O2 headamp output booster PCB

Hello Diyers,

Thank you AGDR for your reply

About the ODAC, I decided to move the PCB DAC under the PCB AMP. Like that I could re install the 2 x 9V batteries inside the enclosure.
And after all install the booster.

About the booster, maybe you could offer a buying option with a bundle PCB + Chips already sold :hug: => an offer for noob :D

My iron is a 40W but but but I have got a thrd hand with a magnifier :shy:

I let you about my ODAC mod
 
About the ODAC, I decided to move the PCB DAC under the PCB AMP. Like that I could re install the 2 x 9V batteries inside the enclosure. And after all install the booster.

I have heard that mouning the ODAC under the O2 PC board was possible. Well that is great! Yours could have the ODAC, batteries, and now booster board all in that standard B2-080 case. :)

About the booster, maybe you could offer a buying option with a bundle PCB + Chips already sold :hug: => an offer for noob :D

Now there is a good opportunity for any third party builders out there. Maybe some of the companies selling O2 boards and kits will eventually make the booster board available.

My iron is a 40W but but but I have got a thrd hand with a magnifier :shy: I let you about my ODAC mod

You would need more iron wattage for the LME49600s. Those helping hands with the aligator chips would probably work. I have one of those but it kept tipping over.
 
Here is an update on this O2 booster board project. I received the pluggable DIP headers from Digikey today. They work, but I think it may be possible to do even a little bit better.

The net result is a bit on the tall side, pushing the booster board up to the top of the top slot. That works, but after some more measuring it appears the next header in the Mill-Max series, the 592 (Digikey #ED17064ND) with the 0.335 inch board stacking posts would be just about perfect and would drop the need for the upside down DIP sockets on the booster board tab. I've ordered some and should know in a few days. The other thing that would work would be a similiar dual-end "pluggable" header without the extension posts, but so far no luck on finding any listed at Mouser or Digikey.

The first photo shows the DIP sockets soldered upside down on a O2 booster board. Next is the pluggable SIP header next to the O2's upright U3 and U4 sockets and the booster board's upside down matching DIP sockets. The next two photos show the O2's DIP sockets loaded with the headers that have been cut into groups of 4. The last two photos show the assembled "sandwich".
 

Attachments

  • IMG_2204.JPG
    IMG_2204.JPG
    168.8 KB · Views: 454
  • IMG_2205.JPG
    IMG_2205.JPG
    135.6 KB · Views: 155
  • IMG_2201.JPG
    IMG_2201.JPG
    153.7 KB · Views: 430
  • IMG_2203.JPG
    IMG_2203.JPG
    170.9 KB · Views: 436
  • IMG_2200.JPG
    IMG_2200.JPG
    159.2 KB · Views: 436
  • IMG_2199.JPG
    IMG_2199.JPG
    155.3 KB · Views: 443
Last edited:
The taller interconnection SIP headers for this project arrived from Digikey today and they mesh the two boards together just absolutely perfectly. :D Thanks again to MisterX for the heads-up on these.

These headers are going to make it ridiculously simple to hook the two boards together. Photos below. I'll include a set of these headers with the boards to prevent a separate order having to be made to Digikey.

I'm going to be gone the next couple of weeks for the holidays. I'll be able to send the at-cost DIY boards out to those interested after the first of the year. Keep an eye on the vendor forum thread for my O2 stuff. :) I also see that I accidentally nuked the spreadsheet of those interested. I'll put that back, but here is a better link to the Google Drive location of all the O2 booster board stuff. Just look in the folder with the most recent date, under "orders".

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

Happy holidays! :santa3:

Photos:

* The new "dual pluggable" SIP headers with the taller 0.335 inch board stacking posts.
* The headers snipped in groups of 4 and inserted into the O2 amp's U3 and U4 sockets.
* (next two photos) the free ends of the SIP headers run through the booster board DIP holes. Remember that the bottom two posts of the U4 header stick out from the end of the O2 booster board on purpose. The fab only allowed 50mm of board width, which only took it to the 3rd row of pins.
* The O2 PCB and O2 booster board partially insterted into the B2-080 case to align the board relative to one another with the case slots... then solder it up!
* (next two photos) showing plenty of clearance between things sticking up out of the O2 booster board and the top of the case.
* (next two) shots showing the completed board with batteries installed and the ground wire re-attached to the center battery terminal.
* thing whole thing hooked up and playing music. Only 20uV of DC offset. I forgot to set the other channel meter to show DC in the photo, but it was similar.
 

Attachments

  • IMG_2223.JPG
    IMG_2223.JPG
    193.1 KB · Views: 138
  • IMG_2221.JPG
    IMG_2221.JPG
    230.8 KB · Views: 149
  • IMG_2219.JPG
    IMG_2219.JPG
    202.3 KB · Views: 128
  • IMG_2218.JPG
    IMG_2218.JPG
    148.4 KB · Views: 111
  • IMG_2217.JPG
    IMG_2217.JPG
    122.9 KB · Views: 115
  • IMG_2215.JPG
    IMG_2215.JPG
    146.8 KB · Views: 121
  • IMG_2214.JPG
    IMG_2214.JPG
    144.3 KB · Views: 123
  • IMG_2211.JPG
    IMG_2211.JPG
    172.9 KB · Views: 126
  • IMG_2208.JPG
    IMG_2208.JPG
    168 KB · Views: 127
  • IMG_2210.JPG
    IMG_2210.JPG
    193 KB · Views: 139
Last edited:
I would also like to try this, put me up for 2 PCB's if still available.

thanks,

I'll add you to the list.:)

I'm back in town. Here are some photos of the rest of the parts added. The LEDs, associated anti-thump resistors (O2 modification) and the latch circuit for the O2's power management circuit. I'll take some photos today of how that latch circuit wires into the O2. These modification parts are separate from the amp circuit and are optional. I cover that in the build instructions.

* The first photo shows how nicely the new stacking posts fit. Just solder them up at this point.

* The next photo shows the two LEDs. One per power rail so you always know if both rails are working now. These are "right angle" side-firing chip LEDs in the BOM. They are in series with the anti-thump resistors next to the big caps. Those provide a resistive load to ground after the O2's mosfets (a load other than just chips) which allows the rails to bleed down linearly after mosfet shut-off, preventing thumps. The LED current comes for "free" in series with those resistors. Two small holes have to be drilled in the O2 front panel above the red LED if you want to see those two new LEDs. I'll provide dimensions.

* The next photo shows the latch circuit components. Keeps the O2's power management circuit from oscillating when the batteries get low and it cuts off.

* Two assembled shots, and a couple of photos with it all working. :) The first closeup photo of the DMMs on the two channels is an instantaneous reading, with music playing, so the readings are bouncing around of course. The second shot of the meters is with the input unplugged (and hence grounded). That tiny amount of AC (237uV & 151uV) is just 60 cycle pick-up in the external wiring since I have the boards out of the case and a fluouresent light nearby for the photo. I should have taken a shot with the meter set to freqency, but that is 60hz and not oscillation. I've also checked it with a scope under all sorts of load capacitance conditions now.

In addition to Shure SRH940, AKG-K500, and ATH-M50 headphones I've got a pair of less-sensitive Beyer DT-770s (250 ohm) on the way which should be here this weekend. Sounds great on all so far!
 

Attachments

  • IMG_2237.JPG
    IMG_2237.JPG
    150.6 KB · Views: 91
  • IMG_2236.JPG
    IMG_2236.JPG
    154.5 KB · Views: 97
  • IMG_2234.JPG
    IMG_2234.JPG
    199.2 KB · Views: 101
  • IMG_2238.JPG
    IMG_2238.JPG
    193.1 KB · Views: 104
  • IMG_2239.JPG
    IMG_2239.JPG
    150 KB · Views: 111
  • IMG_2242.JPG
    IMG_2242.JPG
    180.1 KB · Views: 66
  • IMG_2241.JPG
    IMG_2241.JPG
    88 KB · Views: 79
  • IMG_2240.JPG
    IMG_2240.JPG
    205.7 KB · Views: 92
Last edited:
Not going to release this PCB, unfortunately...

Well after a bunch of testing the last few days I've uncovered a glitch in this project that is bad enough that I'm not going to release the PC boards or Gerber files, unfortunately. My appologies to those who have been interested in the boards! But that is why we test these things. :) I've deleted the "orders" spreadsheet from the Google Drive link.

I may whip up a revised board as time permits later this year. The sound of the OPA140 + LME49600 combination is fantastic. I would encourage anyone else interested in the design to simply whip one up themselves, even if just on a proto-board! I'll leave the Google Drive link up with all the schematic and layout information. Those plus the info in this thread should be enough for anybody to whip up their own version.

Details:

The O2 has a bit of a turn-off thump problem in general, with the way RocketScientist is handling things with his power management circuit. When the PM circuit turns off the mosfets, RS is relying on the resulting voltages on the two power rails on the amplifiers to drain pretty evenly. Tracking each other essentially to prevent any rail-to-rail unbalance that could result in a short lived DC offset votlage to the headphones, and hence a thump. He is also relying on the comparator circuit maintaining its state in the face of falling power supply rails, all the way to zero, which is a tall order.

I've discovered that the turn-off thump is actually even more pronounced with this O2 booster board. That is even with the turn-off thump "fix" modification included on the board which does help with the regular O2 NJM4556A chips. The turn-off thump with this board is not only more pronounced, but sort of rings like a small "boing" sound, which is likely the OPA140 + LME4600 loop not being happy as the power supply rails both collapse.

To me, having headphones connected to the output of an amp at all with collapsing power supply rails to the amp's innards is just not a good way to go. An output relay really is helpful, like I added to my ODA amp. The relay circuit keeps the headphones cut off for a few seconds while the chips thermally stabilize while turning on, then cuts off the headphones before the brunt of the power supply rails collapse while turning off.

What I may do with this board design later in the year is replace the power management latch modification parts with the output relay modification that I posted here way back when:

http://www.diyaudio.com/forums/head...-crc-diode-cap-heatsink-mods.html#post2785867

...if I can find some way to fit a relay in the space available. That circuit does work very well. Results in a nearly silent turn-on an turn-off for the O2. I am still using the circuit on the one of my O2's here shown in the pictures in that link.
 
Last edited:
Why not redesign the original Objective 2 Headphone AMP

Why don't you redesign the entire Objective 2 Headphone AMP, using all
this new circuits? Talk to the original designer and see what he thinks.
The new design could be a mixture of the best from the original design, your
ODA Desktop design, and this new booster work. Take care.:D
 
I've discovered that the turn-off thump is actually even more pronounced with this O2 booster board.

Hi agdr

You can try a few things here.

Reduce amount of capacitance on the board allocated after MOSFETs. If you do it carefully and MOSFETs have reasonably low Rdson, everyone will be happy even with pretty low uF. I think this way it is possible to reduce duration of this thump.

The second thing is that having different capacitance on the positive and negative rails may help to compensate the thump if it has some particular polarity.
 
mrsavage: a redesigned O2 would be interesting! The designer, RocketScientist / NwAvGuy, has dissapeared though and taken his copyright and licensing restrictions with him. No way to get in touch with him for a copyright release. The person who probably had the best shot was George Boudreaux who co-designed the ODAC. He posted a few months ago that he sent RS a package that was signed for, but no reply. So it appears he is missing on purpose. I've had a total of 4 companies contact me over the last year and a half wanting to do an O2-ish amp and not being able to get hold of RS. But ultimately that question of the licensing puts the brakes on it all.

The booster board was a way around that whole issue by making an add-on to the original O2, rather than a new O2 board. Good thoughts though!! I can think of a lot of stuff I would love to see on an updated O2, like lithium batteries with a charge control chip, a DAC and bluetooth connectivity.

Anymore, with the latest crop of headphones having high sensitity and low impedance to work directly with iPods, I think it is becoming a big question whether headphone amps even have a future.

Sergey888: Definitely something to try. I'm kind of curious where the thump is coming from. I was thinking today that opc's LME49990 + LME49600 "Wire" amp doesn't have any significant tturn-off thump and it also doesn't have an output relay. Just collapsing power rails, although no mosfets involved. Since the OPA140 is a precision DC op amp maybe it is especially sensitive to power rail imbalance. I have the parts here to build an OPA827 + LME49600 version up. May be interesting to see if the thump is the same with that. I've been tossing around some ideas about using a depletion mosfet to ground the input of the LME49600, with a resistor in series between the two chips, at mosfet turn-off.

As time permits, if I can think of a way to get the scope or recording DMM to trigger & store amp output voltage at the mosfet turn-off, I might try to record that last second and see exactly what is going on.
 
Last edited:
The power rails imbalance does not matter for opamps until collapsing rails "hit" the opamp common mode input range or internal current sources start to wrap down. You can have a look which of these happens earlier. Since OPA140 has a JFET input, it's common mode range can be quite asymmetrical.
 
Last edited:
The unmodified O2 amp may have transient peak issues on power cycling

Sergey888 - hey thanks. I was wondering if there was something about the JFET opamps vs. the bipolar that I was missing here, on power down. That does sound like one to look at.

I built up the the OPA827 version tonight with some interesting results. The turn-off thump is still there, but it does look like the O2 design itself may the culprit. The booster board here may have just brought an O2 problem to the surface.

I had forgotten the DMM has a fast-repsonding peak-hold function. Just what I needed to capture the thump on the output. I had never tried capturing peaks on an un-modified O2 output before. Here are the rather distressing results...

The first two photos are the test setup and average AC and DC output results on one channel. I have the new Beyer DT-770s hooked up to allow me to crank the output up a little bit. These guys seem to have normal volume levels at around 175mV vs. the 50mV of the more sensitive, lower impedance cans.

The OPA827 has a definite thermal warm up period to DC-stabilize, it appears, much more so than the OPA140. As shown on the meter here the initial DC output offset is around 250uV - 300uV (270uV int he photo), about 10 times that of the OPA140, but still 1/10 of the O2 with the NJM4556A chips. But after about 5 minutes of "on" time warm-up that drops down to around 30uV for the OPA827. Both channels did the same. I noticed this a month or two ago when I did tests on the earlier board I posted with the OAP140 on one channel and the OPA827 on the other.

BUT... the fun stuff is with the peak readings. The next photo is a peak hold reading with the OOPA827 booster board in place, switching the O2 from off to on. Notice that the meter is now on the volts scale, not mV anymore, so that is a whopping -231mV - not uV - fast-moving turn-on transient. I'm not really hearing a turn-on transient, so this must be an extremely fast spike. Still, it is there. Not the best.

The next photo shows what happens then when the O2 is switched from on to off. A positive 197mV spike, which is what I'm hearing, so this one must be longer in duration. Keep in mind that the average DC output offset is still just 30uV or so while the unit is on.

So then I start wondering what the O2 amp's output peaks are with the normal NJM4556As installed. The next photo shows the O2 with the booster board removed and NJM4556A's re-installed. No tricks up my sleeve. :) The next photo shows the resulting DC output offset, the O2's typical 3.7mV-or-so.

The next photo is the off-to-on peak hold reading of the O2 with the NJM4556A chips back in. At first it showed just 14mV or so, but after about 20 seconds that suddenly jumped to around 24mV, than at roughly the 45 second mark, boom - that -231mV negative going fast transient. It was stable from there. I retried this 4 or 5 times just to make sure it wasn't an error, but the O2 behaved the same way each time. Coming up on a minute it is somehow generating a mystery fast moving -231mV transient. That is the same number that popped up with the booster board, so I'm taking this as confirmation that the booster board is just bringing an O2 issue to the surface here a bit more rapidly.

The final photo shows the peak hold reading with the O2 then switched off. This generated the 122mV positive peak on the meter, which is the O2's typical, audible, turn off thump. Longer duration.

I should do the same test on a couple of other O2s here. Maybe there is some issue with just one. But I would be curious what other folks get if anyone else out there wants to test their O2s with a peak hold meter or scope. All of these peak tests were done with the source input unplugged and hence grounded.

In summary, in my view anyway, a 122mV transient is a distressing thing to see being feed to the headphones on turn-off with the normal O2 with the NJM4556A chips. The O2 really needs a on output relay, IMHO. :(
 

Attachments

  • IMG_2258.JPG
    IMG_2258.JPG
    107.4 KB · Views: 46
  • IMG_2257.JPG
    IMG_2257.JPG
    104.2 KB · Views: 45
  • IMG_2256.JPG
    IMG_2256.JPG
    105.7 KB · Views: 84
  • IMG_2255.JPG
    IMG_2255.JPG
    204.2 KB · Views: 238
  • IMG_2253.JPG
    IMG_2253.JPG
    105 KB · Views: 228
  • IMG_2252.JPG
    IMG_2252.JPG
    105.6 KB · Views: 236
  • IMG_2251.JPG
    IMG_2251.JPG
    99.5 KB · Views: 231
  • IMG_2249.JPG
    IMG_2249.JPG
    205.3 KB · Views: 241
Last edited:
O2's power management circuit generates transients

I found the problem. RocketScientists' design of the O2's power management circuit is generating the turn-on and turn-off transient peaks. The circuit is leaving one power rail mosfet on longer than the other, resulting in one power rail being connected and the other disconnected to the amplifier chips for an instant. The rail imbalance generates the transients.

He has the gates of the two power rail mosfets going through cascaded comparators, adding a bit of propagation delay. But the bigger issue seems to be that the comparators have open collector pull-down transistors, then he has an RC pull-up network on each. So the "on" direction for the compators is fast, but the "off" direction is a slow RC charge or discharge. For each power cycle you wind up with one mosfet with a "fast" turn on while the other is experienceing a "slow" turn on.

For example, when the PM circuit turns on, the gate of Q1 is quickly turned on by the first section of U2 pulling down. That turns on the positive rail to the amp chips. But the negative rail isn't on yet. The gate of Q2 has to wait for an RC time delay to eventually turn on.

In the photos below I've just intentionally solder bridged the O2's two mosfets to bypass the PM circuit. The next photos shows the turn-on peaks now. The final photo the turn-off peaks. No 200mV transients anymore. The negative rail transient that showed up at 40 seconds or so is gone too. RocketScientist must have some low frequency oscillation going on with that PM circuit, or an RC section not working the way he is expecting.

I'll see if I can fit an output relay onto the booster board design. That would fix this O2 problem. It looks like the relay circuit will have to delay turn-on a full minute to avoid that negative peak the O2's PM circuit is generating.
 

Attachments

  • IMG_2261.JPG
    IMG_2261.JPG
    102.8 KB · Views: 49
  • IMG_2260.JPG
    IMG_2260.JPG
    101.4 KB · Views: 68
  • IMG_2259.JPG
    IMG_2259.JPG
    174.9 KB · Views: 80
Last edited:
Turbon - yes indeed, that version of an ODA is all mine. :D The consensus definitely came down on the side of all aspects of the ODA being different enough from the O2 as to not be a "derivitative". A couple of law students in different countries even wrote. The ODA most definitely has a relay and it works well.

Speaking of relays... guess what fits? :D Drops right in between the O2's on/off switch and the filter caps. I can mount it upside-down in Sergey888's famous dead bug style! :) The holes for the upside-down 1/4" jack will have to go, along with the PM latch mod circuitry. The rest of that space will be used for the relay circuitry.

The relay winds up kind of close to AC parts, but at least this is the output (low impedance) side which will reduce any AC pickup. The proximity to the two O2 rectifier diodes actually isn't as close as I would have thought. They wind up more on the outer edge of the relay.

I'm going to prototype the circuit and if it all works, including completely eliminating any turn-on and turn-off thumps, I will probably fab a version 3 of the booster board.
 

Attachments

  • IMG_2262.JPG
    IMG_2262.JPG
    105.8 KB · Views: 86
  • IMG_2263.JPG
    IMG_2263.JPG
    147.4 KB · Views: 97
  • IMG_2264.JPG
    IMG_2264.JPG
    162.7 KB · Views: 104
  • IMG_2265.JPG
    IMG_2265.JPG
    136.6 KB · Views: 100
Last edited:
Hi agdr

I just looked through the schematic again and came to a conclusion that reconnecting R3 and R6 to LME output instead of the ground may bring some PSRR improvement.
The ground on the booster PCB is more like power, and it is good of you have no reference to it in any point.
It will preserve the second order feedback roll-off around the buffer.
 
Sergey888 - OK, I'll make that change on V3. Makes sense.

I've been thinking of adding a high value resistor (1 meg or more) to ground on each OPA140 non-inv input though. When the booster board is plugged into the O2 there is a 40.2K resistor to ground on that line on the O2 board. But... before the booster board is plugged in I've been a bit concerned about static damage from having those JFET inputs attached to pins with no return path to ground. Ideally people assembling the booster board will use anti-static grounded wrist straps, but a lot of folks probably won't. In the build instructions I'm going to have soldering in the OPA140s as the last step, making sure the 1 megs are in place first.

Then there is the safety issue if the O2's U3 and U4 socket contacts become intermittant a bunch of years down the road. If those JFET inputs float with no return path for the input bias current the output would probably shoot up to some DC voltage, which would appear at the headphones. The 1 meg resistors would prevent that eventual failure mode.

Once the booster is plugged into the O2 the 1 meg resistors would be in parallel with the 40.2K's on the O2 board, so the overall Johnson noise shouldn't suffer.

Good point about the ground and power. You are quite right, with the change you suggest all of what goes to ground on the booster board will be the IC power rail bypass caps, except these new 1 meg input resistors. I have 3 different ground points (PC holes) on that board right now with build instructions to pick any one of the 3. So far in the builds I've been using the one next to the center ground connection of the two batteries.

But here is an idea. I could separate the ground for the 1 meg resistors and return it to the output ground, in the P2 connector on the O2 by the output jack, since that is where RocketScientist / NwAvGuy has the 40.2k resistors on the O2 board grounded. A snip of that section of the O2 ground foil is attached below.

So the majority of the groundplane on the booster board, with the bypass caps being returned to ground, would ground at the battery center connection. But specifically the 1 meg resistors on the JFET inputs would ground at the O2 output, an inch of trace or so away from the ground on the 40.2K resistors they would wind up in parallel with. That would just mean soldering two ground wires from the booster board to the O2 board rather than one. Comments are welcome!
 

Attachments

  • O2 ground 1.png
    O2 ground 1.png
    24.5 KB · Views: 97
Last edited: