Hi guys,
I have a board here with a NJM4556 based headphone output. Crystal clear in the top, rather wimpy in the bass. The OPA2134 doesn't have enough power.
Are there other good op-amps for headphones? Preferably dual packages, SOIC.
I'm using DC coupling and negative gain with two amps (one chip) for each channel. Paralelled with 12R resistors. Perhaps changing the resistor will also change the oompfh. The power supply is +-5.8V.
Cheers,
Børge
I have a board here with a NJM4556 based headphone output. Crystal clear in the top, rather wimpy in the bass. The OPA2134 doesn't have enough power.
Are there other good op-amps for headphones? Preferably dual packages, SOIC.
I'm using DC coupling and negative gain with two amps (one chip) for each channel. Paralelled with 12R resistors. Perhaps changing the resistor will also change the oompfh. The power supply is +-5.8V.
Cheers,
Børge
I'm driving AKG K701, Sennheiser HD600 and Shure SRH940 at the moment. I wouldn't mind if the amp could work for much smaller drivers as well. I haven't yet tried it on Porta-Pro or in-ear stuff.
Over at Analog they suggested I should check out the AD8397. Samples are underway. Anybody here with experience on that one?
Thanks,
Børge
Over at Analog they suggested I should check out the AD8397. Samples are underway. Anybody here with experience on that one?
Thanks,
Børge
The HD600 is a high impedance model. It isn't too bad sounding on the NJM4556, actually. The lack of oompfh is there on both 32 and 600 ohm cans.
I'm aware of the voltage limitation. I can beef it up a little bit but not much more than 1V or so on either rail before the rest of the circuit becomes tricky to build.
Børge
I'm aware of the voltage limitation. I can beef it up a little bit but not much more than 1V or so on either rail before the rest of the circuit becomes tricky to build.
Børge
AD8397 is great, just be sure to put 2-3 Ohm output resitor, wirewound 5W should keep the op-amp safe from shorting output to the ground. If you're prototyping - be sure to apply power at the same time, if your ground connection fails - AD8397 will burn, JRC4556 is virtually indestructible compared to AD8397. Be sure to use 0.1uF C0G + 10uF tantal caps for decoupling power pins and You will be fine. Great chip for headphone amp.
OPA1688
Take a look at TI's new OPA1688 which is specifically designed to be a headphone driver. I have an amp built with the chip and it sounds *great*. Adydula here on the forum has also been testing it for the last week or two.
The chip is a dual SOIC-8, no output resistor required for up to 500pF of drive load (headphones and cables, see fig 25 & 26 in the datasheet), does include short circuit protection like the NJM4556A for TRS plugs in/out, just 2mA per amplifier of idle current, +/-2.25V to +/-18V (=36V single ended), only 250 micoamps of offset voltage for low turn-on/off pops, FET input so no large IR drop across input resistors to reflect to additional output offset, and no metal pad on the bottom to have to solder.
Apparently TI figures the low Iq results in less idle dissipation and therefore no need for the metal powerpad. By way of comparison the NJM4556A has the same 75mA per channel (datasheet says 70 but I can say from measurements it is 80mA), but a whopping 9-12mA of quiescent currentper amplifier(= x2). At +/-15Vdc rails that equates to over half of the package's rated dissipation (15 - (-15)) * 11mA * 2 amplfiiers = 660mW!) taken up just with idle current.
To me the NJM4556A (in the O2 headamp) has always sounded wimpy on bass. I have never come up with an electrical reason for it. From the datasheet specs and from the measurements NwAvGuy did the bass should sound just great. But my ears have always said otherwise. I can say from the testing so far no such problem with the OPA1688 IMHO.
Thread on the forum here:
http://www.diyaudio.com/forums/vendors-bazaar/280113-new-audio-op-amp-opa1688.html
and datasheet:
http://www.ti.com/general/docs/lit/getliterature.tsp?genericPartNumber=opa1688&fileType=pdf (opens PDF)
One of my last posts in that thread notes that Mouser has them for just $1.59 a chip at the moment! The pinout is the same as the NJM4556A. If you use a DIP to SOIC8 adaptor it should just drop right in if you are using it as a unity gain buffer (as in the O2 headamp). If you are using it with voltage gain note that the compensation cap is lower than most in the datasheet app circuits at 47pF. Johnc124 at TI said he did that for stability with a wide range of headphone capacitive loads, so you might have to reduce an existing 150pF or 220pF from some other chip if substituting into a gain circuit.
As for paralleling I've been experimenting with that too.
No troubles using a 1R output current balancing resistor on each amp half since the output offset voltage is so low, both inherently and the low input IR drop being FET, even multiplied by any voltage gain. So the balancing currents won't be large even at 1R, and in fact 0.5R on each amp half would probably be fine. For both sections paralleled that would give you 0.25R output Z.
TI also has a Spice model available for the chip that plugs right into LT Spice.
Take a look at TI's new OPA1688 which is specifically designed to be a headphone driver. I have an amp built with the chip and it sounds *great*. Adydula here on the forum has also been testing it for the last week or two.
The chip is a dual SOIC-8, no output resistor required for up to 500pF of drive load (headphones and cables, see fig 25 & 26 in the datasheet), does include short circuit protection like the NJM4556A for TRS plugs in/out, just 2mA per amplifier of idle current, +/-2.25V to +/-18V (=36V single ended), only 250 micoamps of offset voltage for low turn-on/off pops, FET input so no large IR drop across input resistors to reflect to additional output offset, and no metal pad on the bottom to have to solder.
Apparently TI figures the low Iq results in less idle dissipation and therefore no need for the metal powerpad. By way of comparison the NJM4556A has the same 75mA per channel (datasheet says 70 but I can say from measurements it is 80mA), but a whopping 9-12mA of quiescent currentper amplifier(= x2). At +/-15Vdc rails that equates to over half of the package's rated dissipation (15 - (-15)) * 11mA * 2 amplfiiers = 660mW!) taken up just with idle current.
To me the NJM4556A (in the O2 headamp) has always sounded wimpy on bass. I have never come up with an electrical reason for it. From the datasheet specs and from the measurements NwAvGuy did the bass should sound just great. But my ears have always said otherwise. I can say from the testing so far no such problem with the OPA1688 IMHO.
Thread on the forum here:
http://www.diyaudio.com/forums/vendors-bazaar/280113-new-audio-op-amp-opa1688.html
and datasheet:
http://www.ti.com/general/docs/lit/getliterature.tsp?genericPartNumber=opa1688&fileType=pdf (opens PDF)
One of my last posts in that thread notes that Mouser has them for just $1.59 a chip at the moment! The pinout is the same as the NJM4556A. If you use a DIP to SOIC8 adaptor it should just drop right in if you are using it as a unity gain buffer (as in the O2 headamp). If you are using it with voltage gain note that the compensation cap is lower than most in the datasheet app circuits at 47pF. Johnc124 at TI said he did that for stability with a wide range of headphone capacitive loads, so you might have to reduce an existing 150pF or 220pF from some other chip if substituting into a gain circuit.
As for paralleling I've been experimenting with that too.
No troubles using a 1R output current balancing resistor on each amp half since the output offset voltage is so low, both inherently and the low input IR drop being FET, even multiplied by any voltage gain. So the balancing currents won't be large even at 1R, and in fact 0.5R on each amp half would probably be fine. For both sections paralleled that would give you 0.25R output Z.TI also has a Spice model available for the chip that plugs right into LT Spice.
Last edited:
The TI TPA6120A2 is about the best chip for driving headphones. It will run on your supply but it would be great to give it a few more volts for your high impedance phones.
Cheers
William Cowan
I built AGDR's version of the cMoy or what i call the super cMoy! with the 1688 chip.
I like many of you have gone thru numerous headphones and amps. The battery powered TI 1688 is nothing but superlative.
I drive high and low impedance cans and its amazing how well it works even compared to other amps.
I have a pair of lowly Audio Technica ATH M50's I just refurbed and they sound absolutley stellar .....for such a low cost you get a really great little amp with this new chip..
Alex
I like many of you have gone thru numerous headphones and amps. The battery powered TI 1688 is nothing but superlative.
I drive high and low impedance cans and its amazing how well it works even compared to other amps.
I have a pair of lowly Audio Technica ATH M50's I just refurbed and they sound absolutley stellar .....for such a low cost you get a really great little amp with this new chip..
Alex
I'll throw OPA1622 into the mix for a step up in performance from the OPA1688 and TPA6120A2, but unfortunately it's not available in an SOIC package.
Both OPA1622 and OPA1688 include thermal shutdown circuitry and are specified to handle a continuous short circuit condition (this is stated in the abs max table in the datasheet). If you short the output and the die temp exceeds about 150C, the part shuts down until the die temp drops below a lower trip point (15 to 20C of hysteresis) and then it turns back on. While we were developing the OPA1622 I left the part with a continuous short over night, taking THD measurements before and after to determine if there was degradation in performance.
Both OPA1622 and OPA1688 include thermal shutdown circuitry and are specified to handle a continuous short circuit condition (this is stated in the abs max table in the datasheet). If you short the output and the die temp exceeds about 150C, the part shuts down until the die temp drops below a lower trip point (15 to 20C of hysteresis) and then it turns back on. While we were developing the OPA1622 I left the part with a continuous short over night, taking THD measurements before and after to determine if there was degradation in performance.
Those supply voltages may be a bit low for that NJM4556 as well. You might try upping the supply voltages first and listening to see if the sound quality improved. The power supply current of older op amps was a strong function of power supply voltage, and therefore by extension so was the open loop gain and distortion. My measurements on the 4556 showed this effect.
Fun fact, you can use this to "dial-in" the amount of distortion you want on some op amps. For example I played with this for awhile with the OPA2604.
Fun fact, you can use this to "dial-in" the amount of distortion you want on some op amps. For example I played with this for awhile with the OPA2604.
OPA1612 makes a decent headphone amplifier. Its output power is slightly less than the OPA1688, and capacitive load capability is also just a tad less. On the flip side, it has substantially lower input voltage noise than the OPA1688 (1.1nV/rtHz vs 7-8nV/rtHz) but still has a reasonable supply current of 3.6mA. It has been selected for a large number of headphone amplifier applications, including "hi-fi" cell phones in Asia.
A few tips:
* The TPA6120A2 requires 10 ohms on the output to prevent it from oscillating. I know a few folks here have experimented with inductors and/or a Zobel network to replace the resistor, but I can't remember how sucessful those efforts were. I know jcx uses that chip in some projects, he would have good info about it.
* The OPA1622 has 1uA of input bias current - no input bias current cancellation, which is probably a good thing for linearity. Not a problem if the input is balanced, like the DAC circuit in the app note. Otherwise you may get a bit higher DC output offset due to the input IR drops times stage voltage gain.
* The AD8397 - the "+/-12Vdc" listed as the maximum operating voltage should be taken lightly.
I'm using it in this projecthttp://www.diyaudio.com/forums/equipment-tools/284202-fliege-1khz-notch-filter-project-pcb.html
and at +/-12Vdc rails ran into some of the same things that AMB and Morsel reported way back when on the Headwize forum (AMB uses the chip in the mini^3 headamp) of instability and overheating. Lowering the voltage solves the problem so you should be OK with your +/-5.9V rails. I'm now using +/-9Vdc and it works like it should. AMB just uses 9Vdc with a virtual ground in the mini^3. Plus as mentioned the chip has no short circuit protection. AMB advises to turn the volume control all the way down when inserting a TRS headphone plug.
Fun with chips!
Last edited:
A few tips:
* The TPA6120A2 requires 10 ohms on the output to prevent it from oscillating. I know a few folks here have experimented with inductors and/or a Zobel network to replace the resistor, but I can't remember how sucessful those efforts were. I know jcx uses that chip in some projects, he would have good info about it.
* The OPA1622 has 1uA of input bias current - no input bias current cancellation, which is probably a good thing for linearity. Not a problem if the input is balanced, like the DAC circuit in the app note. Otherwise you may get a bit higher DC output offset due to the input IR drops times stage voltage gain.
* The AD8397 - the "+/-12Vdc" listed as the maximum operating voltage should be taken lightly.
http://www.diyaudio.com/forums/equipment-tools/284202-fliege-1khz-notch-filter-project-pcb.html
and at +/-12Vdc rails ran into some of the same things that AMB and Morsel reported way back when on the Headwize forum (AMB uses the chip in the mini^3 headamp) of instability and overheating. Lowering the voltage solves the problem so you should be OK with your +/-5.9V rails. I'm now using +/-9Vdc and it works like it should. AMB just uses 9Vdc with a virtual ground in the mini^3. Plus as mentioned the chip has no short circuit protection. AMB advises to turn the volume control all the way down when inserting a TRS headphone plug.
Fun with chips!
My AD8397 project is alive for 2 years now and I don't have any problems with supply up to +/-13.5V. No overheating at all, cold chip. I'm using one channel per package, the other one is terminated.
OPA1688 installed. First impression on K701: no more wimpy bass!
I'm going to give it a couple days of burn-in and then compare it to the original NJM4556. Then my plan is to do knock-out testing with OPA1688, OPA1612, OPA1622 and AD8397.
My circuit is a little short on supply rails, so good output swing is important. After deciding on which amp I'll test the same thing in positive and negative gain configuration. My board allows for both.
Thanks for your input guys!
Cheers,
Børge
I'm going to give it a couple days of burn-in and then compare it to the original NJM4556. Then my plan is to do knock-out testing with OPA1688, OPA1612, OPA1622 and AD8397.
My circuit is a little short on supply rails, so good output swing is important. After deciding on which amp I'll test the same thing in positive and negative gain configuration. My board allows for both.
Thanks for your input guys!
Cheers,
Børge
Thanks for the listening confirmation! Sure sounds that way to me.
Looks like the NJM4556A is a chip that "measures well but sounds bad", at least on the bass end of things. I saw that quote attributed to Nelson Pass once about his experiences measuring and listening to amps. Yet another plus about the OPA1688, it is listed as RRO.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.