An inverting version of NwAvGuy's O2 Headphone Amplifier. B5-080 case with 1/4" and RCA jacks. +/-15.3Vdc power rails, input select switch for ODAC. Many upgrades and improvements over the orginal O2 headphone amplifier.
No new version of NwAvGuy's famous O2 Headphone Amplifier has been introduced since he disapeared in the fall of 2012. agdr Audio has created a brand new version of NwAvGuy's O2 Headamp that contains several significant new designs. The new "O2" fits in the B5-080 case, which is slightly taller and an inch wider than the O2's B2-080 case. The new amplifier is nearly all easy-to-solder through hole parts, similar to the O2, except for a half dozen bypass capacitors under the PC board.
The original NwAvGuy O2 Headphone Amplifier used non-inverting op-amp stages for the gain and output sections. See a writeup here, with the non-inverting writeup below it. Inverting amplifiers lower common mode distortion and harmonics by keeping the voltage levels at the two inputs at zero volts with respect to ground. The voltage level at the input of a non-inverting stage varies with the incoming signal. See a good discusion of the inverting op-amp configuration vs. sound quality here.
The original NwAvGuy O2 Headphone Amplifier also used conventional (not low dropout) fixed (12V) voltage regulators that were not heat sinked. Low dropout (LDO) voltage regulators are used here to prevent going into dropout under heavy loads, a known design problem with the original O2 headamp on 12Vac. They are heat sinked to the case rear panel prevent the overheating the original O2 experiences with transformers over 12Vac. The new regulators (LT3015 and LT3080) are lower noise than the O2's 7812 and 7912 regulators.
Part of the reason no new versions of NwAvGuy's O2 Headphone Amplifier have been designed is (confusion over) the "no derivatives" license that NwAvGuy invoked with his O2, which essentially says derivatives of the design have to be approved by him. But... he has disappeared and is non-contactable for license releases. I would love to be able to contact NwAvGuy for his measurements and design approval. In several countries that issue alone of a license holder (or their representative) being non-contactable for over a year will cause a license to revert to the public domain.
For a design to pass the test as a "derivative" it has to be extremely similar to the original. Nearly the same parts, same (or nearly the same) layout, etc. The amplifier presented here is significantly different to the point where we believe it is not considered a derivative, but a new design. The part that is similar is the overall format of gain stage, followed by volume pot in the middle, followed by output stage. This design uses the slightly wider B5-080 case, with two "9V" NiMH batteriues
The new agdr Audio version of an Inverting O2 Headphone Amplifier has several benefits over the original NwAvGuy O2 including:
- Inverting gain and output stages for harmonic cancellation, vs. the non-inverting stages in the O2. The final output signal of the new amplifier is not inverted due to two inversions along the way from the two stages. Inverting stages keep the op amp inputs at zero volts with respect to ground, eliminating common-mode problems.
- Uses a FET-input LME49880 chip for true "tube sound", along with the lower harmonics of inverting gain and output stages. The original O2 used bipolar input chips and non-inverting stages.
- Double the peak output current capability at 250A per channel using heat-sinked TO-220 BUF634 chips. The NJM4556A output chips in the orginal NwAvGuy O2 Headamp are not heatsinked and have been known to overheat and crack in half in the special case of low impedance + low sensitivty headphones at high volume levels and with music that has sustained peaks. No such problem with this new amp.
- A circuit design trick of using an intermediate value of bandwidth set resistor (rather than leaving the pin floating or tying it to Vee) is used with the BUF 634 chips. The resistor extends the bandwidth just enough to elminate any chance of oscillation in the loop with the OPA2140, but still keep quiescent current to a low value to reduce battery drain.
- No output balancing resistors requiired for true zero ohm output impedance. The O2 requires 1 ohm output resistors (also called ballast resistors) on each paralleled NJM4556A section to force equal current sharing, resulting in 0.5 ohm output resistance. This new amplifier uses just the one output buffer, nothing paralleled, so no balancing resistors required. The loop stability has been tested to insure no oscillation problems with zero ohm headphone drive.
- NwAvGuy's ODAC DAC still fits in the battery area if the batteries are removed, similar to the O2. Both the original ODAC and the new rev B fit just fine.
- A front panel input select switch selects between an internal ODAC (or the rear panel RCA jack used for input) or the front panel 3.5mm jacks. An audio source can be left plugged into the 3.5mm jack all the time, along with the ODAC input.
- Headphone relay for absolutely no turn-on or turn-off thumps. NwAvGuy's O2 used part of the power management circuit for this, which was only partially successful.
- Super quiet low-dropout power supply with the voltage regulators heat sinked to the case
- Higher voltage +/-15.5Vdc power supply vs. the +/-12.0Vdc in the original O2 headamp while on AC power. Still an average of +/-8.4Vdc while on batteries. The extra 0.5V (15.5Vdc vs 15.0Vdc) compensates for the diode logic voltage drop (and mosfet voltage drops) present in the O2. That is why the O2 only delivers about +/-11.5Vdc on average under load to the amplifier chips, rather than the +/-12.0Vdc produced by the voltage regulators.
- Using +/-15.5Vdc power rails produces less THD+N from the datasheet graphs, vs lower rail voltages like the +/-12.0Vdc in the O2, for nearly every op amp that goes up to 15Vdc, including the ones used here.
- Power jack in the rear with the on/off switch in the front (or can easily be panel mounted anywhere).
- The on/off switch is a toggle type rather than slide for improved rail-to-rail power up/down tracking (one of the sources of thumps in the O2).
- Same 10K input impedance as the O2 by default, but can easily be built for 20K or 50K input impedance thanks to the FET input LME49880 chip. FET inputs draw nearly zero current (picoamps) vs. tens of nanoamps for bipolar. The input bias current from bipolar inputs can cause current noise when pulled through the input resistors. Although the inherent voltage noise with FET inputs is a bit higher than with bipolar the current noise is much less. Total input noise is the RMS sum of the two, plus the Johnson/thermal noise of the input resistor(s).
- The wider B5-080 case allows for a front panel 1/4" jack, front panel input select switch, rear panel power jack and rear panel RCA jacks. The RCA jacks can be wired either as outputs for an optional internal ODAC or inputs for the amplifier.
- New and updated power management circuit! Battery voltage detection is now per-battery, rather than across both batteries as in the O2 headamp. This means that the special case of one bad battery (shorted) and one good battery will be properly detected. In the O2 there was one circuit across both power supply rails that could get "fooled" by certain combinations of battery voltages, like 11V and 4V (shorted battery)
- The power management circuit now includes a latch to prevent the O2's "motorboating" sound in the headphones when the batteries get low. The problem was caused by the battery voltage rising slightly when the O2's mosfets turned off and removed the load, which would then fool the PM circuit into turning on again. With the load re-attached the battery voltage would again drop and the mosfets would cut off again. The latch in the new circuit prevents the problem. To reset the latch the O2 power is just turned off for 10 seconds, although the low batteries would need to be recharged (or the amp run on AC) or it would latch again, of course.
- Dual stacked front panel LEDs show the condition of both power supply rails individually. The O2's LED was across both rails.
- The LT1540 comparators used in the new power management circuit have built-in precision voltage references. In the O2 the LED was also used as the voltage reference. The new circuit results in exellent matching of the low-power trip point between the two battery cutoff circuits.
- Current sources are used to power the front panel LEDs so the brightness doesn't change bretween batteries and AC operation. The O2 headamp used a single resistor, alowing the LED brightness to fluctuate, along with the power management comparator trip point.
- 5K volume control and 24.9K ground return resistors for lower Johnson noise, given that the new LME49880 chip is rated for a 600R load vs. just 2K with the NJM2068 in the original O2 headamp.
- The standard Bill of Materials lists a linear taper 5K dual pot for the volume control for the ultimate in channel-to-channel tracking accuracy. A part number for an audio taper 5K pot is also listed and can be used instead.
- Lower distortion and vastly lower DC output offset from an OPA2140 looped around the BUF634 chips in each channel vs, the two paralleled NJM4556A chip sections in the O2 headamp.
- Very low current noise Vishay "RN50" 1/8W resistors are used throughout, rather than the Xicon 1/8W in the O2. NwAvGuy wrote that he initially intended on using the RN50s, but switched to Xicon to lower the pirce. He posted that the difference is noise was small, but measurable. Inverting amplifiers require using larger value feedback resistors than non-inverting amplifiers, which results in slightly higher Johnson/thermal noise. Using low current noise resistors like this helps compensate since total noise is the RMS sum of the various (uncorrelated) noise sources.
- In this new amplifier a separate headphone relay circuit takes care of eliminating any turn-on or turn-off thumps.
- The amp uses the same 0.22uF 63V EPCOS film capacitor for bypassing used in the O2 headamp, but all are also in parallel with 2.2uF 0805 surface mount X7R MLCC capacitors between the leads. The ceramic provides 10x the capacitance for decoupling, while extending the frequency range of bypass in the tens of MHz.
- High quality vertical Switchcraft 3.5mm jacks are used in the new amp, much better than the ones used in the O2. These are the same jack I use in the ODA.
- * This amplifier includes reversed biased power rail Schottky clamp diodes that were left off on the O2 headphone amp. I've added these same diodes to my O2 Booster Board.
So how does this new version of NwAvGuy's "O2" headphone amp differ from my version of the O2 Desktop Amp and my O2 Booster Board? :) This new amp also makes a fine desktop unit and is positioned halfway between the orginal O2 and the ODA. The ODA has several premium design features which raise the price quite a bit:
- This amp has just the one pair of votlage regulators, although they are very good and quiet low dropout units which are heatsinked to the rear panel, unlike the O2. The ODA has a 3-stage power supply: a CRC filter feeds a pair of pre-regulators which in turn feed a pair of LDOs for the ultimate in low ripple, line noise rejection, and low noise generation. The ODA power supply also includes a snubber on the transformer secondary and a fast-responding bidirectional transient suppressor. The extensive power supply in the ODA raises the price and takes a lot of PCB and case real estate.
- This amp uses 1% metal film through-hole resistors, although they are the super-low current noise Vishay RN50 units vs. the Xicon 270 1% metal film series used in the O2 headamp. But the ODA uses 0.1% thin-film surface mount resistors throughout, which are even quieter than metal film. The ODA SMD resistors are 80 cents or so each, which raises the price of the ODA. The O2 resistors are 15 cents and the RN50 resistors in this amp are around 30 cents each.
- This amp uses a gain chip that is low-THD+N specified down to a 600 ohm load, allowing it to power a lower johnson noise 5K pot vs the O2's NJM2068 which is only good for 2K and can only power a 10K pot. But the chip here and in the O2 are both dual chips, two amplifiers on one chip to save space. The ODA uses single chips, one for each channel, which allows higher power dissipation and permits the ODA to use a super low noise 1K dual pot.
- The 1K dual pot in the ODA also requires five 4.7uF film capacitors in parallel on each channel to maintain the frequency response curve. Those capacitors alone add $50 to the cost of the ODA and take a lot of PCB space. The small amps like the O2 and this inverting O2 version only have space for one film coupling capcitor per channel, hence the 10K pot in the O2 and the 5K pot here. The 2.2uF coupling caps in the O2 are upgraded to 4.7uF caps to maintain the frequency response with the 5K dual pot.
- This amp has 2 gain switch positions, similar to the O2. The ODA has a 4 position rotary switch with 4 gain choices.
- The ODA has placeholders for optional damping resistors on the output, attenuation resistors on the input, and an optional Zobel network. This amp, like the O2, is too densely packed for any of that to fit.
- The O2 headamp has a fairly large DC output offset of 3.7mV (=3700uV) or so. This amp uses a DC precision chip (OPA2140) looped around a BUF624 current buffer to produce very low output DC offsets of around 250uV (=0.25mV). The ODA uses an independant DC output offset zeroing circuit on each channel which allows the offset to be reduced to below 10uV (= 0.010mV), and will typically stay in a range of +/-50uV with thermal cycling (the amp warming up). The low DC output offset allows the headphone and IEM transducers to idle at their natural mechanical resting places, rather than being forced 3.7mV worth to one side or the other.
- But this amp does have some similarities with the ODA (and O2 booster board) that the O2 doesn't have! Rear panel RCA jacks, rear panel power jack, input select switch, headphone relay to eliminate thumps, front panel 1/4" jack in addition to the 3.5mm output jack. The rear panel RCA jacks, rear panel power jack, and front panel 1/4" jack were common modifications of NwAvGuy's O2 headphone amplifier. With this new version of an O2 they are all part of the PC board.
- The headphone relay circuit in this amplifier is nearly identical to the one on the O2 Booster Board. The same miniature relay is used, although the relay voltage regulator in this amp is specified for the higher +/-15.5Vdc power rails.
The agdr version of the Inverting O2 Headphone Amplifier is a do-it -yourself proect posted on the headphone forum at diyaudio.com. All the project materials are posted there: schematic, layout, part ID diagram, build instructions, Bill of Materials, panel CAD drawings for Front Panel Express, Gerber files to a make a PC board, photos, etc.
For information on buying a PC board and parts, please see here.
Entire website copyright (C) 2013 - 2017. All rights reserved.