A version of an O2 Desktop Amp (ODA) - Page 2 - 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 14th February 2013, 04:06 PM   #11
jcx is offline jcx  United States
diyAudio Member
 
Join Date: Feb 2003
Location: ..
Quote:
Originally Posted by abraxalito View Post
Looks to me (from the 20kHz THD plot) that the linearity problems are down to the NJM4556's undegenerated input stage. Also there's no PSRR nor open loop gain/phase plot provided, I think some data could be inferred from the 40dB gain reponse plot though with reasonable assumptions being made.
NwAvGuy: Cmoy eBay Headphone Amp
shows ~ unity gain performance of the 4556 - the 20 kHz distortion is much lower than the datasheet Av 30 dB example, ~ 30x lower diff input V yeilds ~ 60 dB lower distortion generation from gm nonlinearity

psrr can be hugely improved in a multiloop - sub regulate the input/global feedback op amp supply

this can require gain in the output buffer if you use much headroom for the regulation - a gain of +2x may be reasonable but does slow the output stage - can be OK as long as the overall closed loop gain is higher, but the local input op amp feedback C may be needed

Last edited by jcx; 14th February 2013 at 04:15 PM.
  Reply With Quote
Old 14th February 2013, 04:12 PM   #12
diyAudio Member
 
abraxalito's Avatar
 
Join Date: Sep 2007
Location: Hangzhou - Marco Polo's 'most beautiful city'. 700yrs is a long time though...
Blog Entries: 109
Send a message via MSN to abraxalito Send a message via Yahoo to abraxalito Send a message via Skype™ to abraxalito
Yes its lower, granted. HF linearity is still an issue though - look at the CCIF IMD plot. Then factor in that this amp is going to be fed from a DAC with significant OOB energy (i.e. the ES9023) typically.
__________________
Seek not the favour of the multitude...rather the testimony of few. And number not voices, but weigh them. - Kant
The capacity for impartial observation is commonly called 'cynicism' by those who lack it.
  Reply With Quote
Old 14th February 2013, 04:53 PM   #13
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
Quote:
Originally Posted by jcx View Post
I would try Class A push-pull bias of the output op amps ~20-30 mA total to ~ 1/2 the combined packages power rating
Good thoughts! I kind of like that idea. I keep looking for things to make this ODA-version a bit different/upgraded from an O2.

The class A bias would probably put this thing in the Lehmann amp category in terms of heat generation but I wouldn't necessarily be opposed to it. I would think about going back to the previous all through-hole layout though, below, that uses the amp board in front and a power supply/ODAC board in back in a longer B3-160 case. On that one I used the DIP8 packages for the NJM4556s and spaced them apart enough to be able to use the Thermalloy DIP8 heat sinks on each.

580100B00000G | 8 Pin Dip Package 30.00 C/W Thermal Resistance Slide On Heat Sink | AAVID THERMALLOY - Future Electronics

I used those on one of my O2 builds and they worked very well. A bigger case would also give the heat generation more surface area to dissipate into the aluminum case. On this through hole version I had enough board space to include the trim pots on the offset-null pins on the OPA627. Squeezing even more DC accuracy out of the pricey chips.

I'm kind of partial to using DN2540 depletion mosfets for CCS although that might be on the noisy side (for the class A biasing). Something to ponder.
Attached Images
File Type: png ODA through-hole.png (40.2 KB, 714 views)

Last edited by agdr; 14th February 2013 at 05:03 PM.
  Reply With Quote
Old 14th February 2013, 05:31 PM   #14
jcx is offline jcx  United States
diyAudio Member
 
Join Date: Feb 2003
Location: ..
Opamp based Amps - Question

shows my push-pull scheme - you need a small stable Vref riding the 1st op op amp output to bias the output buffer pairs against each other
smaller dV, but greater regulation can be had with a Bandgap scheme - differing current ratio in identical diodes


I think multiloop performance with with high frequency noise in the input signal is mostly up to the input device performance - it has plenty of "authority" to servo out IMD products in the audio range to below the O2 -80 dB criteria for IMD coming from sub % output buffer nonlinearity in absence of deadband - and added Class A output bias should eliminate deadband

the input diff pair linearity may be where problems could arise with sufficeintly high out of band signal - the 627 is pretty good
a ~200 kHz low pass input shouldn't harm audio, gives some atteuation of RF


but if you want to see alarming out of band hash - look at SACD DSD with a single bit DAC - they really need 6th order filtering

Last edited by jcx; 14th February 2013 at 05:43 PM.
  Reply With Quote
Old 14th February 2013, 06:30 PM   #15
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
jcx - that heatsink in your link is an absolute work of art! That is fantastic. I didn't know about that thread from way back when. Very good stuff. Thanks for the pointer. I'm on the lunch break now but I'll read more tonight. I see in general, though. Your bias scheme is not just the typical ccs but a push-pull class A biasing.

Yeah the more I think about it your class A biasing idea really rocks. That is something that RocketScientist couldn't have even considered with the O2 due to battery drain. For an AC power only desktop-only version here though it fits right in with the theme, I think.

I've remembered why RS wound up with the NJM4556As now over the OPA551. I think he posted somewhere that the OPA551 was somewhat unstable into capacitive loads, as was the AD8397, but he didn't get that with the NJM4556A.

I've spent a boatload of time over the weekend with noise analysis on the front end, given the pot up front. Nearest I can tell the increased Johnson noise isn't really going to be audible, especially from the various posts that people seem to be using their O2 at 1x and 2.5x gain. The 6.5x wasn't as popular as RS initially thought.

From what I can tell it seems that maximum Johnson noise comes at the midpoint of the 10K pot, with 5K on either side of the wiper. Less rotation and the resistance & noise is less than 4K. More rotation and now the voltage divider starts doing its thing and reducing the noise along with the signal. If anyone has worked that out (noise figures from a pot at the front of the circuit) and has some thoughts to share, please post. At least the current-induced resistor noise will be negligible with just 5pA of input bias current.

Last edited by agdr; 14th February 2013 at 06:33 PM.
  Reply With Quote
Old 7th March 2013, 03:03 AM   #16
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
Talking It works!

The crazy thing works! The PCB's arrived from China on Monday and package of parts from Mouser today.

The project has actually exceeded my expectations and I'm not entirely sure why at the moment. It is completely noiseless, at all pot settings, with a gain up to 12x. The O2 headphone amp has a extremely small background hiss at 6.5x gain and the volume all the way up (input grounded), so this thing actually has less even less hiss than the O2. That is a surprise, given what should be the increased Johnson noise with the pot at the input and the FET input op amp. I just wasn't sure from the noise calculations. Hard to translate that into what is actually going to be audible. And that is with my senstive AKG K550 cans that are full volume at about 80mV.

The 12x gain was a mistake too, I had forgotten the feedback resistor is now 3K for the bass boost circuit, rather than 1.5K in the O2 amp, therefore the resistors to ground need to be doubled. So instead of the 1x, 2x, 4x, and 6x I had intended with the rotary switch I have 1x, 4x, 8x, and 12x. But worked out well for hiss testing. Just a matter of pulling out those resistors and stuffing in double the values.

Nuking the O2's 3mV or so of DC offset was the primary goal of my project and it certainly has achieved that, as the photos below show. I was expecting a low of about 300uV (0.3mV). Instead the 1x position in the photo is just 30uV! The DMM is good for 1uV resolution so that should be fairly accurate. The 4x position is 71uV, the 8x position 98uV, and the 12x gain setting 112uV. Woohoo!

No oscillation at all up to the 100Mhz limit of my scope. I even tried a parallel 680pF capacitor across the headphones to try inducing instability. Nothing. The bass boost didn't de-stabilize it either, even though that is putting a 0.1uF in the feedback loop which causes phase shift. I was really expecting some stability trouble with the bass boost, especially at the high gain setting with the increased load capacitance, but not so.

The schematic and layout that went out to fab is below. Since then I've traded the 1/8" output jack for a 1/4" output jack which seems to be more standard on desktop headamps. I managed to get a front panel RCA jack wedged into this one along side the input 1/8" jack. I used solid polymer electrolytics for the power rail bypasses (390uF) in this one rather than the wet electrolytics.

I also manged to stuff in a third 4.7uF film capacitor in parallel for the input blocking capacitors on each channel, bringing that to 15uF. That was important since I decided to lower the pot from the 10K used in the O2 to 5K (audio taper). The lower pot value may be one of the things that has led to the exceptionally low hiss. But to keep the LPF corner frequency the same as the O2 between the blocking cap and the pot I had to up the blocking caps to 15uF.

The output balancing resistor scheme seems to have worked out very well, too. The O2 uses one 1R resistor from each op amp output to the headphone output to balance the two op amps in the single package. In addition to that I've added another 1R resistor between packages, as the schematic shows. The theory was that the op amps in a package would be well matched but not so much between packages. Measurements this evening tend to bear that out. I measured around 300uV across the two 1R resistors in one NJM4556A package, while both in the other package were around -100uV.

I'm not entirely sure why those DC offset drops are so much lower than the O2's 3mV. I know half of that 3mV was NJM4556A input bias current going through the 40.2K resistor to ground, which is all removed here. The OPA627 will effectively has zero ohms out, of course. But that leaves 1.5mV of inherent DC offset at the chip's input. Apparently the feedback loop with the OPA627 is working extremely well to null that remaining NJM4556 input offset, even though there are 4 op amps now on each channel over two packages.

No significant heating of any of the chips either. Lol - only stuffed one channel to see if it was going to oscillate itself to death or what.

In the photo I'm tapping off the +/-12Vdc from an O2 since I don't have the power supply board for this thing out to fab yet. So all of these numbers are with just the O2's power supply.

I also haven't stuffed the offset trimmer pot holes yet, and at this point I probably won't. I know that using the trimmer adjustment on op amps is one of the quickest ways to kill the PSRR and unbalance the op amp's input stage. With just 30uV of overall DC offset there really is no point adding the trimmer circuit.

jcx - if you happen to read this - I decided that clever class A scheme was a bit more than I wanted to roll into this one, but I have added a position on the PCB for the simple resistor on each package output (after the 1R resistors) to a rail to bias the NJM4556A's into class A that way. I haven't tried it yet but probably will. Looks like 750R or so would do the job.
Attached Images
File Type: png ODA circuit.png (55.0 KB, 621 views)
File Type: png ODA layout.png (46.0 KB, 592 views)
File Type: jpg IMG_1800.JPG (148.7 KB, 622 views)
File Type: jpg IMG_1801.JPG (203.5 KB, 588 views)
File Type: jpg IMG_1802.JPG (201.2 KB, 280 views)
File Type: jpg IMG_1803.JPG (222.3 KB, 265 views)
File Type: jpg IMG_1804.JPG (218.2 KB, 151 views)
File Type: jpg IMG_1805.JPG (221.0 KB, 175 views)

Last edited by agdr; 7th March 2013 at 03:28 AM.
  Reply With Quote
Old 7th March 2013, 03:25 AM   #17
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
...and the matching power supply board that goes in the other half of a B3-160 aluminum case. I've already posted about this in a thread in the power supply section here. In the latest incarnation LM317/LM337's are acting as pre-regulators for low(er) noise LT1963A's and LT3015's. +/-16Vdc.

I've labelled the various sections for any hobbyists out there learning some electronics.
Attached Images
File Type: png ODA power supply board circuit.png (53.3 KB, 242 views)
File Type: png ODA power supply board layout.png (46.5 KB, 231 views)
File Type: jpg ODA power supply Spice circuit.jpg (144.8 KB, 216 views)
File Type: jpg ODA power supply Spice plot.jpg (55.8 KB, 163 views)

Last edited by agdr; 7th March 2013 at 03:34 AM.
  Reply With Quote
Old 7th March 2013, 04:10 AM   #18
diyAudio Member
 
abraxalito's Avatar
 
Join Date: Sep 2007
Location: Hangzhou - Marco Polo's 'most beautiful city'. 700yrs is a long time though...
Blog Entries: 109
Send a message via MSN to abraxalito Send a message via Yahoo to abraxalito Send a message via Skype™ to abraxalito
Quote:
Originally Posted by agdr View Post
Apparently the feedback loop with the OPA627 is working extremely well to null that remaining NJM4556 input offset, even though there are 4 op amps now on each channel over two packages.
Pleased to hear that's doing fine - RS swore blind that paralleling opamps in more than one package could never work due to offset mismatch
__________________
Seek not the favour of the multitude...rather the testimony of few. And number not voices, but weigh them. - Kant
The capacity for impartial observation is commonly called 'cynicism' by those who lack it.
  Reply With Quote
Old 7th March 2013, 09:48 PM   #19
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
Quote:
Originally Posted by abraxalito View Post
Pleased to hear that's doing fine - RS swore blind that paralleling opamps in more than one package could never work due to offset mismatch
Lol! Yeah I pitched more NJM4556A's to him also once via PM or email a year or two ago and got pushback too, but no specifics other than he didn't think it was workable. I'm guessing one problem may have been his 40.2K return resistors. With 4 op amps rather than two that input bias current would double and the voltage drop across the 40.2k would double. The DC offset of the O2 may have gone up to 6mV.

This thing also proves that a FET input on the pot wiper is as good as a capacitor-coupled bipolar op amp input, in terms of keeping DC out of the pot wiper. I'm getting absolutely zero pot turning noise, just as with the O2. I really expected that would work since the pico-amp FET bias current is probably about the same as the leakage current of the O2's coupling cap.

I was a bit concerned about what the pot/FET design would do from a DC standpoint, since turning the pot changes the impedance looking out the non-inverting OAP627 pin from 0R to 5k, vs. less than 1k looking out the inverting input at the feedback network. But in real life measurements last night the DC effect of turning the pot is a couple of micro-volts variation in the amp output DC offset, which then stabilizes after a second or two. A non-event due to that miniscule pico-amp bias current on both inputs.

I just managed to wedge in 4 film coupling caps on each channel on the current layout. That will drop the corner frequency on the input LPF down below that of the O2 by 3Hz or so. One of my goals with this thing is to try to slightly exceed the O2 parameters in a few places, at least those that I can measure/verify with a DMM and scope. I'm thinking about setting the latest up with a single DIP8 socket for the gain amp(s) and use a dual-soic-to-DIP adapter to use 2 OPA827s instead of the 627. I didn't know those things (dual soic adapters) existed until a posting in the O2 thread a week or two ago. That should result in a further small upgrade. The OPA827's only come in SOIC last I looked. But still the dual assembly will be in socket in case one or both fry due to static electricity during assembly, for easy replacement. Either that or do the same with 2 soic OPA627's. The adapter and single DIP8 socket would also save one socket of layout space.

I'm still planning on having the final layout done as 4 layer (this one is just 2 layer and still is completely hiss-free!) with one layer as nothing but a full ground plane, single-point grounded at the output jack ground connection which is essentially the star point. The hope/expectation is the ground plane will help with all those parameters I can't measure, like distortion.

Also thinking of adding a clipping indicator LED. I've always thought that is something the O2 needed.

Last edited by agdr; 7th March 2013 at 10:03 PM.
  Reply With Quote
Old 8th March 2013, 09:23 AM   #20
weslito is offline weslito  United Kingdom
diyAudio Member
 
weslito's Avatar
 
Join Date: Sep 2010
Quote:
Originally Posted by agdr View Post
I'm guessing one problem may have been his 40.2K return resistors. With 4 op amps rather than two that input bias current would double and the voltage drop across the 40.2k would double.
How on earth did you figure that one out?
  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 37 20th October 2014 05:46 PM
Help with BantamDAC and O2 Amp jschristian44 Headphone Systems 2 27th January 2012 06:22 PM
O2 Headphone amp for sale! agdr Vendor's Bazaar 10 9th December 2011 04:05 AM
o2 amp building jtktam Vendor's Bazaar 0 3rd November 2011 10:52 AM
Questions before getting parts for the O2 Amp dewasiuk Headphone Systems 3 12th September 2011 05:33 PM


New To Site? Need Help?

All times are GMT. The time now is 03:10 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