I'm interested, but I'd just like to ask how suitable this project would be for a complete beginner? My soldering experience is very limited (i.e soldering wires together),and I'd probably stick to a wallwart until I could build something like a Sigma 11/22. I missed out on The Wire project aswell, but this suits my needs better as I have an unbalanced setup.
Well... I wasn't going to post this, but since you asked.
At the risk of getting an "I told you so" from jcx...Before I realized that HA-INFO already had a new version out I emailed them back in February saying I liked my NG98 and suggesting they consider switching over to the LME49710 + LME49600 for the next version to take it to the next level. In the process of composing that email I looked at their website and discovered the new TPA6120 version. In addition to pitching the LME49600 in my email I then suggested the new version was an unfortunate change due to the chip change and the lack of analog input jack. Here is what came back. I've edited the personal info out.
_______________________________
From:XXXXX [ha info customer service]
Sent: Sunday, February 13, 2011 9:30 PM
To: XXXXX [me]
Subject: RE: Suggestions for the NG98
HI,
than you for your advice.
NG98MKII use TPA6120A2 as the headphone amplifier, it is better than op+buf method.
i'll give this mail to our engineer, any other questions please write to me.
Regards
XXXXXX
HA INFO
______________________
Hello,
My brain is stuck in the class A world where worst case for the heat sink dissipation is idle. Increase the power into the load and the heat into the heat sink goes down. I will go explore Class AB in the National data sheet.
Noodleing through a 32 ohm load drawing the maximum 250ma current from an audio buffer is 0.25 amps times 32 ohms equals 8 volts. 8 volts times 250ma equals 2 watts connected load. This is about worst case for the loaded audio buffer. This would blow a pair of headphones off your head. This is 2 watts into the load. I do not know how much will be dissipated into the heatsink. I am thinking much less than two watts.
I am going to dig through it!
DT
All just for fun!
Note from the National data sheet.
Note 3: The maximum power dissipation must be derated at elevated temperatures and is dictated by TJMAX, θJA, and the ambient temperature TA. The maximum
allowable power dissipation is PDMAX = (TJMAX–TA)/θJA or the number given in Absolute Maximum Ratings, whichever is lower. For the LME49600, typical
application (shown in Figure 2) with VSUPPLY = 30V, RL = 32Ω, the total power dissipation is 1.9W. θJA = 20°C/W for the TO–263 package mounted to 16in2 1oz
copper surface heat sink area.
Last edited:
@NakAttack, hard to say on the DIY skills. My intent was to have the board be really easy, but if you've never soldered any boards before, it's up to you if you want to take the risk. Once we get the design finalized there will be a decent estimate of the total cost. And you can decide from there.
I could probably better answer the question if I knew more... Are you mainly looking for a good headphone amp or a first DIY project? What headphones/application? I also have someone wanting me to test a few current eBay DIY headphone amp boards/kits. If any of those turn out to perform well I'll be happy to let everyone know. That might be a lower risk (cost) way to get your feet wet? Some of the eBay sellers also have assembled boards (or entire amps with enclosures) but if you go that route, I'd suggest sticking to one that's known to be a solid performer.
@DualTriode, yup close to 2 watts... which is also about what the TPA6120 has to dissipate. And that's not 2 watts to the headphones, that's 2 watts of heat per channel on the PC board for only milliwatts into the headphones.
@Agdr, thanks for the added info from HA. It doesn't really explain why, they just think it's better. It might be better mainly for marketing or other reasons.
And earlier this evening I ordered an NG98 MK II. It will probably take a week or three to get here but once it arrives I'll listen to it, bench test it, etc. And if it's at least decent, it can be part of the detailed comparison. If nothing else it should help evaluate the TPA6120 performs in terms of peak current, how close it can swing to the rails, real world thermal issues with various loads, etc. I'm willing to hack up the NG98's PCB, if need be, to use my own power supply, make modifications, etc. As I said earlier, I can always recycle the enclosure.
@JCX, Did you post a link somewhere to the 436 page paper Operational Amplifier Distortion authored by Groner in 2009? It's an interesting paper if you haven't seen it. And there's lots of other interesting information on his website.
Just to be clear, I fully believe the TPA6120 has some advantages on paper. And your design approach would likely yield an amp that, when using a "distortion magnifier", would also measure better. (For those who don't know, this performs a partial subtraction of the input from the output leaving more relative distortion which is then fed to whatever is analyzing the distortion--it's "half way" audio differencing). I just wish the TPA6120 was more DIY-friendly.
I have the Chinese TPA6120 design on the way and look forward to getting it on the bench. I'll try to keep an open mind when juggling the goals for this project and the TPA6120. Your input and suggestions have been helpful and are appreciated.
I could probably better answer the question if I knew more... Are you mainly looking for a good headphone amp or a first DIY project? What headphones/application? I also have someone wanting me to test a few current eBay DIY headphone amp boards/kits. If any of those turn out to perform well I'll be happy to let everyone know. That might be a lower risk (cost) way to get your feet wet? Some of the eBay sellers also have assembled boards (or entire amps with enclosures) but if you go that route, I'd suggest sticking to one that's known to be a solid performer.
@DualTriode, yup close to 2 watts... which is also about what the TPA6120 has to dissipate. And that's not 2 watts to the headphones, that's 2 watts of heat per channel on the PC board for only milliwatts into the headphones.
@Agdr, thanks for the added info from HA. It doesn't really explain why, they just think it's better. It might be better mainly for marketing or other reasons.
And earlier this evening I ordered an NG98 MK II. It will probably take a week or three to get here but once it arrives I'll listen to it, bench test it, etc. And if it's at least decent, it can be part of the detailed comparison. If nothing else it should help evaluate the TPA6120 performs in terms of peak current, how close it can swing to the rails, real world thermal issues with various loads, etc. I'm willing to hack up the NG98's PCB, if need be, to use my own power supply, make modifications, etc. As I said earlier, I can always recycle the enclosure.
@JCX, Did you post a link somewhere to the 436 page paper Operational Amplifier Distortion authored by Groner in 2009? It's an interesting paper if you haven't seen it. And there's lots of other interesting information on his website.
Just to be clear, I fully believe the TPA6120 has some advantages on paper. And your design approach would likely yield an amp that, when using a "distortion magnifier", would also measure better. (For those who don't know, this performs a partial subtraction of the input from the output leaving more relative distortion which is then fed to whatever is analyzing the distortion--it's "half way" audio differencing). I just wish the TPA6120 was more DIY-friendly.
I have the Chinese TPA6120 design on the way and look forward to getting it on the bench. I'll try to keep an open mind when juggling the goals for this project and the TPA6120. Your input and suggestions have been helpful and are appreciated.
I'm definitely looking for a good headphone amp, as I only have a cheap DAC/amp combo at the moment and I will be purchasing harder to drive headphones in the future. To me DIY will be my only option from now on when upgrading amplification and sources because I want to remove myself from the FOTM/politics bulldust that goes around on Head-Fi, and the financial side of it also helps
The Wire project seemed perfect to me because I am aiming for a "wire with gain" for an amp, so I was considering building an M^3 or a Dynalo when I realised I had missed out, of course after I had some soldering experience. If this project can perform as well as either of those amps and is easier to assemble it would be perfect for me, both as an upgrade and as my first foray into DIY. Regarding the ebay products you're building, if you could let me know that would be wonderful
You bought one! That's great. I'll be very interested to read your review. That TPA6120 may turn out to be a superior chip.
My NG98 took 3 weeks to get here (US) and almost everything else I've ordered from HK and China has taken 6-8 weeks. I think they get stuck in customs.
If you want to have a look at what the TPA6120 can do in a very robustly designed amp, you should try and get in on this group buy: Group buy: QRV09 Headphone amp - diyAudio. but you'll have to act fast. It's a great opportunity to get one of his boards relatively cheaply.
Thanks MrSlim. That's a very impressive Wiki and the photo of the board looks great. I knew about the similar Sjostrom/Peranders QRV07 but the QRV09 isn't on his website. So thanks again for pointing it out.
I'm currently not sure what direction to go. The QRV09 may fulfill my requirements for an ultra low distortion amp and save me lots of work. Like the "Wire", however, it's mostly SMT so it's not very DIY-friendly. I also can't find any pictures of the bottom side of the board showing how he handled mounting the TPA6120.
One thing I like about the "Wire", and the National reference board, are the impressive Audio Precision measurements. I haven't seen any solid measurements of the QRV09 but perhaps some exist? Are there measurements in the documentation package he sends via email? From what I gather, Peranders may only have a Turtle Beach soundcard to make measurements with. His QRV08 was measured by someone else (using a sound card) and the QRV07 hasn't been measured at all that I can find.
I've worked with enough audio projects to know you have to make the right measurements to verify a new design. And there are a lot of measurements that are difficult or impossible to do correctly with just a soundcard and software like RMAA. With a really good sound card/interface, enough extra equipment, software, test jigs/adapters, etc. you can work around many of the limitations, but not all of them. When you're talking about this level of performance, there really is no substitute for a real audio analyzer like an AP or Prism dScope.
I believe the lack of proper measurements is a big reason I've measured so many eBay products with lousy performance. They use all the right chips, etc. but they make some small mistake somewhere (which can be as simple as a poorly routed ground trace on the PCB) and the performance really suffers.
Even companies like NuForce apparently don't properly test their commercial products. They've been caught with their pants down with at least four different products where third party measurements revealed serious problems. And, in the case of a Stereophile review, NuForce admitted they hadn't made the measurements thinking their design was solid. I'm not saying Peranders made a design mistake, but I don't see any proof he didn't either.
I'll send him a message and see if I can get the documentation package. If there are sufficiently impressive valid measurements for the QRV09 it may work well for what I'm looking for if you ignore the SMT hassles and the need for "off board" connectors, volume control, etc. It certainly would save me a lot of time developing a new board so I'm willing to put up with the tedious soldering. I'll also be testing the NG98 with the TPA6120.
I'm currently not sure what direction to go. The QRV09 may fulfill my requirements for an ultra low distortion amp and save me lots of work. Like the "Wire", however, it's mostly SMT so it's not very DIY-friendly. I also can't find any pictures of the bottom side of the board showing how he handled mounting the TPA6120.
One thing I like about the "Wire", and the National reference board, are the impressive Audio Precision measurements. I haven't seen any solid measurements of the QRV09 but perhaps some exist? Are there measurements in the documentation package he sends via email? From what I gather, Peranders may only have a Turtle Beach soundcard to make measurements with. His QRV08 was measured by someone else (using a sound card) and the QRV07 hasn't been measured at all that I can find.
I've worked with enough audio projects to know you have to make the right measurements to verify a new design. And there are a lot of measurements that are difficult or impossible to do correctly with just a soundcard and software like RMAA. With a really good sound card/interface, enough extra equipment, software, test jigs/adapters, etc. you can work around many of the limitations, but not all of them. When you're talking about this level of performance, there really is no substitute for a real audio analyzer like an AP or Prism dScope.
I believe the lack of proper measurements is a big reason I've measured so many eBay products with lousy performance. They use all the right chips, etc. but they make some small mistake somewhere (which can be as simple as a poorly routed ground trace on the PCB) and the performance really suffers.
Even companies like NuForce apparently don't properly test their commercial products. They've been caught with their pants down with at least four different products where third party measurements revealed serious problems. And, in the case of a Stereophile review, NuForce admitted they hadn't made the measurements thinking their design was solid. I'm not saying Peranders made a design mistake, but I don't see any proof he didn't either.
I'll send him a message and see if I can get the documentation package. If there are sufficiently impressive valid measurements for the QRV09 it may work well for what I'm looking for if you ignore the SMT hassles and the need for "off board" connectors, volume control, etc. It certainly would save me a lot of time developing a new board so I'm willing to put up with the tedious soldering. I'll also be testing the NG98 with the TPA6120.
@JCX, Related to your multi-feedback approach, I'm curious if you saw this interesting DC coupled headphone amp design by Samuel Groner:
IC_headphone_amplifier_r1.pdf
The feedback, compensation, and volume adjustment are rather interesting. What do you think of the design--especially the volume control? He doesn't say much about it besides it's untested.
IC_headphone_amplifier_r1.pdf
The feedback, compensation, and volume adjustment are rather interesting. What do you think of the design--especially the volume control? He doesn't say much about it besides it's untested.
the QRV09 is a redesign of the QRV07, I think he just tweaked the board a bit. There doesn't seem to be any obvious differences between the schematics for either of them. I just reread your reply and looked at the web page and realized that the spec's are just that, not measurements. It's just a good time to buy because the group buy price is less than 1/3 what it would cost to buy one of the QRV07 boards from him.. Part of P-A's thing is over built power supplies, in this case using individual regulators for each channel and each rail and dual transformers.
I agree that SMD is more fiddly than through hole, but the more you do it, the better you get at it. If you consider the time it takes to bend the leads on through hole components,making sure they are oriented the right way(for those of us who are anal about that), and cutting the excess leads after, then I'm not sure it takes that much longer to do. I have also thought about the idea of doing the bulk of the SMD work using the Toaster Oven or Electric Fry Pan(I have one to try it with..) techniques, although it would probably be a good thing to try it on a less valuable board first..
I just had a flash... One of the things I find painful about SMD is holding the component in place (holding it down with the tweezer) and not moving it until the soldering is done on at least one end(if it is something like a capacitor). We need something like half of a tweezer that was spring loaded so that it could hold the component in place (a third hand hold down device...)
I agree that SMD is more fiddly than through hole, but the more you do it, the better you get at it. If you consider the time it takes to bend the leads on through hole components,making sure they are oriented the right way(for those of us who are anal about that), and cutting the excess leads after, then I'm not sure it takes that much longer to do. I have also thought about the idea of doing the bulk of the SMD work using the Toaster Oven or Electric Fry Pan(I have one to try it with..) techniques, although it would probably be a good thing to try it on a less valuable board first..
I just had a flash... One of the things I find painful about SMD is holding the component in place (holding it down with the tweezer) and not moving it until the soldering is done on at least one end(if it is something like a capacitor). We need something like half of a tweezer that was spring loaded so that it could hold the component in place (a third hand hold down device...)
Last edited:
@MrSlim Thanks. So far, I haven't found any solid measurements of the QRV07 either. If you know of some please let me know where to look? On his website he just lists the raw specs for the components, not the final design. But I did try to add my name to the PCB buy list for the QRV09. Hopefully I can still get one. So thanks again.
MrSlim, The CK2III is an appealing discrete amp. It really belongs in a different thread, but I'd be curious to know why you chose the CK2III over say these:
Gilmore Pure Class A
Dynalo
Balkishan
I'm still trying to decide between the CK2III, the above 3, and perhaps a couple of others to represent the discrete relatively low feedback design in my comparison.
Gilmore Pure Class A
Dynalo
Balkishan
I'm still trying to decide between the CK2III, the above 3, and perhaps a couple of others to represent the discrete relatively low feedback design in my comparison.
Sure, I wasn't trying to threadjack. To answer your question, I was being pragmatic about it. A friend over at Headfi bought 2x CK2III boards and matched TX's from someone else who changed their mind about building them and he only wanted one, so he sold me the other one. That's how I ended up with the CK2III vs a Dynalo(which I also considered) I've built a few hybrids ( 2 each of the SOHA I and CTH) and thought I should build a solid state amp too. I think the Dynalo is essentially a Gilmore Pure Class A with a couple of improvements(the circuits are almost identical). Ive never really heard of the Balkishan, and considering that I will probably sell the CK2III at some point, depending on how it compares to some of the other amp's I am building, its better to have something "known" in the community, vs something unknown. Since I have a finite number of amps I need, I have to plan to be able to sell one project to finance another as need be.. Since I'll have a number of amps in my hands at the same time, it will be interesting to do direct comparisons of them, particularly given their various designs. Also, I tend to lean towards the path of least resistance, given a choice of a design that I would have to breadboard, vs a design that has a PC board available for it, I'll choose the latter, just for ease of construction.
Last edited:
I've seen Groner's op amp distortion paper - the common mode distortion results are very interesting as they aren't given in datasheets
it is disappointing to see that the OPA827 doesn't substantially improve the 627's common mode distortion
Groner’s headphone amp could take some time to analyze - off hand I would only use discrete output Q for more I,V than op amps can provide, also for headphone use I wouldn't consider anything but Class A bias for a discrete Q output stage
Perander’s QRV09 isn’t multiloop so some of my earlier TPA6120 commentary won’t apply
it is disappointing to see that the OPA827 doesn't substantially improve the 627's common mode distortion
Groner’s headphone amp could take some time to analyze - off hand I would only use discrete output Q for more I,V than op amps can provide, also for headphone use I wouldn't consider anything but Class A bias for a discrete Q output stage
Perander’s QRV09 isn’t multiloop so some of my earlier TPA6120 commentary won’t apply
@JCX, thanks. I figured the QRV09 wasn't multi-loop from what I can see of the supposedly very similar QRV07 design. I just meant that, as far as ultra low distortion unbalanced designs go, it might be a ready-to-go proven solution without needing to spin a board from scratch. I'm not big on re-inventing the wheel when someone else has done a perfectly good job.
So if I wanted to use a TPA6120 in a multi-loop configuration what would you propose? You've done a more elaborate version with your design, so I assume you're thinking something simpler perhaps more like you posted here:
high-loop-gain-composite-op-amp-circuits
Which, in some ways, is a more elaborate version of this (which gets bonus points for making the most of dirt cheap parts):
MINI-HA1-Headphone-amplifier-kit-DIY
Speaking of the TPA6120, Peranders talks about "18 vias" in the link you provided (also below), and even captions a picture that way, but nothing in the picture looks like 18 vias under the TPA6120. Do you know what's going on there? Do you know where results have been posted for that board (QRV07) as a DIY project by various builders?
TPA6120A2 headphone amplifier power supply capacitance for the output THD specified - Audio Amplifiers Forum - Audio Amplifiers - TI E2E Community
I've emailed him for the documentation package for the QRV09 and have not heard back yet. It might be he's no longer accepting orders (the Wiki lists "early April" as the cutoff and he met his goal of 35 boards).
So if I wanted to use a TPA6120 in a multi-loop configuration what would you propose? You've done a more elaborate version with your design, so I assume you're thinking something simpler perhaps more like you posted here:
high-loop-gain-composite-op-amp-circuits
Which, in some ways, is a more elaborate version of this (which gets bonus points for making the most of dirt cheap parts):
MINI-HA1-Headphone-amplifier-kit-DIY
Speaking of the TPA6120, Peranders talks about "18 vias" in the link you provided (also below), and even captions a picture that way, but nothing in the picture looks like 18 vias under the TPA6120. Do you know what's going on there? Do you know where results have been posted for that board (QRV07) as a DIY project by various builders?
TPA6120A2 headphone amplifier power supply capacitance for the output THD specified - Audio Amplifiers Forum - Audio Amplifiers - TI E2E Community
I've emailed him for the documentation package for the QRV09 and have not heard back yet. It might be he's no longer accepting orders (the Wiki lists "early April" as the cutoff and he met his goal of 35 boards).
I was the one thread jacking
Thanks for the answer. I agree about the path of least resistance. It's so much easier, and usually more robust with better performance, to use a properly designed multi-layer PCB. Designing the PCB, however, is another matter. So it's a big help to find one that already (even mostly) meets your needs.
so much to post in reply, some really good conversation here and just a little hey nonny nonny hey
being that i have a bit of experience with headphones, have/had a penchant for paralleling buf634 in the past (the more the better i once thought) and in particular have built and love 'the wire', i can say with all honesty that imo paralleling these buffers is a bit of a waste of time, money and board space, i have jh13 which are 6 driver a side 28ohms nominal professional in ear stage monitors, if anything is going to need the extra current its these babies and i wouldnt punish your ears with them, let alone mine at anything even approaching 50% with my amp (fed with balanced source and set for unity gain).
the above will present less than 4r at some frequencies and as i mentioned in the thread, like this, with +/-12v rails, its as good as any single ended headamp i have heard with those and lineball with high quality balanced, with enough voltage swing in reserve to drive hd600 with the same gusto. as we know volume isnt everything, but i have more than enough options of my own as well as past experience with top flight commercial gear that drives them 'properly' to offer opinion (though of course that is all i'm offering)
i used the newer lme49610 in place of the 600 and added some of the newish small kemet 6.8uf 1206 size polymer tantalum caps soldered directly on top of the existing 1206 cap position, which i subbed 100nf np0 for the 10uf x7r for decoupling the rails. i cannot measure any offset at all past the lead resistance with my fluke, which puts it less than 1/10th of a mv, i would have to break out the HP3400+bench meter to do better than that, but why bother, 10mv is hardly insignificant with low z headphones though, so better aim higher...erm...lower than that.
anyway i'll have a run through other posts which has been a good read and i can only speak highly of the amp, do watch the draw on the power supply though, i underestimated it and ended up with some pretty toasty heatsinks on an Amb sigma22 with pretty standard 1.5-2" board mount heatsinks, (which i have used for powering 2 x 2 channel m^3s bridged) after losing myself with the jh13 for a while and then leaving it for a few minutes with no music playing (wide open though, i use sabres internal volume and no pot) perhaps because they are class AB rather than class A like the m^3, so with the amb the amps sinks took some of the load. your test as you go method will work here of course, though perhaps given the package, try and rig up some adapters or fudge it deadbug style to get an idea before going
opa1642 would be a nice match also, but then i like jfet inputs
actually looking back above, perhaps i have said enough good luck, though i must say if you dont hear some difference in these topologies there is something amiss, how are you planning on leveling the playing field as far as headphone impedance though? some amps will perform better with some headphones, whether you volume match them or not
being that i have a bit of experience with headphones, have/had a penchant for paralleling buf634 in the past (the more the better i once thought) and in particular have built and love 'the wire', i can say with all honesty that imo paralleling these buffers is a bit of a waste of time, money and board space, i have jh13 which are 6 driver a side 28ohms nominal professional in ear stage monitors, if anything is going to need the extra current its these babies and i wouldnt punish your ears with them, let alone mine at anything even approaching 50% with my amp (fed with balanced source and set for unity gain).
the above will present less than 4r at some frequencies and as i mentioned in the thread, like this, with +/-12v rails, its as good as any single ended headamp i have heard with those and lineball with high quality balanced, with enough voltage swing in reserve to drive hd600 with the same gusto. as we know volume isnt everything, but i have more than enough options of my own as well as past experience with top flight commercial gear that drives them 'properly' to offer opinion (though of course that is all i'm offering)
i used the newer lme49610 in place of the 600 and added some of the newish small kemet 6.8uf 1206 size polymer tantalum caps soldered directly on top of the existing 1206 cap position, which i subbed 100nf np0 for the 10uf x7r for decoupling the rails. i cannot measure any offset at all past the lead resistance with my fluke, which puts it less than 1/10th of a mv, i would have to break out the HP3400+bench meter to do better than that, but why bother, 10mv is hardly insignificant with low z headphones though, so better aim higher...erm...lower than that.
anyway i'll have a run through other posts which has been a good read and i can only speak highly of the amp, do watch the draw on the power supply though, i underestimated it and ended up with some pretty toasty heatsinks on an Amb sigma22 with pretty standard 1.5-2" board mount heatsinks, (which i have used for powering 2 x 2 channel m^3s bridged) after losing myself with the jh13 for a while and then leaving it for a few minutes with no music playing (wide open though, i use sabres internal volume and no pot) perhaps because they are class AB rather than class A like the m^3, so with the amb the amps sinks took some of the load. your test as you go method will work here of course, though perhaps given the package, try and rig up some adapters or fudge it deadbug style to get an idea before going
opa1642 would be a nice match also, but then i like jfet inputs
actually looking back above, perhaps i have said enough
good luck, though i must say if you dont hear some difference in these topologies there is something amiss, how are you planning on leveling the playing field as far as headphone impedance though? some amps will perform better with some headphones, whether you volume match them or not
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.