O2 headamp output booster PCB

Sergey888 - thanks for the links, I will definitely read through those. That TI app note is 13 years old so things may have changed. The booster board is completetely stable from my tests with the existing X7R decoupling. I've hit it with all sorts of reactive load combinations in my testing and no oscillations at all.

Yeah I recently read about that trick of selecting a capacitor SRF near a problem oscillation. I may wind up doing that on a Fleige notch filter board I'm putzing with. It has a 7.1mHz oscillation despite some heroic attempts on my part at different decoupling methods. I haven't tried bypassing the decoupling caps with COG yet, which I will probably do just for curiosity sake. The problem with that one probably isn't decoupling. The circuit is essentially a positive feedback loop at the notch frequency it seems, so I may simply have too much trace length involved and formed a parasitic resonant tank somewhere.

Hey as per the above post, I did mange to get those series 10R resistors on the op amp power lines on the booster board! I realized that I could notch out some space on the back (a Vee plane) for some more parts. Great suggestion you had about those. Also fantastic suggestion about that OPA140 chip. That chip has really worked out well. The DC output offset with that chip is just amazingly low. I didn't know that chip existed until you mentioned it!
 
Last edited:
Sergey888 - OK I had a chance to read through those links today. Good stuff! Thanks. I see on thing I missed was not so much where the capacitor's SRF is but the absolute value of impedance vs. frequency on either side of the SRF. That older 2001 TI paper seems to be a bit off in the night on the impdedance vs. frequency values of modern MLCCs, like 100x or so, from the graphs.

The 2.2uF X7R Murata caps that I'm using for the op amp decoupling, these guys

http://psearch.murata.com/capacitor/product/GRM31CR71H225KA88%23.pdf

have an SRF at 4mHz but have an impedance of only about 0.7R at 110mHz from the graph, the bandwidth of the LME49600 which should be the highest potential oscillation on the board. Not bad. In the other direction at 300kHz they are at Z=0.25R, and that is about the point where the 47uF solid tantalums on the board start being useful anyway.

So it looks like you are right, I can stop worrying about the 2.2uFs X7Rs for decoupling and don't need any parallel COG caps there.

On the LME49600s though I'm still pondering that a bit. There I have 10uF X7Rs for decoupling. They are Taiyo Yudens which don't have a Z curve in the datasheet, but the equivalent thing in TDK does:

http://www.mouser.com/ds/2/400/C3216X7R1V106M160AC-265275.pdf

which shows Z at about 2R at 110mHz. I'm thinking that given the 47uF solid tantalums on the board power planes (ESR 2mR) for overall nearby burst energy storage, I may be better off actually going in the other direction and dropping those 10uFs X7Rs to the same 2.2uF X7Rs around the op amps. That way I would get the Z=0.7R at 110mHz. about 3X lower.
 
Last edited:
Hi agdr

m instead of M looks a bit disturbing :) m = milli while M = Mega :D

There is also a difference how to parallel ceramics on the PCB.
If you put to different value capacitors (for example 100n and 1n) in parallel and ground them trough the same via you will not be able to see two resonances. You will see only one common, because series inductance will be common for these two capacitors. If you spread them, ground at different points, you will start seeing two independent notches. Also some unpleasant effect may appear. You may see quite sharp resonant peak in between these two notches. That is where a lossy electrolytic near by handy for damping it.
 
Hey Sergery888! Lol - you are right, the way I have that written it does look like "millihertz". :D I should be writing MHz. I also just spotted a typo in what I wrote, those 47uF tantalums are 20mR ESR, not 2.

Yeah I noticed from the graphs how two or more decoupling caps in parallel result in sort of "circus tent" style peaks over the spectrum. I am probably going to change the layout around the LME49600 back to just one decoupling capacitor on each rail, and then switch the 10uF for the 2.2uF to get that lower Z at 110MHz.

I'm thinking about upping the series resistors on the op amp power rails to 100R from 10R. That would put the corner frequency of the filter with those 2.2uFs at 720Hz. I don't see a PSRR vs. frequency plot for the OPA140, but for the OPA827 the positive rails's PSRR starts going south at 5KHz, while the negative rail starts dropping at just 10Hz.

With just the LME49600 as load on the op amps, nearly all the op amp current draw from the rails would be quiescent. 100R on the rail filters would result in just 200mV of drop from the rails with the OPA140, which should be negligible.
 
The V3.1 O2 booster boards are back from fabrication!

The new V3.1 O2 booster boards arrived! I've built one up. A few photos below and a pile of photos out at the V3.1 project Google Drive link in the "photos" folder. So far it is working perfectly. I want to test play it the entire rest of the week though with various test loads before posting the Gerbers out at the link.

A few people have PMed me about at-cost spare boards. After I've tested it this week I'll make a posting in my vendor forum thread here next week, and then send out PMs next week.:)

Thanks again to Sergey888 for all the great suggestions on this project.
 

Attachments

  • IMG_2579.JPG
    IMG_2579.JPG
    137.1 KB · Views: 254
  • IMG_2632.JPG
    IMG_2632.JPG
    163.5 KB · Views: 99
  • IMG_2623.JPG
    IMG_2623.JPG
    146.9 KB · Views: 86
  • IMG_2622.JPG
    IMG_2622.JPG
    154.6 KB · Views: 224
  • IMG_2621.JPG
    IMG_2621.JPG
    201.6 KB · Views: 228
  • IMG_2616.JPG
    IMG_2616.JPG
    163.7 KB · Views: 237
  • IMG_2580.JPG
    IMG_2580.JPG
    107.7 KB · Views: 245
Last edited:
O2 booster board V3.1 Gerber Files posted

I've finished my testing on the new V3.1 O2 booster boards and everything works. I've gone ahead and posted the Gerber files out at the project's Google Drive link in case anyone wants to fabricate their own boards. That link has all the other materials - schematic, BOM, layout, part ID diagram, and build instructions:

https://drive.google.com/folderview?id=0B67cJELZW-i8Vmp5MDVLNzJxTGc&usp=sharing

Then the "4_22_2014 V3.1 with relay fabricated" folder.

I've also just posted out at my vendor thread with the details for anyone interested in at-cost boards:

http://www.diyaudio.com/forums/vend...llel-njm4556al-two-stage-amp.html#post3924592

If you are already on the order interest list out at the Google Drive link I've just sent you a PM. Please follow up in that vendor thread with any questions related to ordering boards, or send me a PM. This time I'm always including the connection pins with the boards (last time was optional with V3.0) and I'm making the relay available since Mouser is out of stock until September. Digikey has them, so this prevent having to make a separate order to Digikey.

V3.1 here will be the last O2 booster board version... unless some new idea pops to mind down the road.
 
Last edited:
I've quickly built up my 3.1 booster board. I feel like I did stellar job on the soldering this time around. But have unfortunately run into a problem and I suspect it's IC2. I really had a hard time getting that part on the board. Perhaps, I fried the 49600. The symptom I'm seeing is a voltage drop from about 24v on down and then it comes back up and continues to drop. I'm measuring pin 2 on IC2 from the - terminal on BT2. Every thing before the 49600 measures at around 11.76v. What a bummer! I've had pretty bad luck with builds these past few months... d'oh! back to Digikey!
 
I've quickly built up my 3.1 booster board. I feel like I did stellar job on the soldering this time around. But have unfortunately run into a problem and I suspect it's IC2. I really had a hard time getting that part on the board. Perhaps, I fried the 49600. The symptom I'm seeing is a voltage drop from about 24v on down and then it comes back up and continues to drop. I'm measuring pin 2 on IC2 from the - terminal on BT2. Every thing before the 49600 measures at around 11.76v. What a bummer! I've had pretty bad luck with builds these past few months... d'oh! back to Digikey!

Hey lets see if we can get this thing going! :)

First off the bat there shouldn't be anything going any higher than +/-12Vdc. It shouldn't be possible to measure 24Vdc anywhere on the board to ground. Are you measuring from the BT2 terminal in the middle, the inner one closest to the other battery? That is where the jumper wire from JP3 should be soldered, onto the top or back of that terminal. In fact, could you post a picture of the board mounted in the O2 from the top?

Are both of your green LEDs on, if you installed them? What voltages do you measure on either side of R1 to that inner terminal of BT2?

Luckily the LME46000s are pretty tough heat-wise and relatively hard to kill. My first guess about any DC output problems on one channel would be one or more pins on the LME49600 or op amp not being soldered all the way down. I've done it myself. The pins all looked just great viewed from the top, even with the magnifying glass. It wasn't until I looked at the pins from the side close-up with a magnifying glass that I spotted the lack of solder on one op amp pin and one LME49600 pin. In fact, even if the joints all look good, going back and reheating all those IC pins on that channel for a second or two couldn't hurt.

The second thing to double check is if one of the op amps is in the wrong way. I have to admit that I've done that once myself too when building one with the OPA827 chips instead of the OPA140s. Very hard to see the markings and beveled edges on those SOIC8 packages. With the OPA140 the gold bar at the end of IC1 (and dot if it has one) should be facing IC2, while the bar on IC3 goes the other way, facing the relay side of the board.

In looking at the V3.1 schematic just now I've realized the IC2 and IC4 text labels didn't show up on the schematic. I've fixed that and reposted the schematic out on the V3.1 project Google Drive link.
 
Last edited:
Testing the IC2 and IC4 buffer chips while on the board

In thinking about my previous post I've come up a general purpose procedure here for testing IC2 and IC4 after they have been soldered on the O2 booster board. This will help isolate whether the buffer is bad or the op amp is bad for each looped pair, if all the solder connections turn out to be OK. I'll write this up for the IC1 - IC2 pair, but all applies to the IC3 - IC4 pair.

1. First check the power supply rail voltage levels to make sure they are OK when the O2 headphone amp with booster board combo is powered on. If you have the green LEDs soldered on the booster board and they are both lit and the same brightness, you already have your answer. :) That is one of the big reasons I've added the LEDs to the booster board, to serve as a diagnostic on the power rails the O2 never never had.

But to check those voltages directly measure from ground, which will be the terminal of BT2 closest to the other battery (the inner terminal), to pins 3 (negative rail) and 5 (positive rail) of the LME49600 buffer IC2. The pins count up from the side of the booster board closest to the O2 batteries. Pin 3 is the pin in the middle (should be around -12.0Vdc) while 5 is the one farthest from the batteries (should be +12.0Vdc).

2. If the power supply voltages are OK the next thing we need to do is break the loop between IC1 and IC2. Power the O2 off and remove the batteries and AC power input. Cut pin 2 of IC2 close to the pad on the board that it solders to. We will fix it later. Then bend that pin up so that it is flat. Then take some solder wick, or even just the end of your iron, and clear the remnant of that pin off the pad so the solder pad for pin 2 is clean again.

3. Now we need to ground pin 2. If you have test clips that can attach to pin 2 without shorting to the nearby pins you can just use those, with the other end going to ground BT2 terminal closest to the other battery). Mini-hook clips are especially helpful here. Another good way to do this is just temporarily solder a small wire to that pin 2 of the IC2, then temporarily solder the other end to ground. The wire would be safer than risking a test clip touching another pin in IC2.

4. Put the batteries and/or the AC power plug back in the O2 amp and power it up. Then check the voltage to ground on the output pin of IC2, pin 4, which is the 4th up from the direction of the O2 batteries. That output pin should measure anywhere from nearly zero up to +/-60mV (the input offset voltage of the LME49600 buffer with input grounded). Any more voltage than that and your IC2 is bad. But... please double check the soldering on the IC2 pins one more time! I would even advise going back and adding a bit more solder to each pin first and trying the test again. It is really hard to see for certain when the pin soldering isn't making contact.

5. If the IC2 buffer chip tests good - your output is within that +/-60mV - then next do a DC test on the output of IC1. Measure the voltage to ground on pin6 of IC1. The easiest way to get to that pin is the end of C5 which is closest to the pin 2 of IC2 that you cut. That end of IC5 is connected to the IC1 pin 6 output. It should measure just nearly zero volts to ground, like 50 micro-volts or less, which would just show as zero volts on meters that only read down to 1mV. If you get zero then you have a solder joint problem on pin 6 of IC1. We know from grounding pin 2 of IC2 that it produces just +/-60mV from the last test, so if IC2 pin 6 is feeding in the same zero volts all should be OK unless pin 6 is just not making contact.

6. If you measure a higher voltage than 50uV or so on IC1 pin6 (end of C5 near IC2 pin 2) like something in the mV or even volts, next check the input voltage to IC1 from the O2 board on IC1 pin 3. You can get to that pin easily by mesuring the side of R1 closest to the IC1 chip, which directly connects to IC1 pin 3. You should measure nearly zero. If so, your IC1 is bad. If you do measure a substantial voltage on the end of R1 then it could still be a shorted chip, or it could be a solder bridge on the O2 PC board. That isn't likely though if your O2 was working fine before the booster board, unless you've built and O2 up from scratch here for the first time to use with the booster board. To check for this, remove the booster board, put the O2's U3 and U4 chips back in, and power up the O2. Then measure pin 3 of U3 to ground. It should be around 1.5mV - 3mV. If much higher then you have an O2 board short.

7. If your IC1 op amp does test bad that is actually more likely than the buffer being bad. The op amps are much more static & solder-heat sensitive than those big buffer chips. But again, be sure to double check the soldering of all 8 pins on IC1 with a magnifying glass and a close-up look from the sides! Even add a bit more solder to each pin if you can. To get those SOIC package chips off the board the best thing to do is cut each individual pin with flush cutters. Then remove the chip and go back with solder wick and clean all the pin remnants off the 8 solder pads.

8. To repair the pin2 on IC2 that was cut, just bend it down again so the pin is hanging over the solder pad. Depending on how long you were able to cut the pin it may even still touch the pad - if so the fix is easy, just solder it up again. If the pin doesn't touch the pad anymore just take the end of a small wire and tin it up with some solder. Then put a little solder on both the pin2 pad and pin2. Then just heat up that end of the wire and the pin2 and pad and solder the wire on as a jumper from pin 2 to the pad. Then cut off the excess wire.

9. If your LME49600 buffer chip is bad, the best was to get it off is to first cut the 5 pins at the end with flush cutters. Then just heat the end of the tab up until the chip comes off the board. Finally take some solder wick and clean up the pads for the 5 pins. Just like that it is all ready for a new chip.

10. All of the above was for the IC1-IC2 pair. For the IC3-IC4 pair the easy way to get to the output pin 6 of IC3 is the end of C11 closest to IC4. The easy way to get to input pin 3 is the end of R8 closest to IC3.
 
agdr - thanks for the thorough trouble shooting tips.

To start with I wasn't using the center most terminal for GND. I have to take a picture to illustrate how everything is set up. More to come on that. Also, didn't install LEDs -- didn't think I needed 'em. But I see how they're helpful.

Thus far I've re-flowed all pins on IC1-4.

Then once I started using the proper GND terminal and measured to both pins 3 and 5 on both 49600s I got around 12v. This was using AC in only, which is what I'll use for troubleshooting consistency.

I also measured on Pin 2 and on IC2. I got a reading of around 12v. On pin 2 of IC4, I got a rather low reading of less than .5v. I'm not sure if this is expected or not. However, If I switch to the other terminal, not GND, I get the same readings but reversed. Pin 2 on IC2 reads low and Pin 2 on IC4 reads around 12v. Perhaps, this is normal.

I haven't had a chance to go through every step you mention, but once I get some time later this evening and tomorrow I'll have more definitive answers.

Thanks!
 
Last edited:
To start with I wasn't using the center most terminal for GND.

That is some good progress! The photos below show how the three booster board wires should be hooked up. The ground that goes to the battery terminal could actually go to either middle battery terminal - the BT2 terminal closest to BT1, or the BT1 terminal closest to BT2. They are tied together on the O2 PC board.

By the way, accidentially hooking that ground up to the wrong battery terminal won't blow anything up on the booster board. Just move it over to the correct battery terminal and solder it up. I purposely designed for worst case on the board and picked capacitors rated at 35V or higher, just in case they got hit with rail-to-rail voltage like this. There are a number of ways that could happen during a build. Some folks may have been wondering why I used 35Vdc and 50Vdc rated capacitors when 16Vdc would have done (and been smaller and cheaper), being just rail to ground. This is why! :)

That longest booster board wire can be routed more neatly than I have it in the photo. That is just a sense line for the relay circuit and isn't involved in the audio path in any way. You can run the wire right over those 4 O2 filter caps, around them, however you wish. By the way, if for some reason a person's particular O2 build has an R8 on the O2 PC board that is not accessible (mounted on end for instance), you can attach that wire instead to the bottom of R25, that resistor to the left of the U2 chip and blue capacitor. That would be the end of R25 closest to the bottom of the O2 PC board. That bottom end of R25 is connected to the top end of R8 on the O2. Same signal.

The ground wire that is under the front edge of the booster board can actually go to that same end of either of those two resistors sitting between the O2 output jack and O2 red LED. Those two ends are tied together on the O2 PC board. I just have it tacked to the frontmost resistor's end.

From the photo with the long blue wire note that nothing connects to the jumper hole by the connection pins on the corner of the board's "tab". That hole is there to optionally connect to the connection pin sticking off the edge of the board closest to it, for a second attachment point to the O2 negative rail. But it can just be left unconnected as done here. But don't connect any ground wires to that jumper hole.

I also measured on Pin 2 and on IC2. I got a reading of around 12v. On pin 2 of IC4, I got a rather low reading of less than .5v. I'm not sure if this is expected or not.

It does sound like you will need to go through those troubleshooting steps in post #150 for IC2, starting with step 2 where you snip pin 2 near the board and bend it up to disconnect it from the board pad. Then ground that pin to the same middle battery terminal with a soldered wire or test clip. With that done power it all up and IC2 should output somewhere between zero volts and +/-60 millivolts if it is OK. My guess is that IC2 will measure OK from the test and from the next steps in that procedure you will find that op amp IC1 is what is putting out the 12V. That voltage on pin 2 of IC4 might be OK, depending on your meter's resolution. That might be as close as it can get to the 60mV (0.06V).
 

Attachments

  • IMG_2618.JPG
    IMG_2618.JPG
    228 KB · Views: 237
  • IMG_2617.JPG
    IMG_2617.JPG
    140 KB · Views: 245
  • IMG_2619.JPG
    IMG_2619.JPG
    170.2 KB · Views: 247
Last edited:
OK so I was able to run a few more tests.

Nothing after R1 or R8. Pretty much zeroes out from the get go.

Cut Pin 2 on the 49600 and jumpered it to GND. Measured Pin 4 to GND and got 10.9 mV. Also have been getting same voltage (-10mV) on Pin 2 of IC4. So it looks like the 49600 is good.

I have re-flowed IC1 a few times to no avail. I measured between GND and C5 and still get ≈12v. So I'm going to use ChipQuik to remove IC1 and try again. If that doesn't work, time to order a new OPA140... or...

I do have some of the alt chips on the way... OPA827 and OPA1641... so I might opt for one of those and remove IC2 as well.

Edit: I did remove IC1 and resoldered it back on. Same results. Good bye 140, hello Digikey.
 
Last edited:
Edit: I did remove IC1 and resoldered it back on. Same results. Good bye 140, hello Digikey.

I agree, that OPA140 is toast. Pop in a new one, patch up IC2 pin 2, and the board should be off to the races. :)

The OPA827 has slightly lower distortion numbers on the datasheet but slightly higher DC output offset. The OPA140 starts out with 30uV or so of offset on power-on and stays that way. The OPA827 starts out at around 120uV or so when "cold", but by 3 minutes or so thermally stabilizes and is back down to low 30uV or so numbers from then on. The OPA827 also has slightly higher idle current, so the batteries would last a little less long.

I'm actually about to add another chip suggestion to the list, thanks to Fsatsil letting me know about the chip, the AD8610. Looks like that chip has been around a while. Same class of chips as the OPA140 and OPA827 in all the datasheet areas. The specifications clock in somewhere between the OPA140 and OPA827. The one catch is that chip can only be used on the "standard" O2 power rails of +/-12Vdc since the max operating voltage is +/-13Vdc and absolute max is +/-13.6vdc.
 
Just finished my O2 Booster last night and it sounds wonderful with my Sennheiser HD580s:) This is only my second surface mount build so try not to be to hard on me about my soldering job. The build went pretty smoothly except for a bad led that I had to replace and a minor issue with the relay turning on and off. Other then that the hardest part of the build was removing the tiny parts from the cut tape packaging:headbash:. Thank you ADGR for all your help and for designing such a great upgrade for the O2!
An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.