Lexicon DD8 Review (Multichannel Amplifier)

[Francis Vaughan]

I agree, but it also sets a precedent of not testing THD at the limit of the generally accepted range of human hearing

OTOH the generally accepted limit of human hearing is lower than the lowest possible harmonic in the THD measurement. So it isn’t clear it has any meaning. Overall I don’t think THD specs at these frequencies have much bearing on actual audio quality at all. They can be a marker for certain types of bad behaviour in some implementations of amplifiers, but they are not universal in application. Especially not in class D.
Metrics based on transfer function can be mapped to the human ear and audibility. Eventually we might get there.
IMD is more interesting as it shows up bad behaviour that THD won’t see and that may well be audible.

 

[restorer-john]

OTOH the generally accepted limit of human hearing is lower than the lowest possible harmonic in the THD measurement. So it isn’t clear it has any meaning.

I also agree. But one has to be careful allowing a downward trend of "frequency creep" or "lowering the ceiling" simply to appease a technology that plays by different rules. The same attention to detail and minutiae which @amirm has applied to D/A converters, has clearly paid off in spades with products being supplied for reviews improving at an exponential rate. It would be a shame to lose that momentum.

It's a tough one for sure, but my position is harmonics or intermodulation artifacts should be retained, up to and including the top end of the agreed human limits of hearing, regardless of any noises from the vested interests, or their associated fanboi peanut galleries.

Absolutely agree with IMD as more illuminating in many respects, but again, the old tests which used mains frequency (60 or 50 for us) and 7-8kHz at 4:1 were phenomenally useful, but less so now in the world of HF SMPS supplies. 19/20kHz with nothing above and only focusing on the audible range shows us nothing about potential interactions or instability outside the audible band. Is Class D these days as well behaved up there as Class AB wasn't?

 

[pma]

The test is automated so doesn't take any effort. I just think it is confusing when it always shows less distortion at 20 khz than 15 kHz.

This is because of amp's limited bandwidth to some 50kHz (an also maybe because of the AP additional filter used?). The output LC filter reduces the harmonics above 50kHz, which would otherwise spread somewhere to infinity. So the 20kHz has seemingly less distortion.

This is the unfiltered 20kHz spectrum of the A07, of course measured with lower resolution, however we can see most distortion components up to 60kHz and also the carrier frequency of 600kHz and intermodulations at 560kHz and 640kHz. These amps are beasts .

 

[Rebelhifi]

Thanks Amir. I changed out my surround, surround back, and atmos channels from Crown XLS 2000/1502 units to the Crown CT8150. I purchased the Crown CT8150 (used - 2 year old conference room application for $600). I have Hypex DIY NC400 mono for the front 3 channels that are dead quiet with my Klipsch Reference Premier (need to put my ear to the tweeter). The Crown XLS were too noisy (hiss snakes) even with XLS connections (or with no cables connected at all) with the Klipsch surrounds (they were worse on the main speakers). I could hear the hiss at the listening position. The CT8150 is (as measured by Amir) much, much quieter. I wanted to go with Hypex all around (maybe in the future) but any 8 channel option are all +$2000 even at the lower power levels (75 watts). I am currently enjoying the fully black background with zero noise with movies. When things go quiet and they director wants to startle you, the CT8150 sounds more dynamic to me as there is zero (hiss) noise and scare the S%$& out of you. (For movies, the limited bandwidth is not a issue as Audyssey Calibration attenuates the signal). I am pleased with the unit as it is an upgrade from my current situation. Based on the specs, the XLR based CT8150 have a better S/N over the RCA based (Lexicon) by 5 db. This is the hobby! Next, start to replace the Crown with more powerful Hypex (step by step)! LOL.

 

[AndreaT]

Poor performance given the asking price. Should be back to the drawing board.

 

[bigguyca]

Speaking of this, I am thinking of eliminating the 20 kHz test. It screws up the visual display if one doesn't take the bandwidth into account. If there are no objections, I will do that in the next review.

Here is an objection

Dumbing down the reported results of your measurements seems a very bad idea. The core value of this site is high-quality, detailed measurements. Measurements are what make this site unique.

People who understand how the measurements are done and the implications of the results also understand why the results look as they do. This is also the case for measurements presented in Stereophile that "suffer" from the same appearance problem.

Based on the posts on this site it seems likely that most visitors don't review the measurements in detail or actually understand them. That's fine, there is something for everyone here. Dumbing-down the reported results won't cause most visitors to get more information from the measurements. Dumbing down will reduce the value of your measurements for those who do understand them.

 

[pma]

Well put, @bigguyca

 

[MrPeabody]

I don't have an opinion on the question but I had a thought that I thought I'd share, partly because I might be misunderstanding something, in which case someone will enlighten me.

My understanding is that harmonic distortion (or harmonic distortion + noise) when measured at some specific frequency is obtained by playing a tone at that frequency and then analyzing the spectral content (using FFT?) to separate the acoustic energy at that frequency from the total acoustic energy.

For the present discussion, it seems pertinent to point out (without taking a stand on anything!) that if the measured harmonic distortion includes subharmonics, i.e., frequencies below the test tone frequency, then the distortion may be audible even if the test tone is 20 kHz. (Of course if this shows up in the measurements this means that it exists, in which case it would matter even if for some reason it was not apparent in the measurement.)

In fact, the same thing seems likely true even if the test tone is higher in frequency than 20 kHz, unless spectral content above 20 kHz is filtered out before there is any opportunity for the amplifier to add subharmonic distortion. (If spectral content above 20 kHz is not filtered out before the amplifier has the opportunity to add subharmonic distortion, then it would seem desirable to measure distortion for frequency much greater than 20 kHz.)

The second graph I see is the spectral makeup of the output when a 1 kHz tone is used. Unless I am misinterpreting this graph, it suggests to me that in general any pure tone will be adulterated with subharmonics as well as the regular harmonics.

 

[ozric]

I set the gain to max which nicely gave me 29 dB

Any guess as to what the gain is at the center detent? That is how I have the gain set on my DD8, I usually have my preamp up to max volume for some of my quieter sources. I had forgotten there was a gain control on the back

Will have to get back there and turn them up.

 

[voodooless]

Based on the posts on this site it seems likely that most visitors don't review the measurements in detail or actually understand them. That's fine, there is something for everyone here. Dumbing-down the reported results won't cause most visitors to get more information from the measurements. Dumbing down will reduce the value of your measurements for those who do understand them.

I tend to agree with you there. There obviously is already a dumbed down metric called SINAD. If that would not be enough one could devise a number of parameters and score those with starts to get to a more comprehensive overview. I would still like de detailed measurements though.

 

[NTK]

...
The second graph I see is the spectral makeup of the output when a 1 kHz tone is used. Unless I am misinterpreting this graph, it suggests to me that in general any pure tone will be adulterated with subharmonics as well as the regular harmonics.

Looks like those are power supply noise also seen in a few other speaker amps.
Examples:
https://www.audiosciencereview.com/...ulus-286-amp.6443/#lg=attachment20844&slide=0
https://www.audiosciencereview.com/...fier-review.13960/#lg=attachment68042&slide=0

 

[amirm]

Any guess as to what the gain is at the center detent? That is how I have the gain set on my DD8, I usually have my preamp up to max volume for some of my quieter sources. I had forgotten there was a gain control on the back

Will have to get back there and turn them up.

I don't but I would guess down to 20 dB or so. You only need to turn them up if you are running out of volume control range before the amp runs out of power.

 

[amirm]

The second graph I see is the spectral makeup of the output when a 1 kHz tone is used. Unless I am misinterpreting this graph, it suggests to me that in general any pure tone will be adulterated with subharmonics as well as the regular harmonics.

Most of those are power supply noise. You can tell this when they are at multiples of 60 Hz mains frequency which is the case here.

There are a pair of small spikes on either side of the main tone which indicate frequency modulation error (similar to jitter). So those you could consider as distortion. The mains stuff is considered noise. Many amplifiers have such noise components.

 

[amirm]

My understanding is that harmonic distortion (or harmonic distortion + noise) when measured at some specific frequency is obtained by playing a tone at that frequency and then analyzing the spectral content (using FFT?) to separate the acoustic energy at that frequency from the total acoustic energy.

FFT is not used here. But rather, a steep notch filter is used to filter out the input tone. All that remains then is summed and expressed as THD+N. So your general point is correct. I just wanted to clarify the method used here. FFT based scheme is used to exclude noise and express THD alone but that capability is not used in this test.

 

[MrPeabody]

Looks like those are power supply noise also seen in a few other speaker amps.
Examples:
https://www.audiosciencereview.com/...ulus-286-amp.6443/#lg=attachment20844&slide=0
https://www.audiosciencereview.com/...fier-review.13960/#lg=attachment68042&slide=0

Thanks for pointing this out. I looked at it again and it is obvious that the spikes below 1 kHz are harmonics of 60 Hz. I should have known this. I wasn't thinking.

 

[MrPeabody]

FFT is not used here. But rather, a steep notch filter is used to filter out the input tone. All that remains then is summed and expressed as THD+N. So your general point is correct. I just wanted to clarify the method used here. FFT based scheme is used to exclude noise and express THD alone but that capability is not used in this test.

I appreciate that clarification, and it makes perfect sense because if the test is only concerned with determining what percentage of the acoustic output is the signal and what percentage isn't the signal, this is all that is needed.

I should have realized it was power supply artifacts, indicating the presence of 60 Hz ripple and harmonics thereof. Although, I wonder why the harmonics at 180 Hz and higher are stronger than 120 Hz and even 60 Hz itself. It seems to suggest that the power supply filter has stronger effect at 60 Hz and 120 Hz than at 180 Hz and higher.

I'm now wondering whether it even makes sense to measure distortion for any frequency above 10 kHz. If the spectral makeup of the distortion is strictly above 20 hKz, I don't see why it would matter. Except to dogs.

 

[Francis Vaughan]

Part of the problem is likely that we keep talking about harmonic distortion. What is presented isn’t really harmonic distortion in the strictest sense. It is residual noise after the stimulus has been removed where the stimulus is a single frequency. Not quite the same thing. Ignoring stochastic noise there is still other noise that is not correlated with the stimulus and not distortion, such as mains hum.
The next question is how much of the stimulus correlated residual is harmonically related to the stimulus, and how much is correlated via some other basis.
When we discuss conventional continuous transfer function amplifier designs (aka analog, class A, AB, B) the transfer function is constant for a continuous stimulus and the transfer function can be usefully described by harmonic analysis. So Harmonic Distortion makes sense.
If however the amplifier is more interesting in design it may not present a constant transfer function. This is especially possible if the amplifier is non-continuous and samples. Like a Class-D amplifier. Intermodulation is always a possibility. Simple heterodyning of stimulus and switching energy may make for a serious mess. (The self oscillation of the Hypex designs is the secret sauce that addresses this and makes these amps such a step up. )
But we now don’t actually have a single transfer function that is amenable to harmonic analysis. Harmonic Distortion in its strict meaning does not cover all the test results.
In all likelihood some of the junk that looks like the noise floor is stimulus correlated energy.
THD is not a complete metric of distortion. SINAD covers more than harmonic distortion and stochastic noise. THD-N is a bit ambiguous. Which is minimally why IMD is always useful. Again, human hearing models applied to the results would be interesting.

 

[hyfynut]

Guys, as an 8ch single chassis the amp looks very good and compact, however overpriced.

What do you see out there that get's you 8 channels over 100w in one chassis? Maybe a multichannel home theater amp? Actual curiousity here, no snark.

 

[hyfynut]

This is a review and detailed measurements of the Lexicon, first generation DD8, 8-channel amplifier. It was purchased used by a member and drop shipped to me. The DD8 came from collaboration of Harman with TI (chip company) to produce very low weight, switching amplifiers for BMW car audio system. It was then later repurposed in this configuration for the Custom Integration (CI) industry. It is actually rebranded by a number of companies and sold under their label.

Note: our company, Madrona Digital, is a dealer for Harman products which includes Lexicon. So keep potential bias in mind as you read this review.

All 8 channels are provided in one compact, very cool running 1U rack mount chassis:

View attachment 107448

A stout on off switch manages power duties together with heftiest relay sound you have heard! The back will seem unusual for hi-fi market but is precisely what the custom integration channel needs:

View attachment 107449

You can either feed the unit individually per channel or a "bus" input that feeds the same stereo input to all the channels. The latter is very handy for "whole house audio" where the same music plays in every room from ceiling speakers and such. A looped bus output is provided to cascade more of these units.

Output is provided through phoenix connectors which again is common and desired in CI channel as they are secure but small connectors. Individual gain controls are provided for every channel. You can also select mono vs stereo (again, useful in whole house audio where you usually want mono). And auto-sensing for channels not used. Trigger is also provided of course.

The available units are pulled out of existing installs so likely have many hours on them. And if the installer didn't know what he was doing, it may have been subjected to high temps. Indeed the first unit the owner bought did not work right, changing its gain with temperature. I traced it to the front-end of the unit. So know what you are getting if you buy them used.

Brand new, second generation DD8s cost US $3,500 (through dealer channels). This version listed for US $2,500. I see them listed for $1,300 right now on ebay but I believe the owner was able to get it much cheaper.

Lexicon DD8 Measurements
As usual, we start with our dashboard. I set the gain to max which nicely gave me 29 dB which is what I have standardized on:

View attachment 107450

This is above average performance with distortion kept below 100 dB;

View attachment 107451

Noise performance is good as well:
View attachment 107452

Ideally we would get 16 bits at 5 watts but 15 is a good effort.

I was disappointed by the early cut off from the output filter:
View attachment 107454

This is due to rather low switching frequency of 250 kHz (common values are in 300 to 400 kHz), which made the filter harder to design:

View attachment 107455

Crosstalk is fine:
View attachment 107457

Power into 4 ohm is rather low:
View attachment 107458

The protection circuit was aggressive which made getting max and peak power difficult. Still I managed to get some numbers:

View attachment 107460

While we get more burst power courtesy of power supply designed for a lot more channels than two, max power with 1% THD+N still was rather low.

I then tested the amp using 8 ohm and surprisingly, it produces more power there:

View attachment 107462

To see how the power scales, I measured max power using two and four channels. Both were the same using 8 ohm:
View attachment 107463

Most disappointing was varying the frequency and producing a power sweep:

View attachment 107465

As you see, there is huge penalty at higher frequencies. Furthermore, there are power related linearities.

Conclusions
In mid to low frequencies, the DD8 produces well above average performance. Above that frequency though, distortion increases substantially and other non-linearities surface. Output seems to have some load dependency while also rolling off early.

If you can get the unit for a few hundred dollars, then its compact enclosure and general performance may be a steal. You could almost push me to recommend it but I am not going to. I like to see better performance with respect to frequency response and high frequencies. I think for its intended custom install market, it is perfect (our company has installed a lot of these). For hi-fi use, you have to make that decision.

------------
As always, questions, comments, recommendations, etc. are welcome.

Appreciate any donations using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

This is a review and detailed measurements of the Lexicon, first generation DD8, 8-channel amplifier. It was purchased used by a member and drop shipped to me. The DD8 came from collaboration of Harman with TI (chip company) to produce very low weight, switching amplifiers for BMW car audio system. It was then later repurposed in this configuration for the Custom Integration (CI) industry. It is actually rebranded by a number of companies and sold under their label.

Note: our company, Madrona Digital, is a dealer for Harman products which includes Lexicon. So keep potential bias in mind as you read this review.

All 8 channels are provided in one compact, very cool running 1U rack mount chassis:

View attachment 107448

A stout on off switch manages power duties together with heftiest relay sound you have heard! The back will seem unusual for hi-fi market but is precisely what the custom integration channel needs:

View attachment 107449

You can either feed the unit individually per channel or a "bus" input that feeds the same stereo input to all the channels. The latter is very handy for "whole house audio" where the same music plays in every room from ceiling speakers and such. A looped bus output is provided to cascade more of these units.

Output is provided through phoenix connectors which again is common and desired in CI channel as they are secure but small connectors. Individual gain controls are provided for every channel. You can also select mono vs stereo (again, useful in whole house audio where you usually want mono). And auto-sensing for channels not used. Trigger is also provided of course.

The available units are pulled out of existing installs so likely have many hours on them. And if the installer didn't know what he was doing, it may have been subjected to high temps. Indeed the first unit the owner bought did not work right, changing its gain with temperature. I traced it to the front-end of the unit. So know what you are getting if you buy them used.

Brand new, second generation DD8s cost US $3,500 (through dealer channels). This version listed for US $2,500. I see them listed for $1,300 right now on ebay but I believe the owner was able to get it much cheaper.

Lexicon DD8 Measurements
As usual, we start with our dashboard. I set the gain to max which nicely gave me 29 dB which is what I have standardized on:

View attachment 107450

This is above average performance with distortion kept below 100 dB;

View attachment 107451

Noise performance is good as well:
View attachment 107452

Ideally we would get 16 bits at 5 watts but 15 is a good effort.

I was disappointed by the early cut off from the output filter:
View attachment 107454

This is due to rather low switching frequency of 250 kHz (common values are in 300 to 400 kHz), which made the filter harder to design:

View attachment 107455

Crosstalk is fine:
View attachment 107457

Power into 4 ohm is rather low:
View attachment 107458

The protection circuit was aggressive which made getting max and peak power difficult. Still I managed to get some numbers:

View attachment 107460

While we get more burst power courtesy of power supply designed for a lot more channels than two, max power with 1% THD+N still was rather low.

I then tested the amp using 8 ohm and surprisingly, it produces more power there:

View attachment 107462

To see how the power scales, I measured max power using two and four channels. Both were the same using 8 ohm:
View attachment 107463

Most disappointing was varying the frequency and producing a power sweep:

View attachment 107465

As you see, there is huge penalty at higher frequencies. Furthermore, there are power related linearities.

Conclusions
In mid to low frequencies, the DD8 produces well above average performance. Above that frequency though, distortion increases substantially and other non-linearities surface. Output seems to have some load dependency while also rolling off early.

If you can get the unit for a few hundred dollars, then its compact enclosure and general performance may be a steal. You could almost push me to recommend it but I am not going to. I like to see better performance with respect to frequency response and high frequencies. I think for its intended custom install market, it is perfect (our company has installed a lot of these). For hi-fi use, you have to make that decision.

------------
As always, questions, comments, recommendations, etc. are welcome.

Appreciate any donations using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

I missed the graph that describes the high frequency distortion? At what frequency does this start? Also am I correct a sinad of 85 "translates" into a very good distortion measurement of .0056?

 

[jhaider]

What do you see out there that get's you 8 channels over 100w in one chassis? Maybe a multichannel home theater amp? Actual curiousity here, no snark.

Don't forget it's not only 1 chassis, it's one one rack unit high chassis, with passive cooling.