My first post in this forum.
You know, i read posts here for a long time and i appreciate the knowledge of some of the members.
Regarding the post title, i have just COMPLETED my headphone amp based on the tremendous tpa6120a2 chip from TI.
This is my second try, on a more effective layout with care on positioning components, as my first try ended up with lot of oscillation, excessive dc offset and many other problems.
Currently, my built amp is truly monoblock, with one chip per channel with activated only the left channel.
Gain is 3V/V with 0.1 % ONEOHM resistor on the feedback loop, while topology is non-inverting. No series DC blocking capacitor in the signal path.
Power supply is also monoblock, with two cascaded 317 / 337 regulators, two 10000 ìF AEROVOX in the main bank and a 160 VA talema trafo, all these per channel.
This overkill PSU was initially built for my 627 based preamp, but i included the option of a PSU out for future try outs, as happened!
The amp board was installed into an empty box previously housing the circuitry of black box phono stage. The connection with the pre's PSU is via a mogami 2534 run.
Regarding the chip's power pad, i ' ve read a lot for the difficulty of soldering to the ground plane. This was not a problem. I installed the chip flipped over, cut all 15 pins (out of 20), cut a thin layer of copper, fixing it on the board's pad and soldering the exposed power pad on the copper. Perfect solution for me!
Total cost? 0 Euros (!!!), as the chips were TI samples, while the rest of components were in stock from other DIY projects.
The amp is driven from a passive controller based on a hand-selected black beauty ALPS 10kohm.
Currently, offset is stable at 4-5 mV (always) and no oscillations are observed at any case (may be the better component layout?).
For history, i have made lots of headamps, with any kind of configuration, specially with opamps. Never tried tubes, but in the future this is possible.
OK, what about the sound? My headphones is hd-650 Senns (300 ohm). The negative first (has only ONE !): a slight lack of "weight" in the lowest registers, compared to the direct drive from my monoblock 3886 GC.
Now the positives: NEVER have i listened such a natural sound from a headamp. Voices, three-dimensionality, bass definition, highs, harmonical structure are IMHO first class.
The most characteristic feature is the desire to listen to music disk-after-disk, concentrating on the songs theirselves without wanting to criticize anything (and believe me, i am criticizing everything!).
For the end, this amp is a perfect line preamp, surpassing also my 627 preamp especially in the dynamics field and in the naturalness!
I'll give more details on another post.
Nice to meet you all.
Steeve,
Athens, Greece.
You know, i read posts here for a long time and i appreciate the knowledge of some of the members.
Regarding the post title, i have just COMPLETED my headphone amp based on the tremendous tpa6120a2 chip from TI.
This is my second try, on a more effective layout with care on positioning components, as my first try ended up with lot of oscillation, excessive dc offset and many other problems.
Currently, my built amp is truly monoblock, with one chip per channel with activated only the left channel.
Gain is 3V/V with 0.1 % ONEOHM resistor on the feedback loop, while topology is non-inverting. No series DC blocking capacitor in the signal path.
Power supply is also monoblock, with two cascaded 317 / 337 regulators, two 10000 ìF AEROVOX in the main bank and a 160 VA talema trafo, all these per channel.
This overkill PSU was initially built for my 627 based preamp, but i included the option of a PSU out for future try outs, as happened!
The amp board was installed into an empty box previously housing the circuitry of black box phono stage. The connection with the pre's PSU is via a mogami 2534 run.
Regarding the chip's power pad, i ' ve read a lot for the difficulty of soldering to the ground plane. This was not a problem. I installed the chip flipped over, cut all 15 pins (out of 20), cut a thin layer of copper, fixing it on the board's pad and soldering the exposed power pad on the copper. Perfect solution for me!
Total cost? 0 Euros (!!!), as the chips were TI samples, while the rest of components were in stock from other DIY projects.
The amp is driven from a passive controller based on a hand-selected black beauty ALPS 10kohm.
Currently, offset is stable at 4-5 mV (always) and no oscillations are observed at any case (may be the better component layout?).
For history, i have made lots of headamps, with any kind of configuration, specially with opamps. Never tried tubes, but in the future this is possible.
OK, what about the sound? My headphones is hd-650 Senns (300 ohm). The negative first (has only ONE !): a slight lack of "weight" in the lowest registers, compared to the direct drive from my monoblock 3886 GC.
Now the positives: NEVER have i listened such a natural sound from a headamp. Voices, three-dimensionality, bass definition, highs, harmonical structure are IMHO first class.
The most characteristic feature is the desire to listen to music disk-after-disk, concentrating on the songs theirselves without wanting to criticize anything (and believe me, i am criticizing everything!).
For the end, this amp is a perfect line preamp, surpassing also my 627 preamp especially in the dynamics field and in the naturalness!
I'll give more details on another post.
Nice to meet you all.
Steeve,
Athens, Greece.
the tpa6120 is a nice chip to use in a headphone amp, i done a circuit board for one and got a few boards manufactured up for it
http://www.mypage.tsn.cc/dakiller/tpapages/
and some finished pics - http://www.mypage.tsn.cc/dakiller/dtpablack/
i really do enjoy the sound, much better than the opamps with buf634 inside the feedback loop that i've done in the past
http://www.mypage.tsn.cc/dakiller/tpapages/
and some finished pics - http://www.mypage.tsn.cc/dakiller/dtpablack/
i really do enjoy the sound, much better than the opamps with buf634 inside the feedback loop that i've done in the past
I now enjoy the sound from my TPA6120 amp!
It has a very natural sound and is NEVER boring .
The amp is configured as a passive one as i told you.
The only problem i encounter is a slight increase in output offset (from around 5 mV to 30 mV) depending on the pot's position.
This from what i know is not really important.
In the first version of the amp thigs were way worst. I had audible oscillations even when moving around. Some times, even when touching the headphone cord, i had hum problems.
Now fortunately these problems disappeared, but what annoys me is the variations of the output offset.
Can anyone elaborate for the physical explanation of what really annoys this chip, independently of the layout?
Anyway, the sound of this amp is something someone must experience to estimate. Because of this, i want to end up this project with a perfect amp without any problem (i.e. no offesets, no oscillations, no nothing).
I have to mention that i have no caps in the signal path, as i believe that a cap is unnecessary.
For the end, i have to mention that the amp was very easy to build. I'd like to hear opinions from other builders (concerning sound impressions).
It has a very natural sound and is NEVER boring .
The amp is configured as a passive one as i told you.
The only problem i encounter is a slight increase in output offset (from around 5 mV to 30 mV) depending on the pot's position.
This from what i know is not really important.
In the first version of the amp thigs were way worst. I had audible oscillations even when moving around. Some times, even when touching the headphone cord, i had hum problems.
Now fortunately these problems disappeared, but what annoys me is the variations of the output offset.
Can anyone elaborate for the physical explanation of what really annoys this chip, independently of the layout?
Anyway, the sound of this amp is something someone must experience to estimate. Because of this, i want to end up this project with a perfect amp without any problem (i.e. no offesets, no oscillations, no nothing).
I have to mention that i have no caps in the signal path, as i believe that a cap is unnecessary.
For the end, i have to mention that the amp was very easy to build. I'd like to hear opinions from other builders (concerning sound impressions).
P-A,
soon will upload photos !!! I have to re-loan the camera.
Regarding musicality, i have never listened to a more musical headamp.
Especially the high extension is extension-less and very very pure !!
Only problem the slight lack of weight in the lo registers. But the others,
Will keep you all informed. Greetings
soon will upload photos !!! I have to re-loan the camera.
Regarding musicality, i have never listened to a more musical headamp.
Especially the high extension is extension-less and very very pure !!
Only problem the slight lack of weight in the lo registers. But the others,
Will keep you all informed. Greetings
This article studied the THS6012, tpa6120's elder brother:
http://sound.westhost.com/highspeed.htm
Some key words in this article:
"I could not measure the distortion.
Quite frankly, I was disappointed. There was absolutely nothing that sounded of 'amplifier' - this is probably as close to a straight wire with gain as you are likely to find."
While some people like such hi-fi sound, more people like musical sound - some color from an amp such as a tube amp.
http://sound.westhost.com/highspeed.htm
Some key words in this article:
"I could not measure the distortion.
Quite frankly, I was disappointed. There was absolutely nothing that sounded of 'amplifier' - this is probably as close to a straight wire with gain as you are likely to find."
While some people like such hi-fi sound, more people like musical sound - some color from an amp such as a tube amp.
I need to limit the (large) bandwidth of the TPA headamp, as my dac (CS4390-based) has minimal 1st order analog filter.
Due to ultra-sonic content at the output of the dac, the TPA oscillates (DC offset, excessive warming, etc.) at specific gain settings of the passive preamp (10k log).
Does anybody knows how to implement a filter at the input of the TPA for cutting the CS4390’s residual 5.6 MHz signal (= 128 x Fs) that so much annoys the TPA’s performance ?
Thanks.
Due to ultra-sonic content at the output of the dac, the TPA oscillates (DC offset, excessive warming, etc.) at specific gain settings of the passive preamp (10k log).
Does anybody knows how to implement a filter at the input of the TPA for cutting the CS4390’s residual 5.6 MHz signal (= 128 x Fs) that so much annoys the TPA’s performance ?
Thanks.
P-A,
I'm sure for the oscillations because only this source (dac) presents this kind of problem. With the remaining (phono, dvd player and analog out from my cd player) no problem at all.
The dac is a modified one but presents a 5.6 MHz output at -42 dB, as defined by the 1st order analog filter. So no component or pcb layout problems here.
This that bothers me is the extremely wide BW of TPA that may be triggered by HF signal even at moderate levels.
Sorry, no photos at the moment. Soon, will upload photos of the pcb when available.
Thanks.
I'm sure for the oscillations because only this source (dac) presents this kind of problem. With the remaining (phono, dvd player and analog out from my cd player) no problem at all.
The dac is a modified one but presents a 5.6 MHz output at -42 dB, as defined by the 1st order analog filter. So no component or pcb layout problems here.
This that bothers me is the extremely wide BW of TPA that may be triggered by HF signal even at moderate levels.
Sorry, no photos at the moment. Soon, will upload photos of the pcb when available.
Thanks.
Topology is non-inverting.
I used 2k (feedback) and 1k for neg. input to GND.
No SMD parts. The feedback resistor is of highest quality (see my first post).
PCB bank is very close to pins and is red ELNA starget 100 uF in parallel with WIMA 100nF MKS (a pair per leg and per channel).
The analog filter has -3dB @ 48 kHz.
By setting a different cutoff (e.g. 24 kHz) the level alters only -6dB lower, thus no real gain is obtained.
I used 2k (feedback) and 1k for neg. input to GND.
No SMD parts. The feedback resistor is of highest quality (see my first post).
PCB bank is very close to pins and is red ELNA starget 100 uF in parallel with WIMA 100nF MKS (a pair per leg and per channel).
The analog filter has -3dB @ 48 kHz.
By setting a different cutoff (e.g. 24 kHz) the level alters only -6dB lower, thus no real gain is obtained.
You can see photo of the internal of my amp at:
http://www.avforum.gr/showarticles.php?topic=11&subtopic=23782
http://www.avforum.gr/showarticles.php?topic=11&subtopic=23782
When I see the pictures I think I see the problem. Texas recommends SMD parts, groundplane and a real tight design. As you may know the feedback resistor is extremely sensitive when it comes to current feedback and this kind of speed, 1300 V/us! I'll suspect you never can get this work properly with your building style.
I'll supect that the design must look like this or similar. I think you shouldn't underestimate the importance of a good pcb especially when you are using this ultrafast IC. Your input wires are good antennas.
I'll supect that the design must look like this or similar. I think you shouldn't underestimate the importance of a good pcb especially when you are using this ultrafast IC. Your input wires are good antennas.
Minimaist approch, in a player...
I put this in my Sony dvp-ns900v.
Direct to the akm4383.
I pulled out all the chips on the analog board to free up power.
I used an old Tamradio 2400 ohm to 600 ohm xformer.
The sound is so good.
But the offset V comes close to 100mv as I get the linear pot about half way, loud! Over 200mv full turn.
The gain is r2/r1 + 1.
So 10.
The second image I got off the net.
It shows a gain of 100.
I am tring to get an offset I can live with and still not use an input chip or buffer.
I want to next try another Tamradio that is 1 to 1 ( 600 ohms in and out) also CT in, SE out. Thereby reducing the gain needed.
It seems the offset is related to the gain.
I put this in my Sony dvp-ns900v.
Direct to the akm4383.
I pulled out all the chips on the analog board to free up power.
I used an old Tamradio 2400 ohm to 600 ohm xformer.
The sound is so good.
But the offset V comes close to 100mv as I get the linear pot about half way, loud! Over 200mv full turn.
The gain is r2/r1 + 1.
So 10.
The second image I got off the net.
It shows a gain of 100.
I am tring to get an offset I can live with and still not use an input chip or buffer.
I want to next try another Tamradio that is 1 to 1 ( 600 ohms in and out) also CT in, SE out. Thereby reducing the gain needed.
It seems the offset is related to the gain.
Attachments
Sorry for beeing a little bit o.T. forgive me
I´ve built an amp with the THS6012 a while ago, it sounded nothing than superb... Last time surfing on the Analog hompage I read about the AD8397, a high current opamp with quite good but not that impressive spec´s as the Ti THS-type.
Anyway, I put it in the output/filter stage of my EMU0404 - as a line driver it can´t be bad I thought.
But: When I drive my K271s (and those AKG´s are known for their high demands on amplification to reproduce good low freq.)directly from the card, bypassing the THS, I can hear absolutely no difference - even with brain-damaging levels...
Another thing: I made also good experience with an THS4222 in my Prodigy7.1 (MSOP) also a impressive improvement over standard opamps...
I´ve built an amp with the THS6012 a while ago, it sounded nothing than superb... Last time surfing on the Analog hompage I read about the AD8397, a high current opamp with quite good but not that impressive spec´s as the Ti THS-type.
Anyway, I put it in the output/filter stage of my EMU0404 - as a line driver it can´t be bad I thought.
But: When I drive my K271s (and those AKG´s are known for their high demands on amplification to reproduce good low freq.)directly from the card, bypassing the THS, I can hear absolutely no difference - even with brain-damaging levels...
Another thing: I made also good experience with an THS4222 in my Prodigy7.1 (MSOP) also a impressive improvement over standard opamps...
neutron7:
I ground down the unused top and bottom end with a dremel tool. This was to get access to the ground connections that go to the power pad. These are visable on each end of the chip but they are reccesed.
I needed to do this because I epoxied the chip to a heat sink. I exposed the ground leads and made them the centre-point on me star ground.
I doing so I saw that the NC leads all run to the die but are not connected at that point. SO If you did ground the nc leads I think you would cause more net capacitance around the die and the touchie neg input lead.
What are you tring to do with the tpa?
BTW I assume this is a relabled ths6012 and it's appl data would apply. Can anyone can any one confirm this so I don't turn into something worse than a pumpkin.
Can anyone say (help) as to the best, lowest values for r1, r2 to get the offset lower. Im using 1k and 10k now. maybe 100 and 1k (gain of 11).
ALSO could I use a lower value pot? I am going to a 1 to 1 trans. so I could lower the gain to maybe 5.
I ground down the unused top and bottom end with a dremel tool. This was to get access to the ground connections that go to the power pad. These are visable on each end of the chip but they are reccesed.
I needed to do this because I epoxied the chip to a heat sink. I exposed the ground leads and made them the centre-point on me star ground.
I doing so I saw that the NC leads all run to the die but are not connected at that point. SO If you did ground the nc leads I think you would cause more net capacitance around the die and the touchie neg input lead.
What are you tring to do with the tpa?
BTW I assume this is a relabled ths6012 and it's appl data would apply. Can anyone can any one confirm this so I don't turn into something worse than a pumpkin.
Can anyone say (help) as to the best, lowest values for r1, r2 to get the offset lower. Im using 1k and 10k now. maybe 100 and 1k (gain of 11).
ALSO could I use a lower value pot? I am going to a 1 to 1 trans. so I could lower the gain to maybe 5.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.