Go Back   Home > Forums > Amplifiers > Headphone Systems

Headphone Systems Everything to do with Headphones

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Thread Tools Search this Thread
Old 13th February 2014, 08:14 AM   #101
agdr is offline agdr  United States
diyAudio Member
agdr's Avatar
Join Date: Sep 2010
Default O2 headamp turn-on / turn-off thumps completely eliminated

Well here some good news. I've figured out how to completely eliminate all turn-on / turn-off thumps with the O2 amplifer using the V3.0 booster board with the heaphone output relay.

The solution is shown in the pictures below:

1. Unsolder the 100K R13 resistor from the V3.0 booster board. You won't need it anymore.
2. Cut a piece of 24 - 28AWG wire about 3.5 inches long (9cm). The wire just carries microamps so any gauge wire works here. Strip the ends back a couple of milimeters and tin the ends with solder.
3. Solder one end of the wire onto the R13 pad on the booster board that is closest to the edge of the board. The other R13 pad isn't used anymore.
4. Solder the other end of the wire onto the end of R25 on the O2 PC board that is closest to the voltage regulators. From the photo you can see which one R25 is. Just put a small blob of solder on that end of the resistor first. Then heat it up and stick in the end of the wire.
5. The three DMM shots are on the output of one channel, set for recording peak millivolt voltage levels. The first is with the O2 power switch off. The second is after the 8 second relay delay period, with the headphone relay on. You can see the 50uV DC offset measurement on the DMM on the top of the display. The final photo is with the O2 turned off. Absolutely zero turn-on and turn-off transients anymore.

And that is it! Enjoy your new completely thump-free O2 headamp + booster board combo.


From the above posts the turn-on thump is already gone with the new delayed turn-on headphone relay on the V3.0 board. The turn-off thump was reduced 90% as the peak meter readings showed. The turn-off thump is a tough one to eliminate entirely since it is caused by power rails collapsing. In the case of the O2 headamp that is further complicated by the power rails being turned on and off by the O2's power mosfets, which could happen for low battery as well as simply the power switch being turned off.

To get rid of turn-off thumps the relay circuit needs advance notice of the power switch being turned off, before any other circuitry "knows", to kill the relay. In the ODA headamp I did that by looping the relay coil through a spare set of contacts in the amp's on/off switch. No part of the amp has any further notice of a turn-off than the power switch, of course.

The O2 amplifier equivalent of the on/off switch is the gate circuit of the mosfets, fed by NwAvGuy's comparator circuit. The signals at the mosfet gates are the most advance notice of a turn off, either by the on/off switch or a low-battery mosfet cycle, in the O2 amp. The gate of the Q2 mosfet is turned off when the open collector output of the comparator goes to the negative rail, putting zero volts across the Q2 Vgs junction.

When the O2's comparator open-collector output turns off, the gate of Q2 is brought to the positive rail via R8 on the O2 PC board. Actually, it is brought up slowly by the R8-C21 RC combination. Once the gate of Q2 hits 2.1V or so it turns on.

What this modification is doing is essentially replacing the 100K R13 on the booster board (which returns to the positive rail after the mosfets) with a connection to the gate of Q2, and then using that same 270K R8 on the O2 PCB as a pull-up for Q1 on the booster board. So now when the O2's comparator open-collector goes low, to the negative rail, it not only shuts off Q2 on the O2 board but cuts power to the drain of Q1 on the booster board, which in turn turns off Q2 on the booster and cuts the relay power.

When the Q1 mosfet on the booster board turns on, during the 8 second relay turn-on delay period, it now shorts the gate to drain of Q2 on the O2 board. This has the effect of dropping the negative rail voltage by about 3v for those 8 seconds, but the relay and hence headphones are cut-off so it doesn't matter. Then the timing circuit finishes the 8 seconds, Q1 on the booster board turns off, the gate of Q2 on the O2 board is released, full negative rail returns, Q2 on the booster turns on and the relay turns on.

In the other direction, when the O2's comparator turns on the open collector output transistor and shorts the O2's Q2 gate to source, it also now cuts power to Q1 on the booster and immediately cuts off the relay, before the power rails have started to collapse.
Attached Images
File Type: jpg IMG_2348.JPG (58.3 KB, 158 views)
File Type: jpg IMG_2355.JPG (64.4 KB, 151 views)
File Type: jpg IMG_2350.JPG (67.9 KB, 150 views)
File Type: jpg IMG_2349.JPG (67.0 KB, 149 views)
File Type: jpg IMG_2356.JPG (22.7 KB, 146 views)
File Type: jpg IMG_2357.JPG (21.6 KB, 12 views)
File Type: jpg IMG_2358.JPG (22.4 KB, 6 views)

Last edited by agdr; 13th February 2014 at 08:25 AM.
  Reply With Quote
Old 13th February 2014, 10:58 PM   #102
agdr is offline agdr  United States
diyAudio Member
agdr's Avatar
Join Date: Sep 2010
I forgot one step on the writeup above for the thump-less mod. Another one of the 10V zeners in the BOM (D2, the same type used across the relay coil) needs to be soldered into the "unused" R14 pad, with the banded end facing away from the edge of the PC board (toward R12). That R14 is listed as "do not populate" in the BOM. Now it gets populated.

The extra zener is to clamp the gate-source voltage of Q2 on the booster board to 10V, since that mosfet has a maximum Vgs of 20V. Without the zener the new modification would put 21 volts on that gate with fully charged batteries, 24V while on AC, and 30V with the +/-15Vdc O2 upgrade.

A gate voltage clamp is what R14 was originally used for when I posted that O2 modification a couple of years ago, part of a voltage divider with R13. That mod was intended to be used without a voltage regulator. After I sent the board to fab I realized that with the new 12V regulator R14 wasn't needed since the gate voltage wouldn't exceed 12V, but now with the new mod it is needed again. A zener is better than the resistive divider for the voltage clamp. The peak DMM milivolt readings I posted above showing no thumps were done with this new zener in place.

I will add all the information for the anti-thump modification to the V3.0 build instructions and BOM this weekend.

Also - unrelated - the 8 seconds I've been quoting for the turn-on time delay for the relay was from the math and Spice simulation, but I measured it today and is actually about 5 seconds. 5 seconds is better, 8 would have probably been a bit long. Keep in mind you can change that delay to just about anything by changing the values of C16 and C17.
Attached Images
File Type: jpg IMG_2359_1.jpg (95.4 KB, 26 views)

Last edited by agdr; 13th February 2014 at 11:08 PM.
  Reply With Quote
Old 15th February 2014, 02:57 PM   #103
agdr is offline agdr  United States
diyAudio Member
agdr's Avatar
Join Date: Sep 2010
Default THD+N measurements

I just received a new toy, one of the Quant Asylum QA400 audio analyzers. Essentially just a USB soundcard in a box with dedicated software.


The plots below show the stock O2 on the left and the O2 + booster board on the right. I'm still getting familiar with the equipment so I may have a setting or test condition off here, but looks pretty good so far.

The O2 + O2 booster board seems to be matching or slightly beating the stock O2 THD + N within the margin of error, but adding the 93% output DC offset voltage reduction and zero-thump headphone relay.
Attached Images
File Type: jpg IMG_2360.JPG (78.2 KB, 33 views)
File Type: jpg O2 150R load 1.jpg (133.4 KB, 32 views)
File Type: jpg booster board 150R load 1.jpg (132.9 KB, 26 views)

Last edited by agdr; 15th February 2014 at 03:01 PM.
  Reply With Quote
Old 16th February 2014, 10:21 PM   #104
diyAudio Member
Join Date: Feb 2009
Hi agdr

It is possible you reached the distortion limit of the analyzer. You can use a notch filter which may allow you to look deeper.
  Reply With Quote
Old 17th February 2014, 02:27 AM   #105
Limp is offline Limp  Norway
diyAudio Member
Limp's Avatar
Join Date: Oct 2010
Location: Bergen
Or maybe give it a more demanding load than 150Ω?
Looks good, none the less. At least you know nothing is seriously amiss.
  Reply With Quote
Old 17th February 2014, 03:06 AM   #106
agdr is offline agdr  United States
diyAudio Member
agdr's Avatar
Join Date: Sep 2010
Limp - very, very good observation! There may be a longer story. 150 ohms is what NwAvGuy used in his frequency response measurements and plots on his website O2 page. I think there may be an O2 problem there.

While testing doing testing on the booster board I've subjectively noticed on the stock O2 (without the booster) slightly more bass, it seems, with the AC plugged in than with the batteries. Lack of bass has been something a lot of folks have posted about the O2 since it came out. Some have the problem, others seem to swear there is no problem. To try this, just start playing the O2 on AC on some song with a lot of bass and then pull out the AC plug. Then plug it back in. As long as the output swing is less than 5V rms or so (true for most headphones) running on AC or batteries should make no difference at all in the output voltage swing (volume).

Pondering the O2's circuit a bit as to why that would be the case, I noticed something. While on AC the O2 amplifier circuits are looking back into the output impedance of the LM7xxx voltage regulators which are nearly zero for frequencies under 1K, then only goes up an ohm or so to 20Khz.

But while on batteries, the amp circuit is looking back into the batteries. The voltage regulators are removed from the circuit by the diode logic. On a hunch I looked up the impedance characteristics of 9V batteries. For alkalines I've found a graph showing about 3 ohms at 100 ohms. That is huge IF the headphone are lower impedance, like most of mine. At 150 ohms, where NwAvGuy did his FR measurements, 3R (actually 6R round trip for both batteries) would hardly show up. So far I haven't found an impedance plot for "9V" NiMH, but I have seen postings indicating the impedence is worse for NiMH than alkaline, so it could even be higher than 3R at 100Hz.

Then I went back to NwAvGuy's frequency response postings on his blog, rather curious now at what impedance he measured things. And I was rather surprised when I saw it was 150 ohms. That is one of the big problems with a measurements-only amp design - a person might miss a certain measurement that matters, or in this case the test conditions on a measurement.

So I want to try quantifying that by re-running the O2 frequency response measurements with a 32 ohm load. In fact should be as easy as feeding in some low frequency signals and measuring the Vrms output with a sensitive DMM on AC, then again on batteries.

Now normally the 220uF bypass capacitors that NwAvGuy has across the power rails would solve the battery impedance problem. But... the impdenace of a 220uF capactor at 100Hz is about 7 ohms, worse than the battery it is bypassing, even though it is low ESR. It isn't big enough to properly bypass the low frequency signals around the batteries. Oddly enough NwAvGuy at one point posted "do not increase the value of this capacitor, it is already optimal" which itself sounds odd.

There would be a couple of possible O2 amp circuit fixes. The easy one is increasing the value of the two 220uF caps. Mouser has a 620uF that fits in the PCB space. I'm going to get those on order Tuesday. Those would drop the bypass impedance at low frequencies by 2/3. Another would be running the batteries through LDO voltage regulators, but that would be a major modification on another board.

Last edited by agdr; 17th February 2014 at 03:13 AM.
  Reply With Quote
Old 17th February 2014, 03:31 AM   #107
agdr is offline agdr  United States
diyAudio Member
agdr's Avatar
Join Date: Sep 2010
Originally Posted by Sergey888 View Post
Hi agdr

It is possible you reached the distortion limit of the analyzer. You can use a notch filter which may allow you to look deeper.
I completely agree! I'm actually in the process of whipping one up. I did a post on in the QA400 thread yesterday:

QuantAsylum QA400

The noise floor that NwAvGuy came up with for the O2 headamp was more like -140dB, as I recall, using his dScope which he said had a max floor of -160dB. The loopback on the QA400 is around -130dB or so. I'm going to try a Fliege notch filter since it primarily uses resistors, which should make matching easier.

Notch Filter Calculator

I've found a 0.047uF 1% 50V COG/NPO MLCC for the single capacitor value. I originally had some rather huge 0.047uF 1% 650V polypropelene units in there, but COG should work nearly as well.

Today I've added a 50R pot to very slightly vary the center frequency around the 1kHz, taken from a TI filter app note on the Fliege here:

http://www.ti.com/lit/an/slyt235/slyt235.pdf (opens PDF)

The range of the pot should be enough to compensate for the capacitor's 1%. In the QA400 thread a lot of folks have mentioned the importance of a defined bottom to the notch, like -20dB or -40dB, to know what, exactly, to subtract off the harmonics. That TI app note has a good suggestion. Don't use the bottom of the notch, but rather use the side of the well. In the plots below I've found that a 38Hz deviation from Fc will yeild -3dB, while a 4Hz deviation will give -20dB, if I'm looking at that right.

Interesting to note that the QA400 software has a built-in optional rounding function on the signal generator that causes the 1kHz signal to actually be generated at 1.002kHz on mine. From the 4Hz plot below that would automatically cause a -26dB deep notch without having to adjust a thing on the filter (assuming it was pre-nulled for whatever the capacitor's 1% variation causes).

Any suggestions are welcome! Those 10R power feed resistors on the op amps are actually your suggstion from the booster that I finally have the opportunity to try out. In fact I just realized I should give you credit on the board text for that. I wouldn't have thought of that good idea for improving PSRR. I'm using 0805's here so I had enough board space to give it a try. The 10R + 1uF should put the corner frequency more-or-less where the PSRR on the LME49600 ramps up.

Everything is at a notch filter Google Drive link now too:


The LT Spice plots are for deviations from Fc of 500Hz, 100Hz, 38Hz (-3dB), 20Hz, and 4Hz (-20dB)
Attached Images
File Type: png Fliege 1khz notch filter circuit.png (68.9 KB, 13 views)
File Type: png Fliege 1khz notch filter layout.png (303.1 KB, 13 views)
File Type: jpg LT Spice circuit.jpg (169.4 KB, 16 views)
File Type: jpg LT Spice ac plot 500Hz deviation.jpg (26.0 KB, 8 views)
File Type: jpg LT Spice ac plot 100Hz deviation.jpg (30.8 KB, 3 views)
File Type: jpg LT Spice ac plot 38Hz deviation -3dB.jpg (38.7 KB, 2 views)
File Type: jpg LT Spice ac plot 20Hz deviation.jpg (33.4 KB, 2 views)
File Type: jpg LT Spice ac plot 4Hz deviation.jpg (34.0 KB, 4 views)
Attached Files
File Type: zip Fliege notch BOM 3.zip (10.6 KB, 1 views)

Last edited by agdr; 17th February 2014 at 03:49 AM.
  Reply With Quote
Old 17th February 2014, 05:54 AM   #108
diyAudio Member
Join Date: Feb 2009
Originally Posted by agdr View Post
Those 10R power feed resistors on the op amps are actually your suggstion from the booster that I finally have the opportunity to try out.
Because there is no class B output on the board driving low impedance load, unlikely you'll need this RC filtering. From other hand it will not hurt to put 0 Ohm links. Although I would try to ground decoupling capacitors trough a single via next to each opamp, and also move large caps closer to each other, to put their ground terminals as close as possible. It may help to avoid possible problems with return currents in the ground plane.

Makes sense to check if current noise of LME49990 will not dominate over other noise sources. If it is, beside lowering twin-T network impedance you may use an opamp with lower input current noise.
  Reply With Quote
Old 17th February 2014, 10:06 AM   #109
diyAudio Member
Join Date: Feb 2009
I'm being a bit paranoid, but it is also good practice to put positive and negative rails track in parallel to each other. This way total magnetic filed of those two tracks will contain more main signal harmonic than even harmonics. In case if you accidentally got a parasitic inductive coupling between the supply rails and a signal line, there will be a lower chance of picking up even harmonics. Again, it is more important when you have a class B/AB output stage driving a heavy load.
  Reply With Quote
Old 17th February 2014, 03:35 PM   #110
agdr is offline agdr  United States
diyAudio Member
agdr's Avatar
Join Date: Sep 2010
Sergey888 - thanks, all good suggestions. I did the ODA amp that way with parallel power traces for exactly the same reason, B field cancellation. I'll make that change. Also great idea about the single ground point for each op amp's decoupling caps. Yeah, I'm always a bit leery where return currents are flowing with solid ground planes. Under which parts and next to which parts.

Another interesting bit in that TI app note about the 100mHz THS op amp they used producing inadequate notch depth for a 100kHz notch - not enough bandwidth - but was adequate for a 10kHz notch. The LME49990 is 110mHz. I'm also using the same trick I did in the ODA and making good use of the chip's 600R drive capability. Smaller feedback resistors to lower the Johnson noise. Good point about the input current noise. I'll take a look at that and the trade-offs.
  Reply With Quote


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off

Similar Threads
Thread Thread Starter Forum Replies Last Post
Zen headamp. PCB, my ver. Stream Pass Labs 42 24th February 2013 10:07 AM
FS - Fully populated & working O2 headamp Lil Knight Swap Meet 0 16th January 2013 02:30 AM
Grommes S35 booster amp output? Dill Pickle Tubes / Valves 1 9th October 2012 07:23 PM
WTT O2 headamp PCB for CCS PCBs regal Swap Meet 0 13th October 2011 03:42 AM
FS: PCB's for the Morgan Jones tube headamp Nisbeth Swap Meet 0 19th October 2004 07:43 PM

New To Site? Need Help?

All times are GMT. The time now is 05:58 AM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2