O2: High Performance Portable/Home Amp

[upstateguy]

Originally Posted by khaos974

Just a remark, but comparing the O2 to the $1600 DAC1 Pre somewhat bothers me since it seems every article reminds us it.

It bothers me because it's a disingenuous comparison, the $995 DAC1 offers the exact same HPA2 headphone amplification module, moreover the DAC1, as its name indicates is a DAC first and foremost and has 2 of them. In fact, the HPA2 used to retail $150 as a stand alone unit.

Saying that the O2 compares favorably to the $150 HPA2 would be a more realistic comparison.

That's mighty unimpressive khaos.

Edit: http://www.head-fi.org/t/36595/benchmark-hpa-2 <--- something sounds familiar in this thread.

[CrowdAds]

[SpaceTimeMorph]

Originally Posted by khaos974

Just a remark, but comparing the O2 to the $1600 DAC1 Pre somewhat bothers me since it seems every article reminds us it.

It bothers me because it's a disingenuous comparison, the $995 DAC1 offers the exact same HPA2 headphone amplification module, moreover the DAC1, as its name indicates is a DAC first and foremost and has 2 of them. In fact, the HPA2 used to retail $150 as a stand alone unit.

Saying that the O2 compares favorably to the $150 HPA2 would be a more realistic comparison.

Agreed. Although if I remember right somewhere deep in one of his articles, nwavguy stated it wasn't a fair comparison based on the fact that the benchmark unit had an integral DAC section.

[NwAvGuy]

Sorry if the comparison bothers you Khaos. I'm really not trying to be disingenuous. One of the most important attributes for a reference product to compare aginst is that it's widely known which excludes the HPA2. It doesn't do much good to compare an amp to one few know anything about. I used the DAC1 Pre for the following reasons:

• It's listed under Class A Headphones & Accessories in this year's Stereophile recommended components. Please note, they gave it their top honors in the headphones section.
• It's been widely rave reviewed both subjectively and objectively--including its headphone output.
• It was one of the first products of its kind, and hence has been around a while so it tends to be fairly well known.
• It measures very well and its performance has been independently verified in multiple reviews including the headphone output (and is also fully documented with Audio Precision results by Benchmark).
• And, finally, I happen to have one. I'm doing the O2 entirely for free. Do you expect me to go out and buy other headphone amps when I have a perfectly valid one sitting right here?

I've said all along I hope the O2 is compared to other amps in blind and bench tests. I just started with what I have.

Once you pass the point of diminishing returns and meet the criteria I established I would be surprised if any headphone amp sounds much different in a blind test. The Violectric V90 is another likely example. That's the whole point of the O2. It's part of a relatively small group of products that are "way beyond the point of diminishing returns". And, as far as I know, it's the only reasonably priced portable amp that's a member of that club.

[khaos974]

Originally Posted by upstateguy

That's mighty unimpressive khaos.

Edit: http://www.head-fi.org/t/36595/benchmark-hpa-2 <--- something sounds familiar in this thread.

I was mostly referring to the Sound On Sound review of the DAC1

Originally Posted by NwAvGuy

Sorry if the comparison bothers you Khaos. I'm really not trying to be disingenuous. One of the most important attributes for a reference product to compare aginst is that it's widely known which excludes the HPA2. It doesn't do much good to compare an amp to one few know anything about. I used the DAC1 Pre for the following reasons:

• It's listed under Class A Headphones & Accessories in this year's Stereophile recommended components. Please note, they gave it their top honors in the headphones section.
• It's been widely rave reviewed both subjectively and objectively--including its headphone output.
• It was one of the first products of its kind, and hence has been around a while so it tends to be fairly well known.
• It measures very well and its performance has been independently verified in multiple reviews including the headphone output (and is also fully documented with Audio Precision results by Benchmark).
• And, finally, I happen to have one. I'm doing the O2 entirely for free. Do you expect me to go out and buy other headphone amps when I have a perfectly valid one sitting right here?

I've said all along I hope the O2 is compared to other amps in blind and bench tests. I just started with what I have.

Once you pass the point of diminishing returns and meet the criteria I established I would be surprised if any headphone amp sounds much different in a blind test. The Violectric V90 is another likely example. That's the whole point of the O2. It's part of a relatively small group of products that are "way beyond the point of diminishing returns". And, as far as I know, it's the only reasonably priced portable amp that's a member of that club.

I agree that the DAC1 Pre is a good reference point, and that's not what bothers me, what bothers me is quoting the *price* of the DAC1 Pre every time. If a quoted price was needed, quoting the $995 price of the DAC1 would be more honest, and quoting the $150 price of the stand alone amplifier module of the DAC1 would be even better.

[NwAvGuy]

I removed the price from the summary article. Hopefully the O2's in the wild can go hunt down some more expensive amps and my $500 blind challenge stands. Even if nobody takes me up on the challenge I plan to conduct more blind tests with the O2. I'll try to favor pure headphone amps, not headphone DACs, just for you Khaos. And the more expensive the better!

And, FWIW, my DAC1 Pre cost $1600. I can't change that. There are differences between the DAC1 models (like most of the op amps). I'm just reporting what the gear I compared the O2 against actually cost. I haven't tried the older more basic DAC1.

[Oliver Freeborn]

Just a quick note - payments are all orders are now being accepted. Head on over to https://docs.google.com/spreadsheet/...hl=en_US#gid=0 to see what you owe.

Many thanks!

[b0ck3n]

I don't see how it matters what amp you compare it too. The point is it's good enough - it will amplify the signal enough to play really loud with just about any headphone out there, and it does so transparantly. Whether another transparent amp costs $150, $950 or $1600 is irrelevant IMO - the only things that set them apart are the feature sets.

The point, as I see it, is making the statement that transparant, reference grade audio doesn't have to cost more than $100.

[NwAvGuy]

Thanks b0ck3n

Apparently the latest confusion on Heavily-Funded is regarding the O2's crosstalk. This is already documented in various places on my blog but it's kind of spread around. So just to get everyone on the same page. Someone asked:

"Would you please explain to me how output impedance affects crosstalk, or how feedback affects output impedance?"

In a virtual ground amp where the input grounds are isolated from the output grounds, which includes the M^3 and Mini3, the amp used for the virtual ground has no way to correct for ground impedance via feedback. So the entire ground channel amp's output impedance is shared between the channels and forms a voltage divider with each channel's output.

Here's the math. AMB claims -88 dB crosstalk into 33 ohms for the Mini3. Assuming 1 volt (just to make the math easier, it doesn't matter what the output level is) of output into AMB’s 33 ohms: AntiLog(-88/20) = 0.00004 volts (40 uV) of signal in the muted channel is the -88 dB claimed.

Because of the isolated input and output grounds, all ground impedance in the Mini3's virtual ground contributes to the crosstalk. So in the working channel the ground impedance forms a voltage divider with the 33 ohm load. To get the output in the muted channel below the required 40 uV (-88 dB) you have: 33/(1/0.00004 - 1) = 0.0013 ohms (1.3 mOhm)!

The headphone jack alone is rated for up to 50 mOhms! If you add in the copper PCB traces, ferrite bead, and most of all, the output impedance of the OPA690 ground amp, and you end up with about 150 mOhms total which gives somewhere around the –46 dB crosstalk I measured at 33 ohms. Shike measured an even worse -39 dB with RMAA at 33 ohms.

The interesting thing to note is that -88 dB number was using RMAA and included all the other results in the same test run (RMAA measures everything at once). The only way to get RMAA to display -88 dB I know of is to have essentially no load on the Mini3. If that's how the AMB's RMAA test was run that means all the other AMB measurements such as THD, IMD, etc are also completely invalid as well.

And all of AMB's 3 channel amps have impossible levels of crosstalk specified at 33 ohms including the M^3 and beta22. Each amp has even more impossible specs than the next cheaper version.

Someone asked: "I'm struggling to try and wrap my head around how exactly you can use any formula to calculate crosstalk, without actually performing a test to measure it. How do you calculate the impedance of an active ground channel that is referenced to the power rails?"

It's easy to show AMB's numbers are literally impossible and estimate a reasonably best case crosstalk. AMB's number still falls way short of even the best case analysis. Even NASA couldn't manage AMB's claimed numbers with his designs.

It's basic electronics theory--theory that anyone offering amp designs and selling components to thousands of people should understand. The math is above.

The ground impedance is so poor in virtual ground designs it, by far, dominates the cross talk measurement. The other sources of crosstalk are bugs on the windshield by comparison. The volume control, for example, might contribute about 50 uV of crosstalk at 1 volt of output but the virtual ground "system" is contributing 4500 uV of crosstalk by itself. So the math is:

Just the virtual ground = 20*log*(4500uV/1V) = -46.9 dB crosstalk

The virtual ground + volume control = 4550uV/1V) = -46.8 dB crosstalk

It's like testing a car for 0-60 MPH with an extra 3000 pounds of lead in the trunk and then complaining the tire pressure was off and slowing it down. To know what the car is otherwise capable you have to get rid of the 3000 pounds of lead first. And with an amp like the Mini3, M^3, or 3 channel beta22, you also have to get rid of the huge lump of performance destroying dead weight--the virtual ground.

If AMB can show where anything above is wrong I welcome him to contribute here and point out my factual errors. The same goes for anyone who understands the math. And for anyone wanting to know more I would suggest these 3 links:

O2 Measurements - skip to the crosstalk section

Virtual Grounds & 3 Channel Amps

Mini3 Review - skip to the start of the tech section and the crosstalk graph

[dfkt]

Screw the O2, this is how it's done - http://anythingbutipod.com/forum/sho...postcount=5668

[Satellite_6]

HAHAhahaHAHA. nice.

[Marvin the Martian]

Originally Posted by dfkt

Screw the O2, this is how it's done - http://anythingbutipod.com/forum/sho...postcount=5668

Where are the RMAA numbers?

[dfkt]

RMAA measurements are all TO THE MAX, what else?

(Sorry for derailing the thread once again.)

[Limp]

Originally Posted by dfkt

RMAA measurements are all TO THE MAX, what else?

(Sorry for derailing the thread once again.)

You finally manned up and downed that six-pack, then?

[xnor]

Originally Posted by NwAvGuy

If that's how the AMB's RMAA test was run that means all the other AMB measurements such as THD, IMD, etc are also completely invalid as well.

I asked amb a few month ago regarding how he did the RMAA tests and what output voltage (volume) he used, he replied: "I had measured the voltage at one time but I don't remember what it is now."

His tests were done at less than 1.25 Vrms anyway because that's where the input of his M-Audio FW AP clips (according to specs). But he could not confirm that either because "RMAA's concept of "0dB" is not the same as the sound card's input clipping threshold. It doesn't even know where that threshold is."

My thoughts: what the?

[NwAvGuy]

Originally Posted by xnor

I asked amb a few month ago regarding how he did the RMAA tests and what output voltage (volume) he used, he replied: "I had measured the voltage at one time but I don't remember what it is now."

His tests were done at less than 1.25 Vrms anyway because that's where the input of his M-Audio FW AP clips (according to specs). But he could not confirm that either because "RMAA's concept of "0dB" is not the same as the sound card's input clipping threshold. It doesn't even know where that threshold is."

My thoughts: what the?

Yeah, "what the... ?" exactly! It doesn't make sense to me how someone would be oh so detailed down to drawing pretty pictures of the test set up, quoting the model numbers of every piece of equipment used, etc., and then is unable to explain results that are way beyond impossible under the most ideal conditions. Or even what level the tests were run at.

Not only that, he's been called out on a lot of this at least 2 years ago. There were others questioning it long before I entered the scene this year. AMB even admitted a long time ago when challenged his 3 channel designs "double the output impedance". So how can he still claim in many places all over his website regarding 3 channels:

"This results in lower output impedance, greater linearity and reduced stereo crosstalk.”

Both can't possibly be true--"double" doesn't equate to "lower". I have copies of the posts.

And, just to be clear, voltage doesn't matter for crosstalk (as long as it's not so low that noise dominates the measurement or so high the amp or virtual ground is clipping). Crosstalk is strictly a relative measurement between the two channels--something RMAA can actually do fairly well.

[mikeaj]

This is ancient history, but while we're on the subject...a few months back during the Mini^3 debacle, I got some PMs in another (non-audio) forum relating to this subject. I had posted a link to some nwavguy review to an E7 or something like that in a headphone thread on that non-audio forum.

PMs were from some guy who knew Ti Kan personally and had been to his house before. As you might expect he was concerned with the Mini^3 review also on nwavguy's blog. (The guy sending the PMs was nice enough and very reasonable btw, just not understanding much of the electrical theory. I feel a little bit bad posting snippets of PMs below into a public forum without express consent, but it was made from an account with 0 posts, so that makes me feel less bad.)


Anyway, he mentioned a few things then that I thought would be interesting to disclose now. Below quotes are from some PMs.

edit: I was contacted within a day by the 0 post user asking me to take his PM snippets down, so I will of course honor that. I don't think they are any kind of indictment on AMB, just a perspective on maybe his side of the story, which I think is missing here because AMB won't post here.

[NwAvGuy]

First of all, The Mini3's performance, and the credibility of AMB is only on the table because AMB has personally attacked the O2 in some very misleading ways and made arguments the Mini3 is a better overall amp. And now AMB fans are attacking Shike for comparing the O2's crosstalk measurement to the Mini3's crosstalk measurement. As always, I welcome AMB to please join the discussion here on "neutral territory".

The 300 mW number, could probably be "faked" sort of like testing a car for O-60 going downhill with a tail wind. In other words, under highly unrealistic conditions, it might be possible to get a distorted 300 mW out of the Mini3 for the purposes of a brief demonstration with just one channel operating, the battery 30 seconds off the charger, etc. But if the test conditions are that unrealistic, why hasn't AMB revised his spec to be more honest or at least qualified the test conditions? How is he helping the DIY community by continuing to publish misleading specs? In reality, tested in industry standard ways, the Mini3 is only good for a bit over 100 mW per channel.

The crosstalk measurements are very different than the max power issue. Into a real 33 ohm load, the claimed crosstalk for all his amps is literally impossible by a wide margin. I've shown the math and I welcome anyone to please show me where I've made an error.

To get AMB's -88 dB the entire ground system can only have 0.0013 ohms of resistance. The headphone jack alone has more and the virtual ground amp has at least 38 times more. So there's no way AMB can make that measurement work unless the set up is "rigged"--i.e. it's really a 3300 ohm load. Even if there's something wrong with his test set up that he doesn't understand or isn't aware of, he should know enough to agree with the math and question his own results-especially when it's been pointed out to him by multiple people.

Shike also confirmed, on a completely different Mini3, that AMB's crosstalk number is very wrong. He's published that information in multiple places.

JCX (an electrical engineer who works with high-end op amps at his day job), at diyAUdio and elsewhere, has confirmed several of AMB's claims are impossible. See this post at diyAudio. There are more on other forums.

So I'm left to believe that perhaps AMB either has personal friends or loyal followers sticking up for him (which is entirely understandable), or he's essentially turned amplifier testing into a slight-of-hand magic trick. The best measurements are the ones that can be verified. And multiple people have tried to verify AMB's measurements and failed by huge margin.

As long as AMB is publishing information that hundreds or even thousands are using to decide on a design it's reasonable to expect others to try and verify that information. If there's strong evidence he's made an error, he should either correct whatever is misleading, or prove where his numbers are correct. In the case of the crosstalk, however, the math is irrefutable.

[Limp]

It's just like Polywater!

[xnor]

Originally Posted by Limp

It's just like Polywater!

Except that "the scientists who had originally advanced the case for polywater agreed that it did not exist". scnr.


Btw, how's the EHP-O2 testing and pre-ordering coming along? Is there similar interest in finished builds as in PC boards?

[mikeaj]

WAIT A SEC.

I just took another look at the AMB site. Here's the test setup:
http://www.amb.org/audio/mini3/


Is the test load in SERIES with the interface line in? Hot damn, now I totally believe the crosstalk results. I mean, it depends on how the black and red (RCA presumably) cables are hooked up to the dummy load box, and if the bottom end of the loads are really connected to ground or not. It's not possible to tell for certain what's inside that load box, but...still looks either weird or misleading in a bad way.

(Hopefully I didn't just overlook something dumb and made a fool of myself.)


edit: nevermind, active ground channel complicates things. I forgot what type of device we were dealing with.

http://www.amb.org/forum/nwavguy-blo...-t1107-60.html

from AMB:

Of course I use dummy loads in my tests. Those who I worked with (and those who have personally met me and seen what I do in testing) can vouch that the results are real. Differences between NwAvGuy's and my test results are due to several factors -- different equipment, different methodology, different assumptions, and a possibility that his sample of the Mini³ was not working properly. By the way, RMAA (or other computer sound card based) testing involving a 3-channel active ground amp is not straightforward, because the sound card -> amp -> soundcard "loop" would short circuit the amp's active output ground to its input ground if you used just a regular interconnect cable. I use is a concept introduced by Morsel (principal member of the PPA v1 and v2 projects, and co-designer of the M³ and early versions of the Mini³) many years ago. I don't claim invention of the testing scheme, nor do I claim invention of the 3-channel active ground concept. Morsel, Tangent, ppl (Phil Larocco) and KurtW had been designing and testing 3-channel amps this way long before I got into headphones.

Need to hunt down that testing procedure now.

But on second thought, this doesn't make sense, and that shouldn't effect proper loaded testing...