A version of an O2 Desktop Amp (ODA) - Page 7 - diyAudio
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
Reply
 
Thread Tools Search this Thread
Old 17th April 2013, 01:42 PM   #61
uncola is offline uncola  United States
diyAudio Member
 
Join Date: May 2012
I refresh this thread every day for exciting new developments. This is my next amp
Is the design still evolving? Maybe you can can start talking to jds/ephiphany or mayflower electronics to do a limited run.. haha this is me sneakily wanting to get one early :P
  Reply With Quote
Old 17th April 2013, 09:43 PM   #62
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
Some good stuff happened this morning. The PCB fabrication company found a layout mistake I made during their pre-check - forgot to mirror the part numbers for the silkscreen for the parts on the back of the board - so that gives me a chance to send in a version with a few more updates.

Yeah I wouldn't be surprised a few of the O2 fabricators would be willing to build one up and/or supply a PCB or parts kit if anyone is interested. Once I get it tested I'll shoot off a PM or email to the group, including Head-n-Hifi and Audio Poutine here on the DIYA manufacturers forum. An interesting value-add would be for the outfits to populate all the SMD parts but leave the through-hole parts for the DIY builder, as an option. Some of the DIY fun but none of the SMD soldering. I've tried to leave as much stuff as through-hole as possible, in the spirit of what RocketScientist / NwAvGuy did with the O2 amp.

Looks like I messed up the Google Drive link on the Google end for the high resolution PCB pictures. They are here:

https://drive.google.com/folderview?...nc&usp=sharing in the ODA -> 80x160PCB B4 case directory -> most recent date.

Some new stuff that should make it in on this PCB fab revision:

  • I've increased the number of (optional) coupling film coupling caps in the middle on each channel to 6. That is 6 * 4.7uF = 28uF now, with a very good electrical reason. With 28uF I'm able to drop that output stage input bias return resistor (the 40.2k R12 and R13 on the O2) all the way down to 5k, a decrease in Johnson / thermal noise, and still beat the high pass filter corner frequency from the O2. The O2 low end HPF is about 1.8Hz while the 6 caps and 5K resistor here would yield about 1.12Hz. That means the low end of the frequency response for the amp starts dropping at about 10Hz on the O2, where here is would be at about 5Hz. The extra margin here also takes into account that each of those caps is 20% tolerance in value, in case someone gets the luck of the draw and the 20% is all on the low end for each.
The problem is cost! 6 x 4.7uF Wima film caps is about $25 per channel just for the coupling caps, or $50 total for both channels. Aarrgghh. So this is all optional - all those caps are in parallel on each channel. Instead just one 4.7uF coupling cap can be used / populated per channel ($4.50 each) with a 10K resistor (if using the 1K pot) or a 40.2K resistor like the O2 (if using a 10K pot). The HPF corner frequency would still slightly beat the O2 since RS used 2.2uF caps. Then later on as money permits all those remaining parallel film caps could optionally be soldered in and the two resistors changed to the lower values to take advantage of the lower (than the O2 amp) thermal noise.
  • The headphone relay circuit now has its own small 24Vdc supply from a MC78L24G, the 40V input On Semiconductor version of the TO-92 regulator. This solves a problem caused by the variable voltage power supply rails of the amplifier, which would cause variable turn-on delay times with the relay control circuit. The constant 24Vdc supply provides a fixed time delay. The 24Vdc relay coil only pulls about 8mA, well within the specs for the 100mA 500mW regulator chip.
  • The number of power supply voltage choices are reduced to 2 from 3, +/-7Vdc and +/-17Vdc (or any two voltage levels, just set by the LDO resistors that are picked). The reason is a lack of right angle PC mount DP3T switches available at Mouser and Digikey. DPDT is the best I could find that would fit in the space available. An external panel-mounted rotary switch could always be used instead to give more power supply voltage choices. +/-7Vdc should cover most headphones below 80 ohms while +/- 17Vdc should cover most above 80R. Those power supply voltage levels are resistor-settable anywhere in the 5Vdc to 17Vdc range, thanks to the adjustable voltage regulators (and heat sinking!), so instead the switch could be wired up for +/-12Vdc and +/-15Vdc power supply rails, etc.
  • Beefed up the headphone output relay to 3A DC contacts (2A AC).
Attached Images
File Type: jpg ODA circuit.jpg (452.4 KB, 491 views)
File Type: png ODA layout both layers.png (160.2 KB, 457 views)
Attached Files
File Type: pdf ODA vs. O2 differences 7 80x160.pdf (83.0 KB, 30 views)

Last edited by agdr; 17th April 2013 at 10:10 PM.
  Reply With Quote
Old 20th April 2013, 12:02 PM   #63
diyAudio Member
 
Join Date: Feb 2009
Glad to see that O2 got some sort of logical successor. Also glad to see that it moves toward more sophisticated design. Despite that O2 truly served its purpose, as for my taste it needed some extra touch on the schematic side. Using several tricks it was easy to improve distortion performance, even without adding extra active components.
To encourage author of ODA, will drop my fraction of substance on low distortion fan blades
Here is a few pictures of my DAC with onboard headphone amplifier. It is sort of less universal thing that ODA. It has fixed gain and no extra frequency compensation stuff, but it was built for very particular purpose.
There are also few pictures with distortion measurement results. On these photos you can see distortion products of 1k tone, amplitude 3Vpk (~+9dBVpk) driven onto 15Ohm resistive load. Because of measurement technique harmonics are actually attenuated by 6dB. So as a result you need to add 3dB to number you see on the screen to get harmonic level in relation to the main 1kHz tone. It comes to around -149dB for the second harmonic, -153dB for the third and more that -162dB for the forth one.
Attached Images
File Type: jpg 1.jpg (242.4 KB, 472 views)
File Type: jpg 2.jpg (268.5 KB, 451 views)
File Type: jpg 3.jpg (136.8 KB, 419 views)
File Type: jpg 4.jpg (122.6 KB, 103 views)
File Type: jpg 5.jpg (121.6 KB, 58 views)
File Type: jpg 6.jpg (127.6 KB, 48 views)
File Type: jpg 7.jpg (129.1 KB, 54 views)
  Reply With Quote
Old 20th April 2013, 02:59 PM   #64
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
Sergey888 - that is a nice looking PC board and some great results! Double sided, too. Looks like you are using a TPA6120A2 for the headphone chip.

Yeah I have come to appreciate how much performance RocketScientst / NwAvGuy managed to pack into that small space for just $30. It has been kind of a puzzle trying to figure out what could be added to it in any way that the O2 didn't already have. Finally settled on higher rail voltage, more output current, lower DC offset, and additional jacks.
  Reply With Quote
Old 20th April 2013, 03:21 PM   #65
diyAudio Member
 
Join Date: Feb 2009
The PCB is actually 4-layer one. It was next to impossible to resolve routing of some sections with my space restrictions. Yeah, it has TPA6120 in it, but it would not be capable to deliver this level of performance itself. I ended up with some sort of composite amp with multipole compensation.

Some ideas with O2 improvement were actually on the surface, and you picked up some of them. It certainly would not cost much (just few extra passives) to add nested feedback around output, like you did in your project. Also adding a bit of driving capabilities may help to reduce higher order crossover distortion (heavy loaded opamp tend to have these).
Anyway nice to see some good ideas been implemented.

Cheers
  Reply With Quote
Old 20th April 2013, 06:51 PM   #66
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
I agree on the 4-layer board. I'm going to try to do the final run as 4 layer. I want to use one layer as a full ground plane, single point grounded with no signal currents going through it, and one for routing. I've had to use wire jumpers in a couple of spots with just the two layers to resolve some routing issues. Those would go away with the 3rd layer. I could also keep high impedance and large signal lines a bit further apart by doing a bit of routing on the 3rd layer.

Hey I'm open to circuit and layout suggestions! Please post any thoughts. The current schematic below. I may or may not to use them, but I'll certainly take a look at any suggestions. Any distortion reduction and/or compensation circuit ideas would be especially interesting.

Speaking of compensation, I don't think I've posted that I did eventually get the very first circuit in this thread to oscillate with additional output capacitance and long cables, the OPA627 wrapped around two NJM4556A buffers in parallel. Looking into that further I found that in addition to the 100 degrees of phase lag the OPA627 has up to 1Mhz (datasheet gain/phase plot), the NJM4556A has an additional 30 degrees of lag up through 30kHz (-130 degrees, still OK). Which unfortunately then falls off a cliff and goes down to around -80 degrees at about 110kHz. That leaves zero phase margin at 110kHz, -100 - 80 = -180 = oscillation, and into the magic 20 degree phase margin buffer zone at about 90kHz. The oscillation on one channel was 77kHz and the other 80kHz. Sometimes the math actually works out.

If I had continued to use the OPA627 I would have added some phase lead compensation at the inputs to move the phase shift back into the stable zone. The BUF634 appears to have 0 degrees of phase shift up through 1mHz, from my read of the datasheet graph, which is why all those OPA627 + BUF634 amplifiers out there apparently never ran into an oscillation problem. RocketScientist avoided the problem by having the NJM4556A's not in a loop with anything else.

The good news is that so far I've had zero oscillation problems with the LME49990s substituted for the OPA627 on that same board (SOIC adapter), which tells me the phase lag of the LME49990 must be less, or possibly even phase lead. Unfortunately the gain/phase plot is missing from the LME49990 datasheet. I've sent a request into TI for that plot, but failing that maybe I can extract it myself with some measurements. It would be nice to know for certain, upfront.

Two new things on the current layout. The on/off switch is now DPDT with the relay coil looped through the extra set of contacts. That provides immediate disconnect for the headphones when the power is switched off. No waiting for power supply rails to start collapsing to disconnect the headphones. And I realized that with the DC servos the coupling caps (all $50 worth of them!) could actually be completely bypassed/jumpered and the amp run as purely DC coupled. Now there is something that couldn't (safely) be done with the O2 amplifier. The servos have a very long filter insertion delay, about 3 seconds, since they are designed to fine-tune the DC offset which is already low. But the amp has the 12 second power-on headphone relay delay, and the servos are looped back before the relay contacts, so as long as the source is turned on and plugged into the ODA input when the ODA power is switched on the servos would settle in time to null out a reasonable amount of DC on the ODA input. I also forgot to say in my last update post why the beefier relay contacts. They are for fault conditions, especially the TRS jack shorting with plugging or unplugging. The ODA output stage can probably push the better part of an amp for an instant before the current limiting kicks in, enough to weld relay contacts that are not rated properly.
Attached Images
File Type: jpg ODA circuit.jpg (410.1 KB, 94 views)
File Type: png ODA layout.png (164.8 KB, 102 views)

Last edited by agdr; 20th April 2013 at 07:12 PM.
  Reply With Quote
Old 21st April 2013, 01:33 AM   #67
diyAudio Member
 
Join Date: Feb 2009
For my opinion single point grounding may be sub-optimal in this situation, if I understand right what you mean. Nevertheless you certainly need to cut some slots in ground plane.
There are only to places where you will have significant ground currents. It is the fraction of plane from power inlet to main capacitors in PSU (the scariest one, will contain a lot of 50Hz products), and from these capacitors to output connector. You just need to make sure that these currents will not upset low signal part of your design.

Speaking of oscillation. You may consider having and output Zobel network with series inductor.
Cable, as an output load, may affect stability as a extra capacitor, or like unterminated transition line. In both cases you want to have something isolating amp output from load on higher frequencies.
  Reply With Quote
Old 21st April 2013, 07:56 PM   #68
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
Quote:
Originally Posted by Sergey888 View Post
Speaking of oscillation. You may consider having and output Zobel network with series inductor.
Good suggestion! I'll do it. I'll add pads for a zobel at the output, which can then be optionally populated if needed.

Something unrelated, I was sent a link yesterday to RocketScientist / NwAvGuy's listing on his blog from back in November of 2011 about what he was thinking of for the ODA at the time:

NwAvGuy: Objective Desktop Amp (ODA) & DAC

I had forgotten about that listing. Looks like I've hit a lot of it except the inclusion of surface mount parts. The total parts cost is higher though, and yes this one is in fact capable of spot-welding with the high output current capability.

I improved the power supply in a different way by adding more physical space around the noisy AC parts and adding the second let of low noise regulators. I know from some email along the way he was thinking of an on-board AC PCB mount transformer, but later in a blog comment saw that he was unable to find any here in the US that didn't make an audible buzz. Running the wall AC to the unit gets into UL and CSA certification land, which is why I stayed with the wall wart and half wave supply. The wall wart is certified so the ODA doesn't have to be.

Last edited by agdr; 21st April 2013 at 08:07 PM.
  Reply With Quote
Old 22nd April 2013, 04:36 AM   #69
diyAudio Member
 
Join Date: Feb 2009
Good call with safety. All you circuitry in this case can be classified as SELV, so as far as power plug has double or reinforced insulation, amplifier will be literally IEC60950 compliant

As for possible pick up of noise from AC part. There is a general approach which may help. Because frequency of our interest is quite low, all coupling from AC power supply lines (and actually from DC supply lines as well, they consist rectified component of amplifier output current) to signal lines will be mostly inductive. To reduce this effect, beside quite obvious spacing, you want to keep area of return current loops as small as possible, on both sides. In this case having ground planes helps. This way (moving apart, reducing areas) you decrease mutual inductance of nets that may possibly interact.

Having extra layer of voltage regulation may not bring as much improvement. But there is something that I would recommend to do. To eliminate possible coupling of the front end amplifiers and what called in this case, output stage, I would put an extra RC filtering on front end opamp supply lines. Another way of doing this is powering them up from separate regulator. For example you can supply voltage to output stage from the LM317/337, and than use extra regulators to power up the input opamps.

You also mentioned that in one case you had to use phase lead compensation. In addition to that, if this one "is not strong enough" you can create additional second order roll off around first op amp. This way it is easy to compensate it even for gain of +1.
  Reply With Quote
Old 22nd April 2013, 11:47 AM   #70
qusp is offline qusp  Australia
diyAudio Member
 
qusp's Avatar
 
Join Date: Oct 2009
Location: Brisbane, Australia
RC filtering will not do a huge amount to get rid of common mode noise, you need common mode filtering. not easy to predict or model, sorta have to build and measure
  Reply With Quote

Reply


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
Building the O2 Headphone Amp 6L6 Headphone Systems 34 4th June 2014 11:41 AM
Help with BantamDAC and O2 Amp jschristian44 Headphone Systems 2 27th January 2012 05:22 PM
O2 Headphone amp for sale! agdr Vendor's Bazaar 10 9th December 2011 03:05 AM
o2 amp building jtktam Vendor's Bazaar 0 3rd November 2011 09:52 AM
Questions before getting parts for the O2 Amp dewasiuk Headphone Systems 3 12th September 2011 04:33 PM


New To Site? Need Help?

All times are GMT. The time now is 07:02 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