DAC+Amp recommendations for HD 490 Pro

[kayson]

Hi everyone. Excited to have found an audio community that understands and appreciates signal processing and physics

I'm looking to upgrade my $70 gaming headset to something a little more premium. After much research and debate, I've (mostly) settled on the HD 490 Pros. They seem to be a good fit for my use case (gaming, classical music, teams calls) based on their transparency, sound stage, and price point. Side note: planning to get an Antlion Mod Mic USB 2.

The main piece that's missing is a DAC+Amp. My PC's motherboard audio quality isn't great because it picks up a lot of junk, and the driver and interface even worse. My wishlist is as follows, roughly in order, though none are dealbreakers:

* Can drive the 490s with good transparency
* Slim height as I'll be mounting it under my desk
* Balanced output
* Powered over USB (i.e. no additional power cable needed
* $200ish range
* No volume knob (mainly for compactness).

After much searching and requesting recommendations on reddit, which was not very fruitful, I haven't found anything that checks all the boxes. Maybe nothing exists, but nonetheless I'm hoping to get more suggestions.

I was a little surprised that I couldn't find an open source DIY design given that the market has converged to a small number of USB controllers, DACs, and signal chains. My searches only turned up a couple of 5-10 year old abandoned projects. Would be happy to spin a PCB and do some soldering if there were a design readily available.

Right now, top of my shortlist seems to be the iFi Zen DAC 3 though it's a bit bigger than I'd hoped. Fiio K3S or K11 are also options but I hesitate to buy from a company that ships a product with a "24bit" R2R with discrete 0.1% resistors...

Thanks!

 

[JIW]

Hi everyone. Excited to have found an audio community that understands and appreciates signal processing and physics

I'm looking to upgrade my $70 gaming headset to something a little more premium. After much research and debate, I've (mostly) settled on the HD 490 Pros. They seem to be a good fit for my use case (gaming, classical music, teams calls) based on their transparency, sound stage, and price point. Side note: planning to get an Antlion Mod Mic USB 2.

The main piece that's missing is a DAC+Amp. My PC's motherboard audio quality isn't great because it picks up a lot of junk, and the driver and interface even worse. My wishlist is as follows, roughly in order, though none are dealbreakers:

* Can drive the 490s with good transparency
* Slim height as I'll be mounting it under my desk
* Balanced output
* Powered over USB (i.e. no additional power cable needed
* $200ish range
* No volume knob (mainly for compactness).

After much searching and requesting recommendations on reddit, which was not very fruitful, I haven't found anything that checks all the boxes. Maybe nothing exists, but nonetheless I'm hoping to get more suggestions.

I was a little surprised that I couldn't find an open source DIY design given that the market has converged to a small number of USB controllers, DACs, and signal chains. My searches only turned up a couple of 5-10 year old abandoned projects. Would be happy to spin a PCB and do some soldering if there were a design readily available.

Right now, top of my shortlist seems to be the iFi Zen DAC 3 though it's a bit bigger than I'd hoped. Fiio K3S or K11 are also options but I hesitate to buy from a company that ships a product with a "24bit" R2R with discrete 0.1% resistors...

Thanks!

Balanced output is of no benefit other than higher level and higher SNR but at the cost of higher noise because headphones are differential receivers. A good single ended design will do you just fine.

Consider the Topping DX1 (Measurements). It should do around 3.5 V RMS or around 11 dBV for the HD490 Pro which according to Amir's review have a sensitivity of 109.70 DB SPL at 1 V RMS. This is enough for maximum peak SPL of around 123-124 dB SPL. More than enough.

Alternatively, consider a dongle.

 

[kayson]

Balanced output is of no benefit other than higher level and higher SNR but at the cost of higher noise because headphones are differential receivers. A good single ended design will do you just fine.

Consider the Topping DX1 (Measurements). It should do around 3.5 V RMS or around 11 dBV for the HD490 Pro which according to Amir's review have a sensitivity of 109.70 DB SPL at 1 V RMS. This is enough for maximum peak SPL of around 123-124 dB SPL. More than enough.

Alternatively, consider a dongle.

The other benefits for balanced outputs are better noise immunity and protection against ground loops. I don't expect ground loop issues unless there's some horribly designed supply, but my old gaming headset picked up all kinds of junk from the motherboard. Granted, my current headset is a cheap USB DAC with SE output and does better, but it's not perfect.

What makes you say it's higher noise? A signal being SE vs differential doesn't fundamentally change the noise floor. In fact, since all the DAC chips are internally differential, I'd expect SE to have slightly higher noise by virtue of having to do the conversion.

It seems this moondrop dawn pro would best the DX1 in every respect, at least based on the measurements... Not sure how higher impedance of the HD 490 Pro would affect things, though.

(Unofficial) Moon Drop Dawn Pro Balanced DAC/Headphone AMP Measurements

I have measured the Moon Drop Dawn Pro with Audio Precision APx555 recently and I would like to share with you. I bought the Dawn Pro myself. Here's the specifications from the manufacture. Just in case. The Dawn Pro is a small and light DAC/Headphone AMP, Only 42mm * 22.45mm * 12.39mm and...

www.audiosciencereview.com

 

[JIW]

The other benefits for balanced outputs are better noise immunity and protection against ground loops. I don't expect ground loop issues unless there's some horribly designed supply, but my old gaming headset picked up all kinds of junk from the motherboard. Granted, my current headset is a cheap USB DAC with SE output and does better, but it's not perfect.

What makes you say it's higher noise? A signal being SE vs differential doesn't fundamentally change the noise floor. In fact, since all the DAC chips are internally differential, I'd expect SE to have slightly higher noise by virtue of having to do the conversion.

It seems this moondrop dawn pro would best the DX1 in every respect, at least based on the measurements... Not sure how higher impedance of the HD 490 Pro would affect things, though.

(Unofficial) Moon Drop Dawn Pro Balanced DAC/Headphone AMP Measurements

I have measured the Moon Drop Dawn Pro with Audio Precision APx555 recently and I would like to share with you. I bought the Dawn Pro myself. Here's the specifications from the manufacture. Just in case. The Dawn Pro is a small and light DAC/Headphone AMP, Only 42mm * 22.45mm * 12.39mm and...

www.audiosciencereview.com

Noise immunity and ground loop protection is because of differential signalling. Since headphones are differential receivers because they have no ground reference, there is no difference from how the signal is applied. Ground loops and such are introduced before the headphone amplifier and will thus be amplified to the same degree for a balanced and single-ended amp. To avoid this, a USB isolator may work.

DAC output stages typically produce the same level for either positive and negative output and rout the positive to the single ended output or do a differential summing which increases SNR from the ingoing signal by the square root of 2. The outgoing signal also contains the noise from the circuit doing the operation but that might still amount to less overall. Besides, my point was entirely about amplification method.

For the same output level, a balanced amp operated at half the gain per amplifier compared to a single ended design. In that case, the signal is as high but noise is reduced by the square root of 2 due to being uncorrelated. Typically, balanced amplifiers have twice the output of a single ended one and thus a noise floor that is higher by the square root of 2.

The Moondrop has no low impedance power measurements. The product page is also confusing. If it can do 4 V at lower impedance that would be plenty.

 

[kayson]

Noise immunity and ground loop protection is because of differential signalling. Since headphones are differential receivers because they have no ground reference, there is no difference from how the signal is applied. Ground loops and such are introduced before the headphone amplifier and will thus be amplified to the same degree for a balanced and single-ended amp. To avoid this, a USB isolator may work.

But the problem is that with single ended outputs, one of the "signals" to the headphones - is - the ground, right? Any noise dumped onto the ground itself will appear on the headphones by virtue of the amp output having some amount of ground rejection. Isolation would help if that noise is coming from a ground loop, where there's some difference between two separate or joined grounds in the circuit, but not if some aggressor (e.g. motherboard) is making a single ground itself dirty. I suppose it would be easy enough to just try it out and see.

DAC output stages typically produce the same level positive and negative output and rout the positive to the single ended output or do a differential summing which increases SNR from the ingoing signal by the square root of 2. The outgoing signal also contains the noise from the circuit doing the operation but that might still amount to less overall. Besides, my point was entirely about amplification method.

For the same output level, a balanced amp operated at half the gain per amplifier compared to a single ended design. In that case, the signal is as high but noise is only increased by the square root of 2 due to being uncorrelated. Typically, balanced amplifiers have twice the output of a single ended one and thus a noise floor that is higher by the square root of 2

Ah I see what you mean. It's not from being single ended in a mathematical sense but that the single ended output is supply limited to half the range so you're effectively attenuating the whole signal (plus noise).

But your point about the balanced amplification vs single ended to the same output level doesn't make sense. If your balanced SNR is 3dB higher, as you've pointed out, then at the same output level, your noise floor would be 3dB - lower - (3dB higher noise - 6dB gain reduction). Assuming the amplification doesn't dominate the noise, which I think it wouldn't for a well designed noise system (but I come from the land of RF not audio. Even if it did, you could attenuate the output by 6dB to achieve the same effect, though I realize it would partially defeat the purpose of having balanced output in that you can achieve higher power.

The Moondrop has no low impedance power measurements. The product page is also confusing. If it can do 4 V at lower impedance that would be plenty.

You mean higher impedance? Or am I missing something?

 

[JIW]

But the problem is that with single ended outputs, one of the "signals" to the headphones - is - the ground, right? Any noise dumped onto the ground itself will appear on the headphones by virtue of the amp output having some amount of ground rejection. Isolation would help if that noise is coming from a ground loop, where there's some difference between two separate or joined grounds in the circuit, but not if some aggressor (e.g. motherboard) is making a single ground itself dirty. I suppose it would be easy enough to just try it out and see.


Ah I see what you mean. It's not from being single ended in a mathematical sense but that the single ended output is supply limited to half the range so you're effectively attenuating the whole signal (plus noise).

But your point about the balanced amplification vs single ended to the same output level doesn't make sense. If your balanced SNR is 3dB higher, as you've pointed out, then at the same output level, your noise floor would be 3dB - lower - (3dB higher noise - 6dB gain reduction). Assuming the amplification doesn't dominate the noise, which I think it wouldn't for a well designed noise system (but I come from the land of RF not audio. Even if it did, you could attenuate the output by 6dB to achieve the same effect, though I realize it would partially defeat the purpose of having balanced output in that you can achieve higher power.


You mean higher impedance? Or am I missing something?

The headphones respond to the voltage across the terminals. If there is noise or any other deviation from 0 on the ground connection that is not present in the amplified signal, that would only show up on single ended connection - signal S = Pos - Gnd for single ended vs. S = Pos - Neg = 2xPos. I am not aware of that being a problem but it might be. Common mode noise rejection should be the same for balanced and single ended though unless the amplifiers output impedance and the impedance to ground differ significantly.

Balanced SNR is 3 dB lower for the same output level but balanced typically has 6 dB higher output and thus 3 dB higher noise. This assumes any amp's SNR is the same for both levels. For a complete device with preceding circuit, if the amplifier is behind the volume control attenuating an analog signal, the balanced version with higher gain may have lower noise - noise_tot^2 = (volume*noise_pre)^2 + (N_amp*noise_amp)^2 for uncorrelated noise. However, in many DAC/HPAs, volume control is done digitally before conversion and thus noise will most likely be higher for higher gain at the same output level.

EDIT: I see I mistyped it in the previous post. It should have said that it is lowered by the square root of 2 for the same level.

The measurement is directly into the ADC with high input impedance - 200 kOhm on the figure. I have read claims of 160 mW at 16 Ohm which is more than enough current for 4 V at 130 Ohm but the power graph suggests 2 W at 300 Ohm which is 24.5 V and thus unbelievable.

 

[AnalogSteph]

My PC's motherboard audio quality isn't great because it picks up a lot of junk, and the driver and interface even worse.

I don't expect ground loop issues unless there's some horribly designed supply, but my old gaming headset picked up all kinds of junk from the motherboard. Granted, my current headset is a cheap USB DAC with SE output and does better, but it's not perfect.

It puzzles me that you're seeing issues of such a magnitude in this day and age. I wouldn't expect that many current-day boards to be problematic, except if it's a super-cheap 4-layer job or something.

Troubleshooting steps you can try:

1. Unplug front panel audio. Some cases still connect audio and USB/chassis ground, which can result in severe ground loops that at times may even affect the back outputs.
2. Make sure power supply screws have good "bite" and ensure a low-resistance connection between power supply and chassis. People have gone as far as sanding the contact area and resorting to bare metal screws. I don't necessarily trust these modern PC cases in general when it comes to providing good conductivity throughout. I haven't had any issues with my Fractal Define R5, but that's a decade+ old design in itself now.
3. Don't be stingy with GPU power cables. Even if daisy-chaining two 8-pins from one cable may still be within electrical limits, the fewer ground returns are available there the more current will make its way back via the PCIe slot, motherboard and its power connectors.
4. Reseat 24-pin, 4-pin and PCIe power connectors on both ends if the power supply is modular.

Anyway, good luck finding much that matches this combination of criteria, I can't imagine it's very common.

* Can drive the 490s with good transparency
* No volume knob (mainly for compactness).

Well, actually there's a bunch of dongles, but do you really want to be dealing with a stupid dongle on a PC? They're not typically the last word in mechanical robustness, even if I imagine the fancier kind with detachable cables would do better than bread and butter cheapies. Electrically, any decent CS43130 or AK4377 implementation should be fine anyway.

A balanced output can be useful in getting higher voltages out of limited supplies, but as mentioned, there generally is precious little need for it otherwise. The shared ground return contact resistance of a headphone jack might be something like 50 mOhms, that's actually quite irrelevant when dealing with dealing with 120 ohm headphones. The important part is the headphone cable, where I've already seen a full Ohm over 3 meters, so the returns better be kept separate right up to the end. Sennheiser traditionally tends to be pretty good in this regard, and the HD490Pro with its 4-pin Mini-XLR connection is no exception.

You mean higher impedance? Or am I missing something?

No, he did mean lower impedance. Providing the same voltage into lower impedance means higher current, and that makes it harder to drive, so maximum voltage pretty much invariably tanks down there as the output runs into current limiting (we're talking the lower double-digit ohms range in particular). If you want to go easy on the output, the best choice will be the highest-impedance headphone that still is sensitive enough to reach target maximum levels. In that sense, one of the easiest-to-drive full-size headphones I know would be the ATH-R70x, which at 470 ohms nominal is still as sensitive as typical 300 ohm Sennheisers (namely 102 dB SPL / 1 V). (Big ears need not apply though. Anything over 6 cm end-to-end is pretty much out.)

 

[JIW]

It puzzles me that you're seeing issues of such a magnitude in this day and age. I wouldn't expect that many current-day boards to be problematic, except if it's a super-cheap 4-layer job or something.

Troubleshooting steps you can try:

1. Unplug front panel audio. Some cases still connect audio and USB/chassis ground, which can result in severe ground loops that at times may even affect the back outputs.
2. Make sure power supply screws have good "bite" and ensure a low-resistance connection between power supply and chassis. People have gone as far as sanding the contact area and resorting to bare metal screws. I don't necessarily trust these modern PC cases in general when it comes to providing good conductivity throughout. I haven't had any issues with my Fractal Define R5, but that's a decade+ old design in itself now.
3. Don't be stingy with GPU power cables. Even if daisy-chaining two 8-pins from one cable may still be within electrical limits, the fewer ground returns are available there the more current will make its way back via the PCIe slot, motherboard and its power connectors.
4. Reseat 24-pin, 4-pin and PCIe power connectors on both ends if the power supply is modular.

Anyway, good luck finding much that matches this combination of criteria, I can't imagine it's very common.

Well, actually there's a bunch of dongles, but do you really want to be dealing with a stupid dongle on a PC? They're not typically the last word in mechanical robustness, even if I imagine the fancier kind with detachable cables would do better than bread and butter cheapies. Electrically, any decent CS43130 or AK4377 implementation should be fine anyway.

A balanced output can be useful in getting higher voltages out of limited supplies, but as mentioned, there generally is precious little need for it otherwise. The shared ground return contact resistance of a headphone jack might be something like 50 mOhms, that's actually quite irrelevant when dealing with dealing with 120 ohm headphones. The important part is the headphone cable, where I've already seen a full Ohm over 3 meters, so the returns better be kept separate right up to the end. Sennheiser traditionally tends to be pretty good in this regard, and the HD490Pro with its 4-pin Mini-XLR connection is no exception.


No, he did mean lower impedance. Providing the same voltage into lower impedance means higher current, and that makes it harder to drive, so maximum voltage pretty much invariably tanks down there as the output runs into current limiting (we're talking the lower double-digit ohms range in particular). If you want to go easy on the output, the best choice will be the highest-impedance headphone that still is sensitive enough to reach target maximum levels. In that sense, one of the easiest-to-drive full-size headphones I know would be the ATH-R70x, which at 470 ohms nominal is still as sensitive as typical 300 ohm Sennheisers (namely 102 dB SPL / 1 V). (Big ears need not apply though. Anything over 6 cm end-to-end is pretty much out.)

The stock HD650 cable seems to have around 27 AWG wire according to this. The resistance for 27 AWG wire is 174.3 mOhm per meter. For the 3 meters of the HD650 cable, the resistance is 522.9 mOhm.

 

[unpluggged]

The stock HD650 cable seems to have around 27 AWG wire according to this.

This is a Litzendraht cable, so you can't judge the effective cross-section from a caliper measurement.

The resistance for 27 AWG wire is 174.3 mOhm per meter. For the 3 meters of the HD650 cable

The wire is twisted, so the net cable length depends on twisting density.

the resistance is 522.9 mOhm.

One can use a multimeter instead of making guesses.

 

[JIW]

This is a Litzendraht cable, so you can't judge the effective cross-section from a caliper measurement.


The wire is twisted, so the net cable length depends on twisting density.


One can use a multimeter instead of making guesses.

Fair points, but no guessing in the last part not present before. That is just the meter resistance of 27 AWG multiplied by the cable length. If I had a multimeter, I could have used it.

Judging by the picture, insulation is around 20% of cross-section and twisting is at around 30 degree angle. Adding the kevlar, conductive cross section is likely around 75% of total cross section. Total height gain is thus 3*tan(30/180*pi) = 1.73 m. Since the diameter of 27 AWG is 0.3036 mm, there are thus 3*tan(30/180*pi)/0.0003036 = 4803.247 windings which gives a total strand length of 3*tan(30/180*pi)/0.0003036*2*pi*0.0003606/2 = 3*tan(30/180*pi)*pi = 5.44 m.

The cross section of 27 AWG is 0.1021 mm^2 and thus conductive cross section is around 0.75*0.1021 = 0.076575 mm^2. The specific resistance of pure copper is 17.8 mOhm mm^2/m so per meter resistance is 17.8/(.1021*.75) = 232.45 mOhm/m. Thus, resistance of the cable is around 17.8/(0.1021*0.75)*3*tan(30/180*pi)*pi = 1264.86 mOhm.

 

[kayson]

For the sake of posterity: I ended up going with the moondrop dawn pro, modmic usb 2 as planned, and got the balanced cable (no reason not to with this particular dac). I'll report back when it all gets in.

 

[JIW]

Fair points, but no guessing in the last part not present before. That is just the meter resistance of 27 AWG multiplied by the cable length. If I had a multimeter, I could have used it.

Judging by the picture, insulation is around 20% of cross-section and twisting is at around 30 degree angle. Adding the kevlar, conductive cross section is likely around 75% of total cross section. Total height gain is thus 3*tan(30/180*pi) = 1.73 m. Since the diameter of 27 AWG is 0.3036 mm, there are thus 3*tan(30/180*pi)/0.0003036 = 4803.247 windings which gives a total strand length of 3*tan(30/180*pi)/0.0003036*2*pi*0.0003606/2 = 3*tan(30/180*pi)*pi = 5.44 m.

The cross section of 27 AWG is 0.1021 mm^2 and thus conductive cross section is around 0.75*0.1021 = 0.076575 mm^2. The specific resistance of pure copper is 17.8 mOhm mm^2/m so per meter resistance is 17.8/(.1021*.75) = 232.45 mOhm/m. Thus, resistance of the cable is around 17.8/(0.1021*0.75)*3*tan(30/180*pi)*pi = 1264.86 mOhm.

I got a multimeter and measured it. It is 0.6 Ohm from plug to headphone connector for positive and ground and 1.3 Ohm from ground to ground on the headphone connectors. Overload for all else so in excess of a 9999 million Ohm.

 

[AnalogSteph]

So that's a nominal damping factor of about 250. Since the HD650 cable is a 4-conductor affair, shared ground return resistance is almost nil (well, mostly whatever the plug contact resistance is, like 50 mOhms or so). Crosstalk would still be well in the green even if it wasn't, 300/0.6 = 500 and my rule of thumb would be >=100. My trusty HD590s get close to this minimum (100 ohms, ~1 ohm over 3 m). Pretty sure I've had in-ears that were below that - it easily happens with 16 ohm drivers and thin cables.

 

[kayson]

Got my HD 490 Pros (with balanced cable!), moondrop dawn pro, and modmic usb 2 set up. Grabbed an eqapo profile from the HD 490 review thread. Quite happy with the listening experience. Definitely didn't need the balanced cable for power, as with windows at 100%, I had to turn the volume control on the moondrop down a decent amount. But I appreciate knowing that I have 10% lower noise floor nonetheless

modmic usb 2 was... less good. The boom isn't quite long enough for headphones as thick as the 490s, so getting good mic levels was really tough. Ended up moving the mounting point as close to the edge as I could and moving the boom down in their screw mount (which they tell you not to touch) to get closer to my mouth. It will still occasionally pick up a mouse click or keyboard press, but fortunately Discord auto-sensitivity is pretty good.