Modern part upgrades for NwAvGuy's O2 headphone amplifier

This is a list of modern part substitution upgrades for the O2 headphone amplifier that have come along since NwAvGuy disappeared in fall 2012. His last O2 headphone amplifier BOM (V1.1) was published on Dec 2, 2011 on his blog as a spreadsheet download. A lot has changed since then! Scroll down, this list goes on for a few screens.

Revised 5/7/2017

If Mouser and/or Digikey are out of stock at the moment on any of these parts I have them for sale. The price is the Mouser price showing plus 30%, plus actual shipping to you and paypal fee costs.  1/3 of that 30% is due to tax.  I'm in Texas and so is Mouser, so I pay 9.75% Texas tax on shipments from Mouser, along with their shipping cost to me.

O2 BOM V1.1 C6, C7 & C1. Voltage regulator stability capacitors (C6 & C7) and power management hysteresis capacitor (C1).

Modern & upgraded part replacement = Mouser #810-FK24X7R1H105K = Digikey #445-8517-ND. 1uF, 50V, +/-10%, -55C to +125C (same capacitance & voltage as the original O2 BOM V1.1 part) X7R MLCC capacitor with a thin 2.5mm profile to fit well in the O2 board space available. $0.40 each vs. $0.43 for the original BOM part. I actually was the one who made the suggestion to NwAvGuy to increase this capacitor to 1uF from his O2 BOM V1.0 for C6 & C7 after finding that recommendation in a datasheet for the negative rail LM7912 regulator. He adopted the suggestion. The original part he has listed in his final V1.1 BOM was the best thing available at the time, a Y5V temperature coefficient MLCC with a tolerance of -20% to +80% (yeah, wow, that is not a typo!) vs. +/-10% for the updated part and -30C to +85C temperature range for the original part vs. -55C to +125V for the upgrade. A Y5V's capacitance value is much more sensitive to temperature and voltage changes than X7R or COG. X7R is always preferred over Y5V if available and the price difference isn't ridiculous. MLCC capacitor technology is one of the things which have been advancing rapidly over the last few years.

The same new capacitor can be used for C1 on the O2 BOM V1.1. The only difference between the O2 BOM V1.1 C1 and the O2 BOM V1.1 C6 & C7 is lead spacing, 2.5mm or 5mm. The upgraded part is tiny - slightly less than 2.5mm - so this new part has the leads formed wider than the part to fit 5mm lead spacing holes. Bend those leads straight with a pair of needle nose pliers and the part will fit just fine for C1. For that reason I'm not specifying a separate part with 2.5mm lead spacing (they do exist) for C1, like NwAvGuy did , to keep things simple.  The same comments about the X7R temperature coefficient being preferable to Y5V apply for C1.

O2 BOM V1.1 C2, C3, C4 & C5. Power supply filter capacitors.

Modern part replacement = Mouser #667-EEU-FR1V471 (or 667-EEU-FR1V471B, the B is a reel tape while the one without the B is bulk, either is OK) = Digikey P14441-ND for $0.71 each vs $0.28 for the original O2 BOM V1.1 part. 470uF, 35V, +/-20%, 105C, 1.16A ripple current (at 60Hz), 28mR (milli-ohms) low ESR, 16mm tall 10mm diameter (same capacitance, voltage, tolerance, and size as the original O2 part). This capacitor is one of the alternates that NwAvGuy has listed in the O2 BOM V1.1.  It is a low Equivalent Series Resistance (ESR) part with an associated much higher ripple current rating.  The new part is rated at 1163mA at 60Hz (note the datasheet 1.79A is at 100KHz, you have to use the datasheet frequency de-rating table) vs. just 555mA for the original O2 BOM V1.1 part. Lower ESR in a filter capacitor means it can supply more instantaneous current and have a lower internal voltage drop (both good things) than a higher ESR part like the original O2 BOM V1.1 capacitor.  The ESR of the original O2 part is not listed in its datasheet, but from the ripple current rating it is likely 2x, or around 60mR.

The new capacitors are also rated for 8000 hours life at 105C degrees temperature (221 F) vs. just 3000 hours at 85C (185 F) for the O2 BOM V1.1 capacitors. The higher temperature specification actually matters in this application because one of the 4 caps, C5, is nearly touching the U5 & U6 voltage regulators which can run extremely hot (150F to 180F or more) due to not being heat sinked.  In a discussion with NwAvGuy once he considered the vreg heating to be OK since the vregs are rated to over 190F based on junction temperature.  Several people have posted over the years about how hot C5 gets, just due to thermal radiation from the nearby vregs and thermal conduction via the PC traces.  The higher 105C temperature specification on this new part means longer capacitor life and reliability in this application.

O2 BOM V1.1 R14 & R20. Gain stage ground return resistor.

Upgraded part replacement = Mouser 594-SFR16S0001002FR5 for $0.21 = Digikey #PPC10.0KXCT-ND for $0.20 each vs. $0.10 for the original BOM part. NwAvGuy used the low-noise (current noise) SFR series from Vishay for all the other resistors in the O2 headphone amplifier where noise matters. But for this particular resistor, which is the input ground return resistor for the gain stage op-amp he was forced to specify the "industrial" grade CCF series since Mouser didn't stock the 10K in the SFR series at the time.

O2 BOM V1.1 C16 & C21. Mosfet RC delay capacitors.

Modern & upgraded part replacement = Mouser #810-FK24C0G1H223J = Digikey 445-8487-ND for $0.48 each vs. $0.23 for the original O2 BOM V1.1  part. A 0.022uF 50V (same capacitance as the original O2 BOM part) +/-5% tolerance C0G MLCC capacitor with -55C to +125C temperature range. The original part he has listed in the O2 BOM V1.1 is the best thing available at the time a +/-20% tolerance Z5U temperature coefficient MLCC with a +10C to +85C range . It is rather critical that these two capacitors match since they set the turn-off times of the mosfets. One power rail will be on while the other is off resulting in turn-off thumps in the headphones if one mosfet turns off before the other. This newer part is a much tighter +/-5% tolerance and the C0G/NPO temperature coefficient means nearly zero variation of the capacitance with temperature and applied voltage. Because of this the voltage can drop to 50V from 100V. There was no need for this part to be rated at 100V, other than to reduce the voltage dependency of the capacitance due to the Z5U dielectric.

Update! Although I've had several people write that replacingC16 & C21 with the updated part solved their thumps, I've had a report that although changing C16 & C21 to the new part didn't help with a person's O2 thumps, replacing the two mosfets Q1 and Q2 did.  Something to try!  If the mosfets are damaged, bad from the factory, or fakes it is possible the rise and fall times and not within specifications.  Those are the same parameters that C16 & C21 effect.   Always buy your semiconductors from a reputable distributor, like Mouser Electronics or Digikey, not eBay.  Too many fakes out there.  Thanks to Angus for the report. :)

O2 BOM V1.1 VR1. Volume control potentiometer and T18-shaft knobs.

A better quality alternate pot (from personal experience), a Bourns Pro Audio dual 10K, 9mm, 25mW.  I've just run into too many of the Alps 10K 9mm pots with bad wipers right out of the package or wipers that get scratchy after just a few months. The Bourns pot is an exact replacement, but these all have a knurled (T18) type shaft rather than the D shaft on the Alps.  They will require a matching knob. There are several version of these pots available at Mouser and Digikey.  NOTE! the 20mm long shafts use the rear set of O2 PC board pot holes, while the 15mm long shafts use the front set of holes.  Be sure to put a small piece of thin cardboard under the metal front of the pot when soldering into the rear set of O2 PC board holes to keep the metal front from shorting some of the front holes underneath it.   

As for T18 knobs I recommend the slimline rubberized knobs at Mouser:.

O2 BOM V1.1 C13 & C14. Coupling capacitors.

Upgraded part replacement = Mouser #505-MKS23.3/63/5 for $2.99 each vs. $0.86 for the original BOM part. These two capacitors are the coupling caps in the middle of the signal path of the O2. The new caps upgrade the BOM's 2.2uF to 3.3uF to extend the O2's low end frequency response roll-off from 1.8Hz to 1.2Hz. The new capacitors are the same physical size as the older capacitors in the O2 BOM and are the same DC and AC voltage rating. Still stacked film to have a minimal impact on the sound. The new caps are 5% tolerance vs. 10% for the BOM cap to provide better channel to channel matching.

O2 BOM V1.1 D1, D2, D5, D6.Power supply diode logic.

Upgrade to Mouser #625-SB140-E3 = Digikey SB140-E3/54GICT-ND for $0.48. These 40V 1A Schottky diodes are the same price these days as the 30V 1A 1N5818 diodes in the original O2 BOM. The forward voltage drop is lower and reverse leakage current is lower on the upgraded diode (both good things).  Use the Vishay part specified here though, don't use the (cheaper) Fairchild SB140!  The datasheet specifications on the Fairchild part are much worse and not a significant improvement over the stock 1N5818 diodes.

O2 BOM V.1.1 J2 & J3. 3.5mm input and output jacks.

Not an upgrade, just an alternate manufacturer for the same jacks. Mouser 490-SJ1-3555NG (black) = Digikey #CP1-3555NG-ND (black), Digikey #CP1-3555NG-BE-ND (blue, Mouser doesn't stock), #CP1-3555NG-PI-ND (red, Mouser doesn't stock), and #CP1-3555NG-GR-ND (green, Mouser doesn't stock). The quality of the original O2 BOM V1.1 Mouser Kycon jacks has proven to be not the best, so here is an alternative from CUI to try and/or if Mouser is out of stock. Most likely however these are made in the same Chinese factory as the Kycons.

Note on the jack's internal switch configuration!  You will find these jacks, and pretty much all 3.5mm jacks, available with no internal switches (SJ1- 3553NG), one internal switch (SJ1-3554NG), and two internal switches (SJ1-3555NG).  The O2 normally requires two internal switches for the J2 input jack.  NwAvGuy uses those witches to ground the inputs when nothing is plugged in to prevent noise pickup.  If you are using an ODAC wired into the P1 holes adjacent the J2 input jack you have to cut those two ground traces to keep the jack from shorting the output of the ODAC when nothing is plugged into the jack. Another solution that doesn't require trace-cutting is to use the jack with no internal switches.

The J3 output jack PC board wiring on the O2 doesn't connect to the switch pins at all, so the jack with two internal switches can still be used, OR either of the other jacks can be used.  NwAvGuy chose to use the same jack in both input & output positions (discussed it with him via PM at the time) just to simplify the BOM.

O2 BOM V1.1 AC-to-AC Wall Transformer.

The "normal load" transformer for the O2 should be the Mouser #553-WAU12-500 (12Vac 500mA). The "heavy load" transformer should be the Mouser #553-WAU12-1000 (12Vac 1A), although Mouser has the beefier #553-WAU12-1500 (12Vac 1.5A) priced $2 less for some reason and it will work great (but is physically larger and heavier). The 12Vac 200mA wall transformer in NwAvGuy's original BOM was never a good idea, and I believe he realized that with the BOM note on the Dec 2 2011 BOM about the 14-16Vac transformers. The current rating of the WAU12-200 is inadequate for heavier loads and the transformer will eventually die (the primary or secondary windings "fuse" and open up) under such loads.

Any of these 12Vac transformers have another problem - 12Vac is not enough to keep the O2's +/-12Vdc voltage regulators from going into dropout with heavier loads since NwAvGuy did not use low dropout voltage regualtors. The larger transformers will help with that problem by having lower secondary resistance, which equates to a higher minimum ripple voltage going into the voltage regulators.

In NwAvGuy's blog he says the transformer voltage can be increased to 20V, which it can, but there is a catch. Going up in voltage significantly increases the power dissipation (temperature) of the O2's un-heatsinked voltage regulators. Turns out with the O2 running at full rated output current, which would be relatively rare (only extremely inefficient low-Z headphones would require that) , the voltage regulators can't handle a higher voltage transformer than 12Vac without overheating. At lower output loads you can get away with 14Vac or 16Vac, such as the Mouser #553-WAU16-500 (16Vac 500mA).

Update! Triad no longer makes AC-AC transformers as of mid-2016, so there a none available at Mouser anymore.  Jameco Electronics has a good selection of AC-AC transformers. (  For the O2 the best choice with the 14Vac 500mA Jameco #171408 currently at $8.95 plus shipping.  We have these in stock too.  An alternate is the 13.5Vac 1 amp Jameco #2138978 for $4.95.  We also stock these.

Higher O2 transformer voltages than the 12Vac 500mA minimum cause more voltage regulator dissipation. If possible always use no higher than a 14Vac transformer with the O2, ideally rated at 1A but 400mA minimum. More than 14Vac will buy you nothing in audio terms with the O2 since that fully covers the regulator's dropout voltage, it will just cause additional voltage regulator heating. Heating (power dissipation) goes up by the square. Every volt over 14Vac causes a considerable increase in voltage regulator power dissipation with no audio benefit at all. has (14vac 500mA) transformers now that Mouser no longer carries AC to AC wall transformers.  

I originally listed the O2 BOM V1.1 12Vac transformer voltage above because Triad did not make a 14Vac AC-to-AC model.   Their next step up was 16Vac.  I felt that the potential voltage regulator dropout under heavy loads with a 12Vac wall transformer was the better trade-off vs. excessive votlage regulator heating with a 16Vac transformer.  But now, with the Jameco 13.5Vac and 14.0Vac wall AC-to-AC transformers, you can have the best of both worlds!  Just enough voltage to cover the O@ votlage regulator drop-out requirement without excess voltage causing unnessary heating.

O2 BOM V1.1 U1. The O2's gain stage op-amp.  

Use the lower noise "DD" versions of these chips as mentioned in NwAvGuy's BOM notes. They are currently priced exactly the same as the noisier "AD" types at Mouser. Mouser #513-2068DD which sells for the same $0.60 as the "D" version.

Update 1: Using the LME49990 chip on an adapter (no longer recommended!).  We had some early sucess with this chip, which was a bit surprising that it didn't oscillate given the chips' huge 110MHz gain bandwidth vs. just 10MHz for the NJM2068.  NwAvGuy did not design the O2 board with the type of power rail bypassing needed to supress ocillations on such a high bandwidth chip.  Yet, it worked and sounded great, for a while anyway.  After TI bought National we ran into reports of the chip oscillating, showing up as heating to 140F and higher.  Sure enough, our more recent batches all oscillate so we removed that substitution recommendation.  The chip has now been discontinued and is no longer for sale by TI making the whole excercise moot anyway.  If you have an LME49990 installed and it is working and not heating - several poeple have reported this - then enjoy!  You lucked out.  The LME49990 has better specs than the NJM2068 in all areas.  Lower THD+N.

Update 2: Other chips for the gain stage.  So far every other dual audio-specified op-amp I've tried in place of the NJM2068 has oscillated.  I don't currently have an alternate recommendation.  This includes the OPA1612, TI's flagship low-noise and low-THD bipolar audio chip.  If some one is aware of a chip with better specs than the NJM2068 that doesn't oscilalte, please send a note at the contact link!

O2 BOM V1.1 U3 and U4. The O2's output op-amps used as unit gain (1x voltage gain) current buffers.  

Use the lower noise "DD" version of this chip, the 4556ADD, as mentioned in NwAvGuy's BOM notes.  These were Mouser #513-4556ADD but Mouser stopped selling them.  We have a stock of these chips that sell for $3 each.

Update 1: Replacing U3 and U4 with the OPA1688 chip (try it if you want but check for oscillations). The OPA1688 has recently been introduced and is the chip we use in the single and dual Super CMOY pocket amplifiers. The OPA1688 is a surface mount chip.  We sell OPA1688 chips pre-soldered onto DIP-8 adapter boards.  The OPA1688 is a fantastic headphone driver chip with specifications very similar to the O2's stock NJM4556A output chips, U3 and U4. The OPA1688 chips are FET-input and have a much lower DC output offset voltage, both of which help to rather drastically reduce the DC output offset from the stock O2 3.6mV-or-so per channel.

BUT.. there is a catch. The NJM4556A has proven to be stable powering a very wide range of load capacitances. Load capacitance come from your headphones drivers and the headphone cable. Too much capacitance on the output of an op-amp will reduce the phase margin to the point where it oscillates. I've tested the NJM4556A chips into 2nF (= 2000pF), a very high capacitance, more than your are ever likely to have with headphones and cables, and they are stable. But for the OPA1688 be to stable into larger capacitive loads it needs either the 47pF feedback capacitor shown on the datasheet for gains greater than two, on in the case of the O2 where the gain is 1x in the output stage the more complicated feedback compensation arrangement show in the Super CMOY schematic, a capacitor in series with a resistor from the inverting input to ground and a series feedback resistor to result in the total 1x voltage gain.

Without these compensation capacitors the OPA1688 will only be stable (won't oscillate) into lower capacitive loads. I'm guessing around 100 - 200pF, but I haven't done exact tests and I'm not aware of anyone who has. It is entirely possible that some headphones, especially with shorter cables, will be a low enough capacitive load for the OPA1688 to be stable even without the compensation capacitor.  You can certainly try it and find out easily enough if your U3 and U4 are in sockets. If the OPA1688 oscillates the chip will likely become quite hot just a idle with no sound going through. If the chip does oscillate it will often be at a very high frequency, such at 1MHz, way above the audio band, so you will not hear it directly. But often oscillation can cause cracking in the sound, or a tinny sort of cast to the high end.  There is really no easy way to know upfront if your OPA1688s will oscillate into your particular headphone/cable load capacitance.

Update 2: Replacing U3 and U4 with the AD8397 chip (not recommended).  A question has come up about this substitution.  The AD8397 is a high current chip which appears at first glance to be a good one to use in a headphone output stage like this.  BUT... the chip does not have the output short circuit protection that the NJM4556A and OPA1688 chips have.  Short circuit protection is absolutely essential because headphone TRS plugs short on their way in and out of the headphone jacks, both 3.5mm and 1/4".  A good DIY pocket amp out there uses the AD8397 in the output stage but notes that the volume must be turned all the way down when inserting or removing the headphone jack.  That works, but easy to forget and just once is all it takes to fry the chip.  Worse still +/-12V is the absolute maximum rating on the chip.  The O2 usually exceeds that on one or both of the rails since the votlage regulators are +/- 0.5V.  Even if they are below 12V my personal experience with the chip near +/-12V has not been good in terms of oscillations.  I use that chip in other filter projects and don't go above +/-9V with it.

O2 BOM V1.1 C8 & C9. Power supply (mosfet switches) bypass and power rail voltage drop slow-down capacitors.

Modern & upgraded part is Mouser #661-EKYB250E331MHB5D = Digikey 565-3887-ND for $0.44 vs. $0.28 for the BOM part. The new part is 330uF vs. 220uF for the original, still 25Vdc, and the same exact size at 8mm diameter, 11.5mm tall, and 3.5mm lead spacing. The original O2 BOM part is a low Equivalent Series Resistance capacitor and so is the new part, with 0.087R resistance at 100Khz vs. 0.117R for the original, a 25% improvement. Ever-lower electrolytic capacitor ESRs have been one of the ongoing market improvements in the 3 years since the O2 BOM was released. The lower the ESR the better job the capacitor will do bypass in the power supply.

The particular part has a bit of mystery surrounding it since NwAvGuy made the statement in his blog "do not increase the value of this capacitor. It is already the optimum value". Just looking at the schematic, as an engineer, it is not immediatley clear why he wrote that. But after some time spent with his power management circuit trying to find circuit methods to further eliminate the O2's turn-on and turn-off thumps, it appears these 220uF capacitors also have the job of slowing down how fast the power supply rails after the mosfets drop after the mosfets cut off power. The goal there was likely to help make the voltage reductions track each other better, since with his circuit one mosfet turns off before the other. The new 330uF capacitors will further help slow down the rail drop.

O2 BOM V1.1 BT1 & BT2. Rechargeable "9V" batteries.

Modern higher-capacity replacement is Maha Powerex 8.4V 300mAh "9V", part #MHR84V at or many other online retailers such as Amazon. $11.73-$14.95 each vs. $4.20-$5.50 each for the original O2 8.4V 250mAh "9V" Tenergy batteries. The updated batteries at still NiMH chemistry with the same datasheet charge rate as the original Tenergy batteries, so the O2's internal charging circuit will still work just fine and the same way. The capacity is just (300mAh - 250mAh) / 250mAh = 20% greater. The new batteries are also specified as low impedance, a good thing, at just 0.5ohms at 1Khz. The new batteries are heavier though than the original Tenergy batteries, which can be a concern if carrying the amp for long distances.

The new ones are rated at 1000 charge cycles vs. a datasheet rating of just 100 charge cycles for the originals. Many folks don't realize that at just 100 charge cycles they need to completely replace the original batteries in the O2 twice a year if they use it on batteries every day. After 100 charge cycles the batteries still work, but the capacity become significanly reduced. The batteries also start exhibiting more of the "voltage rebound" behaviour when cut off, which can fool the O2's power managment circuit and cause it to oscillate when the batteries are low. With the new batteries you are good for at least 2 years rather than the 6 months of the originals.

NOTE - possible point of confusion! Maha Powerex also makes a 9.6V 230mAh "9V" NiMH battery that looks very similar to the 8.4V 300mAh battery on the outside. You can not use the 9.6V battery with the O2's existing charging circuit without some modifications. Be sure to read the text on the batteries to make sure you are using the 8.4V 300mAh version. For those folks interested in a bit more involved modification here, I have instructions posted in the "O2 mod" thread on>headphone forum on how to use the 9.6V batteries. The value of the O2's 1W charging resistors have to change along with some other things.

Entire website copyright (C) 2013 - 2017. All rights reserved.