O2: High Performance Portable/Home Amp

[Rapide_23]

Originally Posted by NwAvGuy

Of course, if you're a certain eBay vendor, you just permanently configure the fashionable DAC chip for 24/96 and it totally falls on its face when using USB at 16/44 (I have just such a DAC sitting right here).

Certain ebay vendor?

[CrowdAds]

[NwAvGuy]

Originally Posted by Sorensiim

Ooohh.... That purple one sure is screaming for an acryllic enclosure. But that would require me to vastly improve my soldering skills

That's actually the uberO2 Purple Edition which uses crystal aligned oxygen free cryogenic treated copper while the board dialectric has been enhanced with an ultra pure form of unobtainium. The board was developed especially for certain members of a large headphone forum.

Originally Posted by Oliver Freeborn

I'm also talking with Rapid r.e. black versions and they are currently getting me a quote. However, it looks likely that there will be a higher MOQ for them.

Black is nice for the enclosure (and I'm also rather fond of BoxEnclosure's blue). But, for a machined front panel, silver is best so the machining (raw aluminum) doesn't show. If the panels are made in high enough volume, they can be anodized or painted after they're machined which solves that problem.

Originally Posted by Rapide_23

Certain ebay vendor?

It's a semi-established Asian brand with a significant following. I've been very positive about Sansa, Behringer, Beyer, Benchmark, Sennheiser, and said nice things about several other companies like HRT, Leckerton, Grace Designs, etc. But, despite all that, I've been heavily accused of having some hidden agenda to "destroy the reputation" of "perfectly good" companies. So, for now, I'm not looking to add anyone new to that list and give my critics yet more ammunition.

When I first tested the DAC in question on the dScope I was kind of shocked--especially when I found out the cause. The DAC chip only works correctly at 96 Khz using the S/PDIF input. It has very significant problems via USB or 44/48 via S/PDIF which is how most probably use it.

[xnor]

Originally Posted by NwAvGuy

But, despite all that, I've been heavily accused of having some hidden agenda to "destroy the reputation" of "perfectly good" companies.

Tbh, I'd rather see the reputation of products "destroyed" that don't earn it in the first place. It's the hype reviews and FOTMs that create the illusion of quality and, even if those who accuse you don't get it, that is more of a harm than presenting hard facts. Reviews without facts are like reading a fantasy book - entertaining, at best.

Nice to see problems getting fixed immediately, NwAvGuy. Good work!

[Satellite_6]

I think I have some unobtainuim sitting around. Maybe I could use it to way down my O2 for fun. (In the real world I think it's called galena)

Seriously tho, can't stop now. The flea infested cat is out of the bag.

[MrSlim]

Originally Posted by Soaa

Thanks for the input. Are there any good compact DACs that you recommend? Any opinions on Headstage's DAC cable?

I look forward to more articles on DACs from you.

I didn't dig all the way through the thread to see if this was answered, but in case it hasn't, I wanted to say that I have the cable DAC, and I think it's one of the best bang-for-the-buck DAC's out there. They're not afraid of listing the components either, which is nice, and they are very good quality.

[mikeaj]

Originally Posted by MrSlim

I didn't dig all the way through the thread to see if this was answered, but in case it hasn't, I wanted to say that I have the cable DAC, and I think it's one of the best bang-for-the-buck DAC's out there. They're not afraid of listing the components either, which is nice, and they are very good quality.

Indeed it seems many companies like to flaunt their name-brand components, but who knows how the performance really is.

btw for those wondering, I think you're talking about this:
http://www.headstage.com/USB-DAC-Cab...30::10134.html

It's an interesting-looking device, that's for sure: a mini USB DAC mostly inline of a cable. Like some other options, it looks like this one only supports 48 kHz. Bummer.


Then again, foobar2000 (and other software music players) can do resampling themselves, or through plugins. I've not tested it, but I suspect that their resampling routines shouldn't effect the sound much, and you could use that to resample 44.1 kHz to 48 kHz--if the DAC only supports 48 kHz--instead of relying on something else to do the job. I hope.

[NwAvGuy]

Thanks for the DAC tip. I'm ideally looking for something a bit more mainstream and mass produced if possible. The sort of thing you can buy on Amazon.com, etc. If that turns out not to be possible, niche low volume products like the Cable DAC might be worth looking into.

I do, however, take issue with the quote "the only serious portable high-end headphone amplifier" on this page:

http://www.headstage.com/The-Arrow-Amp:::1012.html

[Satellite_6]

Probably beating a dead horse here, but how can a DAC not support 44.1? (rhetorical)

Ridiculus.

[Soaa]

Originally Posted by NwAvGuy

I do, however, take issue with the quote "the only serious portable high-end headphone amplifier" on this page:

http://www.headstage.com/The-Arrow-Amp:::1012.html

Well, dfkt did measure and review it, saying it's one of the best ones around... It appears to be a well engineered product, having been through several revisions, and sports pretty interesting features. And it is thinner than the O2 can ever dream to be.

I wonder how it would measure on the dScope, compared to the O2? Might be an interesting comparison.

[upstateguy]

The whole gang's here !

Glad to see everybody.

[NwAvGuy]

Originally Posted by Soaa

Well, dfkt did measure and review it, saying it's one of the best ones around... It appears to be a well engineered product, having been through several revisions, and sports pretty interesting features. And it is thinner than the O2 can ever dream to be.

I wonder how it would measure on the dScope, compared to the O2? Might be an interesting comparison.

EDIT: Thanks for the links Satellite!

I agree it might be an interesting comparison. Based on what I know so far, here's an engineer's view of why I questioned Headphonia's claim the Arrow is "the only serious portable high-end headphone amplifier".

The website says it's available in June but still says pre-order. What's the story? Is it shipping?

It's cool it has crossfeed, some EQ, 3 gain settings, auto shut off, and a microcontroller (although the latter might increase the noise floor). The size is also appealing. And I like the idea of variable power supply voltages so you can exchange run time for maximum output voltage. The battery life numbers are impressive. But when it comes to the actual audio performance, I'm not as enthusiastic.

The website says it uses a single Li-Ion battery with a DC-DC converter to step it up to 12 volts. And it uses the same AD8397 that's in the Mini3. That's only good for 3.8 volts RMS output versus over 7 V RMS with the O2. That's a big difference (nearly 6 dB).

The website says it works with "full size/high ohm" headphones but that should be qualified. The Arrow, for example, will fall short with the Beyer DT880-600 which need at least 5 V RMS. The O2 doesn't fall short with 5V on battery and 7+ V on AC power.

It's not clear if it's using a virtual ground, capacitor coupled output, or a true bipolar power supply. The first two will seriously compromise its real world performance--especially into low impedance loads. If it is a true bipolar supply I would have expected them to have mentioned it as that's a significant advantage. The fact it will run with "no voltage boost" in its lowest power mode implies it's NOT a bipolar power supply.

The AD8397 is not short circuit protected and loves to blow up. So they either needed to add elaborate protection circuitry, or more likely, they added series resistors in the outputs raising the output impedance to unacceptable levels with many headphones. Or, they tried the Mini3's band-aid, and put the resistors inside the feedback loop which creates stability and distortion problems. If they did none of the above, the AD8397 is on borrowed time as headphone plugs create a short circuit every time you plug, and unplug, them.

If it uses a virtual ground or capacitor coupled output, I can just about guarantee the O2 will out perform it in just about every way. If it's a bipolar power supply (unlikely), there's still the short circuit/output resistor/output impedance issue and the lack of voltage swing for high impedance headphones. There might also be stability issues with the relatively high-strung AD8397.

Finally, there's no such thing as a "4.2V" Li-Ion battery. They're 3.7 or 3.8 volts. Someone's getting creative with the website marketing.

All the O2 parts come to $75 including case, batteries, and AC adapter. Add a generous $25 for shipping and you're at $100. You can still buy three O2's for the price of one Arrow. Or buy one O2 and have $200 left over to spend on enjoying life more. I realize I'm comparing DIY to a finished amp. But I believe, if the O2 takes off, you'll be able to buy a finished amp, or at least a kit with a pre-soldered circuit board, for around $100.

I would need to get an Arrow on the bench to find out the real story. I would be surprised if it outperforms the O2 in any meaningful way and the opposite seems more likely. At the least, the O2 will drive a wider variety of headphones to realistic volumes.

Also, there are almost no meaningful specs on the Headphonia web page. If Headphonia has confirmed specs based on measurements, I'm sure many would love to see them. Test results from an audio analyzer would be even better.

All that said, it's entirely possible the Arrow is worth $300 to some people for its size, feature set, etc. That's especially true if the audio performance turns out to be respectable within its limitations.

If the designer, or someone from Headphonia, is here on ABI I would welcome their comments on what I've outlined above.

[Satellite_6]

Hmmm. . .

front page

thread

RMAA (link in thread)

[NwAvGuy]

Thanks Satellite. I just typed "Arrow" into the search box and didn't find it. Duh.

I probably need to investigate more how dfkt makes his RMAA measurements. But RMAA has no concept of voltage, current, or power. So it makes a huge difference what absolute level the tests are run at. This test for example shows the 3rd harmonic over my standard threshold of -80 dB. But what level is it?

Arrow 12HE 16 Ohms THD

And I'm not sure what's going on here with "Imp" but - 3dB at 40 hz isn't good at all:

Arrow 12HE 16 Ohms FR

Hopefully dfkt can help answer the above.

[mikeaj]

Don't blame it on the imps, the poor critters! Seriously though, it seems to be an issue with the way the gain setting is implemented instead.

300 ohms, bass 0, CF 0, gain variable, imp 0

Actually it seems like there are:
Three settings of bass boost (including none)
Three levels of crossfeed (including none)
Three levels of gain (1.2dB, 10.5dB, 20.9dB)
Three levels of output impedance (10 ohms, 70 ohms, 120 ohms)

That's what an "imp" is. Some of those IMD graphs don't look so tasty either.

[NwAvGuy]

Ouch! Well if the minimum output impedance is 10 ohms, the Arrow officially rates a solid FAIL for any headphones under 80 ohms. So all you Grado, AKG, Denon, Shure, Ultimate Ears, Westone, HiFiMan, Etymotic, etc. owners can forget it unless you want messed up frequency response and/or under damped bass. See Headphone & Amp Impedance for why.

10 ohms flat destroys the frequency response of any balanced armature IEM I know of. And it's enough to degrade the bass performance and still create frequency response issues with lots of other headphones including many full size cans.

But, sadly, it's not that uncommon. The FiiO E9 has the same problem (but the E7 and E5 are well under 2 ohms). In this case it's probably necessary for stability and short circuit protection for the fragile AD8397 op amp used in the Arrow.

[eroarex]

so anyone still not selling the full kit? as usuall the farnell/mouser shops in my country (though i live in eu) delivery rates are double than the cost of the parts so it won't happen

so sad

[NwAvGuy]

Originally Posted by eroarex

so anyone still not selling the full kit? as usuall the farnell/mouser shops in my country (though i live in eu) delivery rates are double than the cost of the parts so it won't happen

Give the O2 some time. It was just fully released to the public 6 days ago. If the amp is worth owning, and proves to be popular, I'm sure there will be kits from someone.

Nobody else even has an O2 yet besides one person in Australia who built one without a PC board. Those who might offer a kit are smart to wait until the design is more proven. Otherwise they might spend $3000 for parts for 100 kits only to have the design change. So I don't blame them for waiting.

[NwAvGuy]

I know this old history, but some earlier comments from another forum were just forwarded to me. They were made by someone with a commercial interesting in a competing product to the O2. I'd like to officially address them as the person who made them is unwilling to post in this official O2 thread. His claims:

"Dual 8.4V batteries are specified, let's say that batteries are charged and does not fall below the rated voltage, we have 16.8V total spread, minus the forward drop of the two 1N5818 schottky diodes (0.5V each), this reduces the available voltage down to 15.8V. And let's assume no further voltage drops in the power supply -- in fact there is some drop across the MOSFETs, but probably small enough to ignore for this discussion. The NJM2608 opamp is not rail-to-rail. It could only swing to about 2V above the negative rail and about 1V below the positive rail. The signal's negative peaks will therefore clip first and effectively we have a maximum output voltage swing of 11.8Vp-p."

First of all, it's fascinating he's considering "fully charged 8.4V" batteries to be 8.4 volts. In trying to defend his design's erroneous specs, he's claimed a "fully charged 8.4V" battery is 10+ volts. Interesting! But let's work with 8.4 just to be consistent.

The measurements are all in the first article and I included the graph below. The O2 managed 5.3 V - 5.5V RMS on batteries measuring 9 volts. Taking even the lower number, that's 15 V p-p, not 11.8 as he claims. And even if we subtract another 0.6 volts from each rail for more 50% charged 8.4V batteries (the number he uses) that's still 13.8 V p-p or about 4.9 V RMS.

By comparison, the O2's closest competitor the Mini 3, running on a nearly fully charged battery, can only manage less than half the voltage swing (2.3 V RMS). That's 1/4 the power, and 6 dB less volume into the same 150 ohm load at the same 1% THD (the accepted measurements for power at clipping).

His error is not using the correct values. For example he's using the 1 amp drop for the diodes but they're only operating at less than 0.1 amps in the above test and have much less drop (there's a graph on the datasheet showing just that). He either is trying to deceive people with the wrong numbers or perhaps doesn't understand how diodes work. And he apparently never bothered to look at my measurement graphs or he's just blindly assuming they're wrong with no proof to the contrary.


"The volume pot is not at the input to attenuate the input signal, so the input opamp "sees" the full output voltage from the source. Let's assume a standard Redbook audio CD player's output voltage at 0dBFS of 2Vrms, which is 5.7Vp-p. Even at the lowest gain setting of 2x, the opamp output will be swinging 11.3Vp-p which is right at the verge of clipping. If you switch the gain any higher, it will clip rather severely."

This horse has been well beat. Not many people will run the O2 from battery while using a home source which is his "straw man" argument above. And, even at 2.5X gain, with reasonably charged batteries, they still can and I published the graph proving it. At 2X gain it's a breeze. Again, he's using the wrong peak-to-peak voltage swing (see above). The proof has been published earlier in this thread. On AC power the O2 doesn't overload until nearly 3V input at the default gain.


"The designer states on the schematic "Input 4V RMS max". If that was true, we'd need to be able to swing 22.6Vp-p at the output of the opamp for a gain of 2x (and more for higher gains). Neither a battery-powered nor a wall-powered version of this amp could do that without clipping."

As already explained, the 4V number was in reference to damaging the op amp from too much voltage. I thought that would be fairly obvious as I state elsewhere, in several places, how the max input voltage is related to the gain chosen. I've since corrected the schematic to be explicitly clear what the input range is.


"This is irrespective of the headphone sensitivity or volume pot position. While similar concepts have been used in commercial gear, the input stage in them would be powered by much higher supply voltages in order to avoid clipping, but that is clearly not an option here due to the batteries."

Commercial gear places the volume pot "in the middle" just as I did for some very good reasons. And the "much higher supply voltages" are usually only 25% higher (15 vs 12) than the O2 on AC power. Because the O2 uses a professional-grade gain structure, it has far lower noise than the Mini3. It also doesn't make "rustling" noises when you change the volume setting. And it has lower DC offset and far lower distortion due to its multi-stage design.

Everything in audio design is a trade off. Even "ultimate" designs trade-off cost and size for their performance. In this case the trade off is using the O2 on AC power when using home AC powered sources. And for someone with a highly unusual source, they may need to simply clip two resistors to lower the O2's gain. That's it. In exchange for those relatively minor trade-offs, the O2 delivers far better performance than the Mini3, can drive far more headphones with much higher output power, and even has far better battery life. All for a lot less money.

If the person making the above comments doesn't agree, I would very much encourage him to join the discussion here. Here's the graph comparing the O2 on battery to the Mini3 real world apple-to-apples:

[Soaa]

Well, the O2 thread on Head-Fi got locked just a minute ago. Not much of a loss, since there was barely any actual discussion there. It seems their admins didn't even care about the integrity of the thread, removing informational links on a whim. If anything, the thread there being locked means less ridiculous claims to refute.

[b0ck3n]

AMB is what the Head-Fi administration calls a "real scientist". "We're in trouble when real scientists don't want to join the discussion" - indeed, they don't return once their erroneous claim have been refuted.