Warning for Denon owners: your xx09/x89 AVR may have a major bug

And the the bug is that Audyssey MultEQ may be completely broken! So the feature that should help your sound the most could be ruining it. You could be affected by this and not even realize it, as I was, until I looked into it further.

Cause/Solution Found! (Sep. 30, 2009 Update):

After sending my [second] AVR-1909 to Chris from Audyssey for testing, he confirmed the primary problem of lossy DD/DTS bitstreams being wrongly decoded when MultEQ is enabled. After sending his findings to Denon engineering in Japan, they found it to be a firmware bug in early units! It was then inadvertently fixed along with an update for an iPod Video problem, but units manufactured as late as Nov. '08 could be affected.

Models affected: 1609/589, 1709/689, 1909/789, and 2309/889. The 2809/989 is not affected since it uses MultEQ XT.

For the 2309 only (not 889!): Serial numbers xxxxx15241 and lower were manufactured with the buggy firmware. If you haven't already installed a newer firmware that resolves the issue (e.g. iPod Video fix), please see jdsmoothie's firmware updates post and contact him to install the update yourself. Again, this does not include the 889! Only the 2309 has the RS-232 jack required to install firmware updates.

For all other models, you will have to contact Denon Customer Service and either take your AVR to a Service Center or create an SRA Request to send your AVR to Denon to have the firmware updated.

Check the firmware to quickly determine if yours has the bug (versions verified on the 1909 and 2309 so far). To view firmware information:

1) With the AVR in Standby (red ring), power off completely with the small ON/OFF button
2) Press and hold the DYN VOL and VIDEO SELECT buttons on the front panel
3) Power it back on with the small ON/OFF button; it goes into Standby (red ring)
4) Release the DYN VOL and VIDEO SELECT buttons, then turn it on with the large On/Standby button (green ring)
5) Once it's powered up, press STATUS to cycle through to the firmware versions
***NOTE*** STATUS will continue to function this way until you cycle the power OFF and ON with the small button (Standby doesn't cancel this mode)

Fixed firmware versions are as follows (lower/older have the bug):
For 1909 and 2309 (probably 789 and 889 also): [Main: 00.90, DSP1: 45.43]
For 1609/589 and 1709/689: To be determined/verified, please report!


To summarize: The main problem is incorrect decoding of lossy DD/DTS bitstreams when MultEQ is enabled, resulting in overblown bass. Some other issues I suspect are related, because they didn't exist on the AVR-590 I tested, and I'm hoping they're resolved with the new firmware: 1) Incorrectly calibrated speaker trims and/or master volume setting, affecting the operation of Dynamic EQ/Volume; and 2) The overall level of bitstreamed Dolby TrueHD is about 2-3dB lower than player-decoded PCM. (I was not able to compare DTS-HD MA.)


Revised original post and information:

For those who haven't seen my posts in the Audyssey thread or the 1909 thread, I finally discovered last month that my problem of inconsistent and wrong audio was solved when I happened to turn MultEQ Off. I sent my unit to Denon to be fixed and they sent back a refurb that has the exact same bug, so I'm now confident that others out there are affected as well, since I've had two units that are (and I may have to fight Denon to do something about it Update: Denon has said they can replace it, with a new unit this time).

On Sep. 4th I picked up a new model AVR-590 from Best Buy (temporarily ) to check it and absolutely rule out anything specific to me (which I already had; I know what I'm doing). And yes, although I wasn't exactly sure what I'd find, MultEQ works perfectly as it should, unlike the two 1909s.


The primary problem is with MultEQ enabled (Audyssey curve, Flat, or Bypass L/R), bitstreamed Dolby Digital and DTS signals decoded by the AVR sound much different than the same soundtrack decoded externally by a player and sent to the AVR as PCM, primarily in the bass from the subwoofer (not sure about other speakers): nowhere near accurate, boomy, overblown, bombastic, etc. It's not a subtle difference at all, especially with Dynamic EQ On (anyone could notice the difference then!).

How to check for the bug:

Lossy listening tests. The simplest is to just use a few movie scenes that you're familiar with and have a good amount of bass (preferably action, explosions), or anything else you think has enough bass to hear.

- Make sure MultEQ is enabled, of course, by the indicator.
- It is strongly preferred to turn Dynamic EQ On (Red light, not Green) to magnify any differences, making it very easy to tell if your AVR has the bug.

If using HDMI audio and a player that can output multi-channel PCM, remember:
- Cannot be used with 1609/589 or 1709/689 (no HDMI audio support).
- You must use a lossy Dolby Digital or DTS soundtrack; not PCM (on the disc), Dolby TrueHD, or DTS-HD.
- Your player must be able to internally decode that soundtrack to PCM; some will only bitstream DVD/lossy audio.

If using non-HDMI audio (only option with 1609/589 and 1709/689) and/or a source that can output 2-channel PCM, remember:
- You cannot use a multi-channel soundtrack (e.g. 5.1), because changing to PCM results in a 2-channel downmix, which will sound "weaker" -- not because of this bug, but because channels are no longer discrete, [some] automatic DRC, loss of LFE channel, etc.
- You must use a Dolby Digital 2.0 track to compare bitstream and 2-channel PCM. (DD 2.0 audio could also come from a HD DVR/STB, for example, instead of a disc, as long as the box can be changed to output PCM instead of the Dolby Digital bitstream.)


It doesn't matter, but I'd recommend first having your player decode the soundtrack and send PCM. Listen to some of your reference material (do it a couple times to get a feel for the sound). Then, switch the player to output bitstream, and don't touch anything else, including the volume. Repeat for each of the few things you're checking -- since bass will be at different frequencies, the bug may be more or less obvious depending on the frequency.

Did it sound exactly the same both ways? Then you're fine, since the bug should have been obvious. If things sounded different, it doesn't mean you have the bug, since differences between player and AVR decoding could result in a different sound. So first thing to do is set MultEQ to Off and listen again. Any difference wouldn't be as obvious now with Dynamic EQ subsequently Off, so listen a little more carefully. If bitstream and PCM now sound exactly the same with MultEQ Off, you have the bug! (If both ways still sound different, check and turn off things like Dynamic Range Compression, Night Mode, etc. in the player and AVR, and check speaker levels for each signal format on the AVR, until you get each to sound the same with MultEQ Off. Then start over with MultEQ and Dynamic EQ On. )


If you are not able to do a comparison with PCM, you can try some subjective tests (with lossy DD/DTS formats):
- With Dynamic EQ On, it should be fairly easy to tell if bass just "seems" overblown.
- You could try comparing MultEQ enabled vs Off and listen for relatively "more bass" overall with MultEQ (but not Dynamic EQ!). Keep in mind this may not be very reliable, since MultEQ is supposed to change the frequency response, which could be all you're hearing...


If you could please check your AVR and report the results (with model and firmware versions), I'd appreciate it!

Like I said, you should be able to check with "real material" and determine just by ear if you have the bug, but for something more scientific, download and burn the DIY Audio Test DVD. There you'll find subwoofer noise similar to the AVR's internal test tone (5.1 DD Tests > Calibration > Levels Check > LFE), which is a steady sound and easily revealed the bug for me. Compare bitstream and PCM using the same "rules" as above (but Dynamic EQ isn't "needed" for these). Or you could use some of the Manual Tones for LFE at different frequencies, to compare by ear or with an SPL meter.

With non-HDMI audio (only option with 1609/589 and 1709/689): You cannot use the "subwoofer noise" since it's in the LFE channel. You can use the Manual Tones for the Left, Center, or Right channels. Check a spread of a few frequencies, and to measure the subwoofer, make sure those frequencies are "far enough" below the crossover for the speaker corresponding to the tone you're using. The correct result is for PCM to measure 3dB lower than bitstream as a result of 5.1 to 2-channel downmixing.

I used that LFE Calibration tone with the AVR's subwoofer pre-out connected to the computer and used Room EQ Wizard's Spectrum analyzer (1/12 octave RTA, 16 Averages) to show a visual representation of the output:

MultEQ Off: PCM vs Bitstream
- Both 1909s and the new AVR-590 sample look like this


MultEQ On: PCM vs Bitstream (Plus the 2 MEQ Off measurements averaged and smoothed)
- Both 1909s look like this, with bitstream completely screwed up


MultEQ On: PCM vs Bitstream
- New AVR-590 sample looks like this, perfect; bitstream and PCM are identical as it should be

Thanks for your reports!

@tingham - Yeah, I'm sure it could easily go unnoticed. Like I said, your 2809/989 model may be different enough with MultEQ XT that it may never see this issue!

I only got this one unit from BB (newer 590) and it's fine. The only "so many" units it's happening on are the two 1909s: one of the first pre-ordered from J&R last Aug. and the refurb replacement Denon sent. Sure, will keep the forum updated...


@zzoli - So it sounds like your 1909 has the bug too?! To be clear, that "night and day difference" (like I hear) was after you adjusted the LFE setting to be equal (0) for both input types? And you have the PS3's Dynamic Range Control turned Off so the PCM output isn't "tamed?" (Sounds like you're fine there from the MultEQ Off result. )

Regarding just setting the PS3 to PCM and sleeping well... I'm not sure, with mine, that everything is exactly as it should be with PCM either, though it's pretty close compared to bitstream. However, don't you have any other sources where you need the AVR to decode a bitstream, such as Dolby Digital from HDTV broadcasts, etc.? Those would also be affected by this problem!

Thanks for the confirmation, as well as your other replies here. Yep, "night and day difference!"

Right, I understand having set PCM LFE to -10 for SACDs, since from what I understand, they don't have the LFE track recorded 10dB low, thus don't need the 10dB boost like other LFE content... Or something.


I said I wasn't sure everything was as it should be with PCM either... If you saw my post in the Audyssey thread after comparing it to the AVR-590 that works right, I'm sure there's also a problem with PCM. Or specifically, with mine, the 1909's channel levels set during auto setup are a lot higher than the 590, and the 1909s are louder (nearly 10dB) at a given volume setting. They should be equal since calibration is supposed to be for reference level (needed for Dynamic EQ to function properly). As a result of the lower volume setting on the 1909s, more "boost" is being applied, causing even PCM's bass to sound "overblown" with Dynamic EQ! Compared to the 590: higher volume setting for same output, less "boost." I was shocked to hear the difference compared to the 590 (whose output sounds balanced and "correct" to me), after thinking PCM sounded "weak" on the 1909 -- only when compared to the bitstream insanity!

With PCM and Dynamic EQ, do you think your volume setting seems "too low" when it's about as loud as you'd want to go?

Sure, see rmm2112's posts. Just don't use a DTS-HD MA soundtrack since it looks like the BDP-S350 can't decode it to PCM.


Proven to work...? Or proven to have the bug?


Sounds like you did it right and probably aren't affected. When I first mentioned MultEQ being the cause in the Audyssey thread, batpig (who also has a 789) said he didn't hear a difference. He should post here and reconfirm.

No reports about a 1609 yet. Since it doesn't accept audio over HDMI, maybe things are different enough that they wouldn't be affected by this bug.

I have updated the first post with directions for checking with non-HDMI audio. Basically if you're doing a listening test by ear, you need to use a DD 2.0 soundtrack to avoid hearing downmix differences when outputting a 5.1 track as 2-channel PCM.

Or if measuring the L/C/R Manual Tones from the DIY Audio Test DVD, expect a 3dB reduction when that 5.1 track is downmixed to PCM if things are working right.

As zzoli said, there's no overall level difference, etc. Of course setting the LFE level to -10 for whatever bitstream will tame the bass that's coming from LFE, if any. Turning down the subwoofer would help even more since the bitstream/PCM differences are the same whether the content comes from the LFE channel, or redirected from the mains.

No matter what, the [bitstream] result would be unbalanced and wouldn't match the correction curve that MultEQ is trying to apply (the PCM result is close, if not exactly correct). But everything is fine with MultEQ Off!


It's still a bug in the AVR no matter what's actually causing it.


Not possible. (I understand about the simple cause.) There isn't a double boost or anything. If you wanted to make those curves look that different (user settings-wise, I mean), you couldn't possibly accomplish it. That's why it rules out any sort of user issue, at least...

And it's not related to redirected bass content from the mains getting boosted with the LFE, since that graph came from pure LFE. I've also played individual bass tones from main channels (on that disc), and the bitstream/PCM differences are identical, except everything's 10dB lower without the LFE boost, so nothing is wrongly being boosted there.

I believe that Audyssey/MultEQ can do a much better job than anything manual. Also that most of the time if the result is disliked, it's probably user-related (except for this bug, of course).

Ignoring that irrelevant part, you completely missed the point anyway. As zzoli said: "The Audyssey implementation should be input signal agnostic." MultEQ is applied as one of the last things, before DAC, I guess, and has nothing to do with the input signal, since everything's PCM... when... it gets to... MultEQ.

Wait, or is it? Didn't I read that lossy formats are decoded in the "floating-point" domain in the DSP instead of actual PCM (integer based)? Hmm, not sure now... just came to mind, but wonder if it's related. Now that makes me wonder if bitstreamed lossless would have this bug! I assumed it would 'cause "bitstream," but lossless would be converted to PCM rather than a floating-point representation (if that's what happens with lossy)...

Well, back to the quoted post -- it also couldn't be caused by the way Audyssey setup was done. You could do everything wrong with mic placement, etc. and then some, which could totally screw up the sound for your room. However, no matter what, the output should/would be exactly the same with bitstream or PCM.

Precisely! It was my observation that the 2 signal plots looked like flipped/mirror images of each other.

Interesting, thanks. That seems pretty similar to what I'm seeing with both 1909s. My speakers were also set to +8 or a bit more (less on surrounds). But the 590 set them 4-5dB lower, and the sub 10dB lower!

Like I said, this allowed me to set the 590's volume higher to get the same output level. On the 1909s, I don't think I'd want to go to -20 (certainly not with bitstream and Dynamic EQ! ). -30 was pretty loud, and not sure if I would have listened to a movie higher than that (haven't done any serious movie "watching" since getting the new AVR last year, LOL, just TV!). With the 590 though, I had The Matrix at -20 and it seemed like a fine, "normal" level.

I feel the volume is miscalibrated with my 1909s after seeing the 590, and it seems yours is the same, which if true, would result in too much Dynamic EQ boost and slightly overblown bass, even with PCM. I found the 590 with Dynamic EQ to be noticeably "easier" on the ears -- less harsh/fatiguing sound.

Thanks for posting the updates jd! Did you just get this additional info. from Denon about the manufacturing date and that it happened to be fixed with the iPod problem...? Sounds like it was just an "accidental" fix then? Don't understand how testing couldn't catch such a fundamental flaw with the audio processing. It's not like some obscure configuration is needed to trigger it!

Wow, Nov '08 is much later than I expected when Chris from Audyssey said "early models." I thought it'd be just a few weeks worth at most. But man, tons and tons of people bought them before Nov, and then however long the problem units were around still in the retail channels after that!

I'll update my first post with the latest information in a bit.