM) for most measurements
• Multitone Analyzer tool (by @pkane) for multitone measurement
• Google Sheets for impedance calculations
• DIY resistive dummy load

All of the digital input measurements were done with the volume set to 45/60, as this gave me the most usable dynamic range for measurements (i.e. 0dBFS peak was close to max output power). All of the analog input measurements were done with the volume set to 51/60 (which resulted in approx. 29dB gain).

1kHz SINAD
Digital input, 48kHz sample rate:

Compared to previous measurement, this time I managed to get a slightly cleaner result with no power supply noise, by playing around with grounding - though as you can see it doesn't impact the SINAD end-result (since it is anyway distortion-dominated).
Note there is a hint of non-harmonic distortion around the 1kHz tone. Compared to the recently measured Sabaj A1 2022 there is less noise here (and a more uniform noise signature), but more distortion.

Here's the same thing with analog input with volume set for the typical 29dB gain (right channel only):

As we can see, distortion and noise both increase with analog input (likely due to AD conversion, though results might be better with less gain), significantly decreasing SINAD.

Multitone response

We get only 10,8 bits without noise and distortion; audio garbage (TM) is uniformly distributed across the audible spectrum

Frequency response and load (in)dependence

As we saw in post #1 already, there is severe load dependence with this amp. This is of course caused by the very high (and rising) output impedance / low damping factor:

The 0,54Ohm output impedance (at 1kHz) gives a damping factor of only 15 into an 8 Ohm load - which decreases further as frequency goes up.
Here's again the plot from the Sabaj A1 2022 measurement thread that shows what happens with the frequency response of this amp if you connect it to a real loudspeaker (Revel M16 in this case):

FR difference with digital input vs analog input:

We can see a few things here:

• We're at -0,8dB at 20kHz regardless of input type or sample rate
• There is some filter ringing with 48kHz sample rate and much less with 192kHz sample rate

If we zoom out a bit we can see that the analog input is probably internally digitized to 96kHz sample rate:

Power into 4Ohms (20kHz BW, one channel driven)
There is almost 70W power into 4Ohm before clipping onset, slightly more if we allow 1% THD:

As a bonus, here's the left channel with analog input:

As expected - higher noise (almost +28dB!?) and distortion but power is the same.

Power into 8Ohms (20kHz BW, one channel driven)
We get almost 40W into 8Ohm before clipping onset, slightly more if we allow 1% THD:

Power vs distortion into 4Ohms (45kHz BW, one channel driven)

The wideband distortion vs level plot is surprisingly tidy, with relatively similar performance regardless of signal frequency!

Conclusion (take #2)
Looking at objective amplifier performance, this is far from stellar (and e.g. output impedance is downright terrible). However I was pleasantly surprised to see a few things done right - like the uniform wideband distortion vs level plots and use of high sample rate on the analog input. Given there's quite a rich feature set, and if power output is enough, I'd say you could go worse than this for a compact AIO stereo

 

[Meteli]

Hi all and first post.
I write from Finland and for years here have being floating these Denon Ceol players. Nothing interesting there, but many of them where sold as "PREOUT" -version with Genelec speakers that are so common here, you can pickup pair as you do food shopping. Only problem is; Genelec speakers are active and many use them beside television, this being before soundbars where more common, but there really where any suitable cheap preamplifier for those untill Denon made these "PREOUT" version that do have all parts in them, but amplifier section is disabled. It hapend I found one from recycle bin and thought I would give a try.

CD was not working. It worked 30seconds to an minute, but after that there was no amperage going in. So as I was watching TV in livingroom I took it apart and pulled out CD mechanism.

It turned out to be usual and since I had found one of those cheap plastic players sie of the cabin road (recycle bin) I took mechanism from that.

Hook it up with original CD loader, littlenewspaper underneath not causing short and it looked good!
These CEOL players got hidden menu's for checking CD mechanism and other, you can save new parameters in, very clever idea to let machine itself tell repairer how thing are working.

 

[Meteli]

As some of you noteice; cheap CD/radio -players are top loaders like those fancy high end CD players I being told, so I took this replacement CD mechanism and gave it a little grind with bench grinder, so CD will lock between spindel and magnet and actually loads it.

Also I added washers under screws to support the springs that cancel vibrations as original springs have died. Cheaper rubber busings are better after all...

Hey and it works!

 

[Meteli]

Now to the interesting part:
As I was there I took pictures how Denon made these "PREOUT" versions.
They simply used jumber wire from preamp and soldered them to "Analog In" connectors.

Here you can see where they soldered:

Other end:
Subwoofer line out is bit crude, but does it's job:

They also cut out that red wire, I have not look what it is and did not open maind board to see what's done on other side, but this is interesting way to bypass Denon amplifier and use these only as pre amplifiers. I use this in livingroom where guests can use NFC to hook up their phone and play musc if they will also these little boxes accept CD's, USB sticks (allthough not wery fast) got radio and some network things.

 

[Meteli]

I am not sure how they disable amplifier from working, since speaker terminals are dead on these. If there is more interest I can open this little box furter to see if it is something done to board itself and not done with software.