Before proceeding to solder one has to apply some sort of insulation over the via(piece of tape,girl-friend's nail polish) so that the body of the switch doesn't touches the via.Since the switch is already soldered try raising it a bit(very carefully or you may damage the tracks) so that it doesn't touches the via.
BTW I've used jumpers (gain= 1X, 2X & 8X) to avoid this problem.
see #3147
http://www.diyaudio.com/forums/head...eadphone-amp-diy-project-315.html#post3377815
Thanks! I'm on the process of fixing, but is having a hard time getting the switch off the board (so I can cut off the bottom left leg (closest to the switch head), correct?).
Also, one little question. The sound on my self built O2 seems to be a little muffled and less detailed compared to a stock O2 amp. Are there any suggestions of what may be wrong? Perhaps too much solder, or some components are not actually making contact, but instead making contact through the solders? Or perhaps because I cut the power jack because it won't fit, and the electricity might be flowing through the solder onto the board instead of directly onto the board?
I noticed something strange while testing my O2. With the amp switched off, putting a sine wave on the input produces a DC voltage at the output. The DC increases in proportion with the input voltage. Using a 1KHz 2.5Vp-p sine wave produces -1.53VDC with around 500 ohms output impedance. The strange thing is, this happens even with the volume at minimum.
Anyone got any clue what's going on here? Seems very strange that it's doing this with no power and volume at minimum.
Edit: It's also worst in high gain mode (5X). Low gain (1X) gives around -0.9V.
Anyone got any clue what's going on here? Seems very strange that it's doing this with no power and volume at minimum.
Edit: It's also worst in high gain mode (5X). Low gain (1X) gives around -0.9V.
Attachments
Last edited:
Atilla ran into that problem in Feb of last year, post #2012 in this thread. There are some more details in the posts after that one. I don't think the cause and resolution were ever found. My best guess was something in the NJM2068 chip was rectifying the incoming signal, even with no power to the chip, then sending that back out the power pins where C8 and C9 then charged up to some DC voltage, partially powering on the output NJM4556A chips.
With the power to the O2 off the mosfets would be off, which would cut that output power rail segment off from the power managment circuit and voltage regulators (load), leaving C8 and C9 as the defacto power rail filter caps for the resulting isolated power rail segment past the mosfets.
You have a good test setup there! I would be curious what the voltages at pin 4 and 8 (power pins) to ground are on the NJM2068 when it happens. Even better, it would be interesting to see a waveform to ground on those pins. I wouldn't be surprised if it is rectified 1kHz, being smoothed then by C8 and C9. Might also just be a single ended voltage being produced, with the v- pin as essentially ground and the rectified 1Khz coming out the V+ pin, or vice versa.
Last edited:
weslito - putting 1N4148 diodes on the NJM2068 power pins as below might be one fix if that turns out to be the problem. The diodes would block current from flowing in the reverse direction when the power rails are off. I picked silicon 1N4148's here instead of Schottky's since the reverse current leak is so much great with the Schottky's.
Just bending the power pins up on a sacraficial NJM2068 and tack soldering diodes to them would be a way to test it without having to cut any traces. Then the free ends of the diode lead stuff back down into the IC socket holes. I forgot that was how I tested the NJM4556 feedback resistor mod.
wiinippongamer also just let me know that a fix I came up with in January for power management turn-off transients he was getting while on batteries has worked after 6 weeks of testing. Just a 10K resistor soldered across the power rails at C8 and C9 to put a small load on that post-mosfet segment once the mosfets cut off. The 10K acts as a bleed resistor for those caps.
The theory was that the NJM4556A and NJM2068 chips are only specified down to about 2.5Vdc, so anything below that as the rails bleed down is kind of no man's land. I was thinking the chips may cut off their power draw hard below 2.5Vdc and leave the rails charged, without a bleed resistor, and that appears to be the case. Adding that 10K bleed load might also help with this problem.
Just bending the power pins up on a sacraficial NJM2068 and tack soldering diodes to them would be a way to test it without having to cut any traces. Then the free ends of the diode lead stuff back down into the IC socket holes. I forgot that was how I tested the NJM4556 feedback resistor mod.
wiinippongamer also just let me know that a fix I came up with in January for power management turn-off transients he was getting while on batteries has worked after 6 weeks of testing. Just a 10K resistor soldered across the power rails at C8 and C9 to put a small load on that post-mosfet segment once the mosfets cut off. The 10K acts as a bleed resistor for those caps.
The theory was that the NJM4556A and NJM2068 chips are only specified down to about 2.5Vdc, so anything below that as the rails bleed down is kind of no man's land. I was thinking the chips may cut off their power draw hard below 2.5Vdc and leave the rails charged, without a bleed resistor, and that appears to be the case. Adding that 10K bleed load might also help with this problem.
Attachments
Last edited:
Two things, is there an output jack that fits the pcb can accept TRRS plugs? I drives me nuts sometimes.
Any ideas to bring down the volume to attenuate the signal? With my current IEMs I can't imagine going over 12 o'clock on the volume pot. So I need about 0.1 gain / 10x attenuation. Guess I'll measure the pot before to determine the right value.
My goals: sustaining good quality (noise, distortion) > easily to do without to much cutting traces etc. >> can be neutralized by the gain switch.
There are two solutions I can think of.
- increase R7 and R3 to 9k, decrease C11 and C12 to a value that fits both low and high gain (22pf) and use R17 and R21 to switch 1k parallel to R14 and R20.
- put 9k between the pot and pin 1 and 7 of the NJM2068, use R17 and R21 to switch 1k parallel to the pot.
Any other solutions? Which one is best?
Any ideas to bring down the volume to attenuate the signal? With my current IEMs I can't imagine going over 12 o'clock on the volume pot. So I need about 0.1 gain / 10x attenuation. Guess I'll measure the pot before to determine the right value.
My goals: sustaining good quality (noise, distortion) > easily to do without to much cutting traces etc. >> can be neutralized by the gain switch.
There are two solutions I can think of.
- increase R7 and R3 to 9k, decrease C11 and C12 to a value that fits both low and high gain (22pf) and use R17 and R21 to switch 1k parallel to R14 and R20.
- put 9k between the pot and pin 1 and 7 of the NJM2068, use R17 and R21 to switch 1k parallel to the pot.
Any other solutions? Which one is best?
Can I borrow your ATF20B?
Strange but intrestingly- I've got many books but no book-case! Nobody lends me a book-case
Namenlos, have you are already configured it for unity gain by removing R17 and R21? If so, the easiest way to achieve even lower gain would be swapping R7 and R3 for a higher value and resizing C11 and C12 appropriately, as you say.
You can calculate the voltage gain this would give you with:
R14/(R14+R7)
Using 9K will give you:
10000/(9000+10000) = 0.53
You can calculate the voltage gain this would give you with:
R14/(R14+R7)
Using 9K will give you:
10000/(9000+10000) = 0.53
book·case [book-keys]
noun
a set of shelves for books.
Origin:
1720–30; book + case2
Bookcase | Define Bookcase at Dictionary.com
Pls see attch. image(The offer does not include the beautiful model)
Attachments
Increasing R7, R3 will increase the Johnson noise according to RS (critical in gain stage). You can solder 2 pairs of resistors in the empty 6 holes of VR1 vol. pot such that they are in parallel with VR1. One can start with 4.7K+4.7K or any other suitable configuration(not to go below total combination of 2K, the lower limit for JRC2068 loading).One can get a rough idea of the value of voltage divider resistors by measuring the resistance of VR1 w.r.t. vol. knob position.
Attachments
book·case [book-keys]
noun
a set of shelves for books.
Origin:
1720–30; book + case2
Bookcase | Define Bookcase at Dictionary.com
Pls see attch. image(The offer does not include the beautiful model)
I don't have a bookcase, but I do have a benchcase
Attachments
Then I would have to add a resistor between the pot and the NJM2068 right?
ODAC with O2 grounding
Hi all, first post... ...this thead is something of an epic, but I did try the search function. If this has already been asked before, I apologize.
I just received an ODAC to use with my O2, and I'm pondering how to enclose it. NWAVGuy seems adamant both in the project notes and in the schematic that the O2 should be connected to the enclosure at one point only (the input jack), presumably to avoid ground loops. However, if I fit the ODAC into the same case and then expose the USB socket via a hole cut in a metal panel, it seems that if the USB socket touches the panel the O2 is now connected to the enclosure at both the USB jack (via the signal ground wire from the ODAC) and via the 'regular' thin wire front-panel connection. Is that right? If so, is this not really so important after all?
I also have a couple of general questions about grounding in an amp like this. I'm not clear if the Triad AC/AC adapter's secondary is grounded at one end, or if the whole arrangement floats. But in any case, I'm assuming that on battery operation with a battery-powered source everything must be floating with respect to to mains earth - unless a mains-powered source is connected. If I run on batteries and then connect a USB source, for example, signal ground is now referenced to the USB ground of the source. Is that correct?
Thanks. I'm more than happy to do the reading / thinking myself if someone wants to point me in the right direction!
Hi all, first post... ...this thead is something of an epic, but I did try the search function. If this has already been asked before, I apologize.
I just received an ODAC to use with my O2, and I'm pondering how to enclose it. NWAVGuy seems adamant both in the project notes and in the schematic that the O2 should be connected to the enclosure at one point only (the input jack), presumably to avoid ground loops. However, if I fit the ODAC into the same case and then expose the USB socket via a hole cut in a metal panel, it seems that if the USB socket touches the panel the O2 is now connected to the enclosure at both the USB jack (via the signal ground wire from the ODAC) and via the 'regular' thin wire front-panel connection. Is that right? If so, is this not really so important after all?
I also have a couple of general questions about grounding in an amp like this. I'm not clear if the Triad AC/AC adapter's secondary is grounded at one end, or if the whole arrangement floats. But in any case, I'm assuming that on battery operation with a battery-powered source everything must be floating with respect to to mains earth - unless a mains-powered source is connected. If I run on batteries and then connect a USB source, for example, signal ground is now referenced to the USB ground of the source. Is that correct?
Thanks. I'm more than happy to do the reading / thinking myself if someone wants to point me in the right direction!
Last edited:
Oh, and one more before I forget, although this is really a 'pure ODAC' question... ...the traces from the line-out header to the pads for the 3.5mm socket on the ODAC board have jumpers on the PCB that need to be bridged with solder to get the signal to the 3.5mm socket. Why not just run the traces to the 3.5mm socket and the line-out header with no jumper? If the 3.5mm socket is not fitted, what's the disadvantage to having the unused pads still connected to the output signal?
This can't work, you just increase the load of the preamp. You have to create a voltage divider somehow.
The easiest way is to increase the values of R3 and R7, which form voltage dividers with R20 and R14, as described here:
http://www.diyaudio.com/forums/head...-crc-diode-cap-heatsink-mods.html#post2794079
I find that I need around 10K for 50% attenuation with my sources. When you increase those resistor values you have to correspondingly decrease the values of C11 and C12 to keep the corner frequency of the input RF filter the same, as the link describes.
Below is the AC sim plot of the O2 amp RF filter with the original values. The second plot is with R3 and R7 increased to 10k, for a 50/50 voltage divider, and C11 and 12 decreased to 22pF. The capacitors should be 50Vdc MLCC COG types with 2.5mm lead spacing, like Mouser part number 594-k220J15C0GF53L2.
Ideally it would be best to add a 10K resistor in series between the first op amp and the 10k pot, since the buffer stage following has 1x closed loop gain and won't increase the resistor noise. But that requires cutting PCB traces. The assumption here is that you are probably using 1x gain anyway in the first stage since you need attenuation, so even that won't matter. Easiest just to replace those two existing resistors and caps.
Attachments
Last edited: