Hey guys, sorry I haven't made it back in a timely fashion! Been traveling a bit and logging in remotely plus a couple of other projects in the works. I should be able to log in tomorrow and give a better reply.
fmcclell - I'm sorry about the delayed reply!
I just realized there is a much easier way to do what you are shooting for. Since you want your pre-amp output to simply be the output of the pot (after the coupling capacitor), you can just connect the pre-amp RCA jacks (via the JP21 connector with R62 and R57 removed) directly to the output of the amp itself. That works because the output stage of the amp is just a 1x current buffer - the output voltage of the entire amp is the same as what is after the pot. Plus it has that DC offset null circuit so you will get the low DC offset, and since it is the output of the amp no problem driving capacitive loads.
But here is the really slick part. You can do what you want to do with just one panel-mounted DPDT switch at this point (or just use two of the 3 poles in that 3PDT you've already ordered). 🙂 The trick is those two jumpers in the "R88" and "R89" positions on either side of the 3.5mm output jack, layout picture below. Those resistor positions are for an optional damping factor resistor on each channel, which is just a resistor in series from the output of the headphone relay to the two output jacks. If not using a damping factor resistor those two are just jumpered across.
So what you can do is first remove those two jumpers. Then run wires from the bottom end of those "R88" and "R89" positions (the end toward the relay) to the center terminals of your DPDT switch. Those lines are the output of the headphone relay, which is the output of the amplifier. Then on one side of the DPDT wire that to the outer two holes of JP21. Remove the R62 and R57 SMD resistors under the board to break the connection to the output of the pre-amp chip, since it won't be used here at all. Those two resistors are used for what we were going to do, add series resistance to the output of the pre-amp chip to isolate output capacitance. But here they work just fine to break the connection between the pre-amp chip and J21 entirely. So when the DPDT is flipped to that side the output of the ODA goes to your pre-amp outs and on to the speakers.
You will also need to add a jumper across the "R90" position on the bottom of the board, if it doesn't already have a zero-ohm resistor in it. That just grounds the ground side of the pre-amp jack. I can't remember if I put a zero ohm resistor there or not on your board, but an easy way to to it is just put solder on both pads and solder the end of a small gauge wire right across them.
Then on the other side of the DPDT switch wire that back to the other two holes in the "R88" and "R89" positions to feed the ODA output back on its way to the headphone output jacks. With the DPDT in that position it is just like having the jumpers in for R88 and R89 again.
You probably won't get any thumps when you toggle the new DPDT switch for speakers/headphone since the DC output on the amp is pretty much nulled out and the circuitry is already thermally stabilized from being on. If you do get a thump for any reason a way around that is turn the ODA off before switching the new DPDT. That will give you the 3 second turn-on delay and accelerated turn-of for either speakers or headphones.
Another reason you probably won't get thumps is R84 and R85, the minimum load resistors I've included in the amp circuit. Those are 2.49K loads on the amp outputs at all times, even when the relay is switched off. That helps prevent DC offset wander during output switching.
Please let me know if this sounds like it will achieve what you are after! It turns out that taking the signal off right after the capacitor would be a bit of work since all the traces are under the PC board. A small wire could be soldered to one leg of the coupling capacitor on each channel, but again since the output stage is a 1x current buffer the output of the amp is the same signal as what is out of the coupling caps.
You won't need the pre-amp chip at all with this method. No SMD chip to change! 🙂 In general though your method for removing those (cut each lead then desolder the lead remnants) works great. I've done that several times over the years. Chip Quick also does a good job.
I just realized there is a much easier way to do what you are shooting for. Since you want your pre-amp output to simply be the output of the pot (after the coupling capacitor), you can just connect the pre-amp RCA jacks (via the JP21 connector with R62 and R57 removed) directly to the output of the amp itself. That works because the output stage of the amp is just a 1x current buffer - the output voltage of the entire amp is the same as what is after the pot. Plus it has that DC offset null circuit so you will get the low DC offset, and since it is the output of the amp no problem driving capacitive loads.
But here is the really slick part. You can do what you want to do with just one panel-mounted DPDT switch at this point (or just use two of the 3 poles in that 3PDT you've already ordered). 🙂 The trick is those two jumpers in the "R88" and "R89" positions on either side of the 3.5mm output jack, layout picture below. Those resistor positions are for an optional damping factor resistor on each channel, which is just a resistor in series from the output of the headphone relay to the two output jacks. If not using a damping factor resistor those two are just jumpered across.
So what you can do is first remove those two jumpers. Then run wires from the bottom end of those "R88" and "R89" positions (the end toward the relay) to the center terminals of your DPDT switch. Those lines are the output of the headphone relay, which is the output of the amplifier. Then on one side of the DPDT wire that to the outer two holes of JP21. Remove the R62 and R57 SMD resistors under the board to break the connection to the output of the pre-amp chip, since it won't be used here at all. Those two resistors are used for what we were going to do, add series resistance to the output of the pre-amp chip to isolate output capacitance. But here they work just fine to break the connection between the pre-amp chip and J21 entirely. So when the DPDT is flipped to that side the output of the ODA goes to your pre-amp outs and on to the speakers.
You will also need to add a jumper across the "R90" position on the bottom of the board, if it doesn't already have a zero-ohm resistor in it. That just grounds the ground side of the pre-amp jack. I can't remember if I put a zero ohm resistor there or not on your board, but an easy way to to it is just put solder on both pads and solder the end of a small gauge wire right across them.
Then on the other side of the DPDT switch wire that back to the other two holes in the "R88" and "R89" positions to feed the ODA output back on its way to the headphone output jacks. With the DPDT in that position it is just like having the jumpers in for R88 and R89 again.
You probably won't get any thumps when you toggle the new DPDT switch for speakers/headphone since the DC output on the amp is pretty much nulled out and the circuitry is already thermally stabilized from being on. If you do get a thump for any reason a way around that is turn the ODA off before switching the new DPDT. That will give you the 3 second turn-on delay and accelerated turn-of for either speakers or headphones.
Another reason you probably won't get thumps is R84 and R85, the minimum load resistors I've included in the amp circuit. Those are 2.49K loads on the amp outputs at all times, even when the relay is switched off. That helps prevent DC offset wander during output switching.
Please let me know if this sounds like it will achieve what you are after! It turns out that taking the signal off right after the capacitor would be a bit of work since all the traces are under the PC board. A small wire could be soldered to one leg of the coupling capacitor on each channel, but again since the output stage is a 1x current buffer the output of the amp is the same signal as what is out of the coupling caps.
You won't need the pre-amp chip at all with this method. No SMD chip to change! 🙂 In general though your method for removing those (cut each lead then desolder the lead remnants) works great. I've done that several times over the years. Chip Quick also does a good job.
Attachments
Last edited:
agdr, Thanks, good to hear from you again -- we were starting to worry 🙂 No worries though, I understand about being busy. Sounds like what you are suggesting should work fine -- I'll have to sketch it up, but should be an easy way to wire in the circuit. I think I had looked at wiring it similar to what you suggested at first, but then started thinking about it and wound up with the other approach. But your suggestion looks like it will be easier to implement and should do everything that I need.
I just got some time tonight and am starting to solder the board. Just came back to the computer to look up where one of the parts goes (R70) as it was not obvious to me from the silkscreen. I remember seeing that you have posted some excellent pics of the assembly that I was going to try to find. That is when I saw your post here. Thanks again. I'll let you know if I have any other problems.
I just got some time tonight and am starting to solder the board. Just came back to the computer to look up where one of the parts goes (R70) as it was not obvious to me from the silkscreen. I remember seeing that you have posted some excellent pics of the assembly that I was going to try to find. That is when I saw your post here. Thanks again. I'll let you know if I have any other problems.
fmcclell - those build photos are in posts #278 - 290, #301, #302, and #368. I have the first part of the #1 post in this thread updated with a bunch of current info like that, if anyone is looking for the links to the Google Drive files and other stuff. 🙂 Keep that "part ID diagram" in mind too out on the Google Drive link. I have a yellow circle around each part and its part number to make it clearer which number goes with what part. The build instructions also call out each part individually.
In case you are still looking for that R70 it mounts on end and is between the relay and the IC11 regulator, right next to that reg chip.
Please do post or PM if any build questions come up!
In case you are still looking for that R70 it mounts on end and is between the relay and the IC11 regulator, right next to that reg chip.
Please do post or PM if any build questions come up!
Hi folks,
so I have been testing various opamps in my O2 during the past 2,5 years and here are my subjective observations (I have spent many many hours with each of the opamp listed below). Maybe somebody will find it useful... But as usual, YMMV 😉
- stock 4556/2068: sharp sounding with unnatural piercing highs at times
- 3x NE5532 (DIP): extremely boring, neutral
- 3x NE5534 (dual browndog): boring, very similar to NE5532
- 3x LME49720 (DIP): sounds too good, almost Hi-Fi sounding
- 3x AD8599: there's something really wrong with the highs (first time I've ever heard grainy sound)
- 3x LME49720 (TO-99, original from TI): unusable, picks RFI as hell
- 3x OPA1611 (dual browdog): very midrange-focused, unnatural at times (I have yet to try OPA1612)
- 3x LME49990 (dual browndog): balanced, with a slight touch of LME49720 character at times
- 3x OPA1662 (browndog): very musical, very big soundstage (probably the biggest I've ever heard), overall balanced/natural presentation
- 3x OPA827 (dual browndog): natural, similar character as OPA1662, but with smaller soundstage
- 3x MAX9632 (dual browndog): boring, neutral, almost reminds me of NE5532
My personal favourites (top to bottom) are currently OPA1662, OPA827, LME49990, MAX9632 and NE5532.
M.
so I have been testing various opamps in my O2 during the past 2,5 years and here are my subjective observations (I have spent many many hours with each of the opamp listed below). Maybe somebody will find it useful... But as usual, YMMV 😉
- stock 4556/2068: sharp sounding with unnatural piercing highs at times
- 3x NE5532 (DIP): extremely boring, neutral
- 3x NE5534 (dual browndog): boring, very similar to NE5532
- 3x LME49720 (DIP): sounds too good, almost Hi-Fi sounding
- 3x AD8599: there's something really wrong with the highs (first time I've ever heard grainy sound)
- 3x LME49720 (TO-99, original from TI): unusable, picks RFI as hell
- 3x OPA1611 (dual browdog): very midrange-focused, unnatural at times (I have yet to try OPA1612)
- 3x LME49990 (dual browndog): balanced, with a slight touch of LME49720 character at times
- 3x OPA1662 (browndog): very musical, very big soundstage (probably the biggest I've ever heard), overall balanced/natural presentation
- 3x OPA827 (dual browndog): natural, similar character as OPA1662, but with smaller soundstage
- 3x MAX9632 (dual browndog): boring, neutral, almost reminds me of NE5532
My personal favourites (top to bottom) are currently OPA1662, OPA827, LME49990, MAX9632 and NE5532.
M.
mireque - thank you for sharing your listening impressions! 🙂
I agree about the LM49720, it just doesn't sound quite right to me in anything I've tried it with. Interesting about the RF pickup with the metal can TO-99 version. Someone has posted about that before. Grounding the can may solve it, but only if National doesn't have the chip substrate or one of the leads connected to the can. I have a few of the LM49720HA here but haven't had much of a chance to mess with them.
I also agree about the OPA827! That is turning out to be a fantastic chip in my opinion. The 827 is one of the two chips I use in the O2 Booster Board project, along with the OPA140. Lately I've been building those with the OPA827 as standard unless the goal is absolute minimum power draw.
The LME49990 also has been great sounding in the ODA, to my ears anyway, and I use two of them on an adapter to replace the NJM2068 in O2's with Booster Boards these days. opc uses that one in his Wire amp and the posted listening reviews have generally been really good. I agree though, I would strongly suspect that the LME49990 is a derivative of the LME49720, especially given the historical time frame when it was introduced. Maybe the result of a project at National to "fix" a few things about the LME49720.
Interesting comments about the OPA1662! I haven't tried that one. In fact, I'm not sure I knew it existed. I'll get some and try them out. I think I've been getting it mixed up with those OPA1611/1612s, which seem to be re-branded OPA211's. Maybe the reason I haven't done more with the 1662 is not being available in a single package, so it isn't a pin compatible drop in for the LME49990.
I agree about the LM49720, it just doesn't sound quite right to me in anything I've tried it with. Interesting about the RF pickup with the metal can TO-99 version. Someone has posted about that before. Grounding the can may solve it, but only if National doesn't have the chip substrate or one of the leads connected to the can. I have a few of the LM49720HA here but haven't had much of a chance to mess with them.
I also agree about the OPA827! That is turning out to be a fantastic chip in my opinion. The 827 is one of the two chips I use in the O2 Booster Board project, along with the OPA140. Lately I've been building those with the OPA827 as standard unless the goal is absolute minimum power draw.
The LME49990 also has been great sounding in the ODA, to my ears anyway, and I use two of them on an adapter to replace the NJM2068 in O2's with Booster Boards these days. opc uses that one in his Wire amp and the posted listening reviews have generally been really good. I agree though, I would strongly suspect that the LME49990 is a derivative of the LME49720, especially given the historical time frame when it was introduced. Maybe the result of a project at National to "fix" a few things about the LME49720.
Interesting comments about the OPA1662! I haven't tried that one. In fact, I'm not sure I knew it existed. I'll get some and try them out. I think I've been getting it mixed up with those OPA1611/1612s, which seem to be re-branded OPA211's. Maybe the reason I haven't done more with the 1662 is not being available in a single package, so it isn't a pin compatible drop in for the LME49990.
Last edited:
OPA1662 is TI's recent design for audio applications, from 2011/2012.
And the another plus for OPA1662 is its power consumption - only 1.5mA per channel, so I have been able to run my O2 from batteries for roughly 3 working days (3x8).
I highly recommend this opamp - it's somehow musical, very detailed, balanced, controlled and tight, clean, and I still wonder about that huge 3D soundstage. Also it doesn't make everything sound right - flaws in recordings are immediately obvious...
M.
And the another plus for OPA1662 is its power consumption - only 1.5mA per channel, so I have been able to run my O2 from batteries for roughly 3 working days (3x8).
I highly recommend this opamp - it's somehow musical, very detailed, balanced, controlled and tight, clean, and I still wonder about that huge 3D soundstage. Also it doesn't make everything sound right - flaws in recordings are immediately obvious...
M.
Either
SO8 to 8-pin DIP Adapter (p/n 970601A)
or
(LME49880 compatible)2013 Version SOIC to DIP-8 Convert SMD PCB Adapter 49pcs - DIYINHK
I have experience with both and they're just fine.
M.
SO8 to 8-pin DIP Adapter (p/n 970601A)
or
(LME49880 compatible)2013 Version SOIC to DIP-8 Convert SMD PCB Adapter 49pcs - DIYINHK
I have experience with both and they're just fine.
M.
While these test results are certainly interesting, why would one use the same type of opamp in all three positions, rather than looking at one stage a a time (their demands are fairly different)? Also, would mentioning headphones used, source levels, gain setting and mains/battery operation really be too much to ask for?
I don't think plugging in random opamps is such a good idea. While 4556s get along with 1 ohm combining resistors, other types may be less than happy like that depending on load. I would not use 5532s with less than about 47 ohms, the MAX9632 datasheet recommends isolation resistors for >100 pF as well. (I assume that some of the sonic phenomena noted were due to spurious oscillation - the "too good" sound of the '49720 may be high levels of 2nd harmonic. And then the higher parasitics for the TO-99 case totally pushed things over the edge.)
Granted, the '2068 breaks a bit of a sweat at high gains and can't get as close to the rails as some other types (i.e. clips a bit earlier), but if in doubt you can just install a '5532 there instead, or whichever type you like as a line-level amp, or just ease up on feedback network resistor values a bit (while maintaining decent impedance balance). It is quite clear that the '2068 is a fair bit happier with the 2.5 kOhms at 2.5x gain when compared to the 1k7-ish at 6.5x, which is heavier loading at reduced spare loop gain (no wonder the '5532 with its more robust load driving abilities does better then). IMO it's quite difficult to make full use of the low-noise properties of this chip at low gains unless sticking with a unity-gain buffer or adding a buffer within the loop. Not like the O2 gain stage would need 'em, source noise is going to swamp everything anyway...
The one result that does not particularly surprise me is the AD8599. Samuel Groner's testing shows highish distortion levels, with the 600 ohm load in particular (certainly much higher than in the TI NE5532). If it doesn't even like 600 ohms very much, we can assume that lower load impedances will get progressively worse. Definitely not a good candidate for the output buffer stage.
The OPA1662 is quite impressive for its moderate current draw (3 mA for stereo), but I don't think it can seriously replace the '4556 in the output buffer. I would extrapolate 6 Vrms, 150 ohm THD at 10 kHz to be 0.002-0.004%, probably dominant 3rd (the stock O2 sits at about 0.002%, dominant 2nd). With its good performance into 2 kOhms and rail-to-rail output, it should make a brilliant gain stage amp though.
I don't think plugging in random opamps is such a good idea. While 4556s get along with 1 ohm combining resistors, other types may be less than happy like that depending on load. I would not use 5532s with less than about 47 ohms, the MAX9632 datasheet recommends isolation resistors for >100 pF as well. (I assume that some of the sonic phenomena noted were due to spurious oscillation - the "too good" sound of the '49720 may be high levels of 2nd harmonic. And then the higher parasitics for the TO-99 case totally pushed things over the edge.)
Granted, the '2068 breaks a bit of a sweat at high gains and can't get as close to the rails as some other types (i.e. clips a bit earlier), but if in doubt you can just install a '5532 there instead, or whichever type you like as a line-level amp, or just ease up on feedback network resistor values a bit (while maintaining decent impedance balance). It is quite clear that the '2068 is a fair bit happier with the 2.5 kOhms at 2.5x gain when compared to the 1k7-ish at 6.5x, which is heavier loading at reduced spare loop gain (no wonder the '5532 with its more robust load driving abilities does better then). IMO it's quite difficult to make full use of the low-noise properties of this chip at low gains unless sticking with a unity-gain buffer or adding a buffer within the loop. Not like the O2 gain stage would need 'em, source noise is going to swamp everything anyway...
The one result that does not particularly surprise me is the AD8599. Samuel Groner's testing shows highish distortion levels, with the 600 ohm load in particular (certainly much higher than in the TI NE5532). If it doesn't even like 600 ohms very much, we can assume that lower load impedances will get progressively worse. Definitely not a good candidate for the output buffer stage.
The OPA1662 is quite impressive for its moderate current draw (3 mA for stereo), but I don't think it can seriously replace the '4556 in the output buffer. I would extrapolate 6 Vrms, 150 ohm THD at 10 kHz to be 0.002-0.004%, probably dominant 3rd (the stock O2 sits at about 0.002%, dominant 2nd). With its good performance into 2 kOhms and rail-to-rail output, it should make a brilliant gain stage amp though.
Last edited:
Thanks for your input. In my experience the time period needed to seriously compare one audio gear to another is about 1 to 2 weeks before I can get fully accustomed to the various sound signatures. Also comparing 10+ opamps in one day is silly, just because people can get tired and then brain perception can get seriously skewed...
As for the setup - source impedance 30 Ohms (from an RME D/A converter). O2 gain settings 1.0x (I have 1.0/2.5x) with 99% of the time powered by batteries. O2's volume pot set exactly to "9 o'clock". Sony MDR 7520 headphones @ 24 Ohms used, always at default gain of 1.0x. Listening in a silent room with no external disturbances occurring and I've tried to have fresh ears / brain.
I didn't observe anything wrong with NE5532 / MAX9632, just that they always sounded to me as extremely flat, smooth and boring with 95% of music I threw on them. But for monitoring / sound engineering purposes I think this type of sound signature is preferred, so I highly recommend these opamps for such purposes (mastering / monitoring). I keep wondering about NE5532, albeit its very old design, to sound this flat.
You have a good point on 2nd level harmonic with LME49720, it seems to make sense. While LME49990 was coming close to MAX9632 in sound signature, at times I just felt it's making some music I know really well sound better than it really is (i.e. that slight hint of 49720 "Hi-Fi sound").
I still don't understand why OPA1662 sounds to me that huge (audiophiles would call this "3D") in comparison to other opamps I've tried... But it's very very musical and very engaging sound signature, but still managing to sound bad with bad music. Dunno what to think about this opamp.
I have yet to try some more opamps, will report again sometime...
I am getting a blank board soon and was wondering if anyone has a recommendation if I should build the 15v version over the 12v for Mad Dog Pros. The specs list them only at 50 ohms, but they can handle up to 3 watts if I recall correctly. Would I get benefit for the 15v or would I be risking damaging them? I am having trouble finding the full specs right now.
Congratulations on your decision build up an ODA! 🙂 It's a fun build.
You are right, those tech specs for the MD Pros are hard to find. I finally located them here. Scroll down a ways then click on the "additional information" tab. 50 ohms and 98dB/mW sensitivity.
Plugging that into the spreadsheet attached below gives 7.07Vrms to hit 120dB, an "absolute maximum peak" dynamic range number that is often used for headphones. So you would be best off with the standard +/-12.5Vdc version of the ODA which can swing up to 7.25Vrms. The +/-15Vdc build also can only be used with 300 ohm or higher headphones, to stay withing power dissipation limits for the output chips.
Interesting to note from the spreadsheet that the current at the 7.07Vrms peak is 141mA, officially just a hair over the maximum for the O2 amp. But I'm increasingly convinced that having at least 2x the maximum current capability needed in the headphone amp makes a sound difference (better), but I'm not sure why technically yet. That one is purely subjective from lots of listening (lol - I'm sure NwAvGuy would bang his head on the table here, being 100% objective with everything 😀 ). The ODA is good for over 450mA per channel at low THD+N, which actually has been measured by the one set of dScope measurements that were done.
Last edited:
Hey Turbon,
Hows the ODA build going?
I didn't read anything lately from you on it.
A couple guys over at headfi asked me about boards etc...told them to check
out this thread amd pm agdr!
All is well here in NC the ODA is working very well for hours at a time daily.
Alex
Hows the ODA build going?
I didn't read anything lately from you on it.
A couple guys over at headfi asked me about boards etc...told them to check
out this thread amd pm agdr!
All is well here in NC the ODA is working very well for hours at a time daily.
Alex
With good reason. You could try a DBT - add additional holes and switches to activate/deactivate the current buffers.
It'd be more believable that noise is proportional to sqrt(current amplification), though no idea if that has any basis in reality.
Hi Alex! I'm glad you posted, I've been meaning to PM Turbon myself to see how things are going. 🙂
I do have bare ODA PC boards (still the same V2.1, no changes) and I even have one built up with the SMD-only parts soldered on, leaving the 1/2 that is through-hole for DIY. If anyone is interested just PM me. 🙂 All the project materials are in the Google Drive link in the first post of this thread - schematic, layout, Part ID diagram, photos (lists the posts numbers of the photo build here in the thread too), build instructions, BOM, etc.
I do have bare ODA PC boards (still the same V2.1, no changes) and I even have one built up with the SMD-only parts soldered on, leaving the 1/2 that is through-hole for DIY. If anyone is interested just PM me. 🙂 All the project materials are in the Google Drive link in the first post of this thread - schematic, layout, Part ID diagram, photos (lists the posts numbers of the photo build here in the thread too), build instructions, BOM, etc.
Hey Moragg! One thought is it may have something to do with the amount of (negative) feedback needed to correct the output signal. The large transistors in big current buffers (or many paralleled small transistors here) may need less correction at currents that are 1/3 - 1/2 into their range vs. smaller transistors running near their full-on point, with the typical class AB output stage. Just a guess though! 🙂
But you are right, switching some of the paralleled buffers off and on would make for an interesting listening test. The S/N is should improve with the paralleled chips since the signal adds linearly while the noise adds as RMS.
But you are right, switching some of the paralleled buffers off and on would make for an interesting listening test. The S/N is should improve with the paralleled chips since the signal adds linearly while the noise adds as RMS.
Last edited: