Finding a short between the negative rail and the ground may be somewhat challenging. There is very little ground traces or pour on the bottom of the board, while the top is an almost uninterrupted ground plane. Because of the that, the short is most likely to be at the top of the board and may be under e.g. a capacitor. It may be just a tiny drop of solder in a wrong place.
For starters, I would recommend looking carefully at the top of the board around where the negative rail is located. They are not too many: pin 4 of the opamps. pin 12 of the comparator and C52 nearby, diodes D2 and D22 near the output connector, collectors of the PNP output transistors (those facing opamps), small ceramic caps and 10 kOhm biasing network resistors that are right in front of those transistors, and Q52, C53 and R60 near the power connectors.
If visual inspection doesn't reveal the short, the easiest way is to connect a current limited 5-15V power supply to the negative rail board, set the limit at say 1 amp, switch it on, and trace the voltage drop along the negative rail traces on the bottom with a DMM. Use the power connector's pin 4 (negative rail) as the reference point for the measurements. Increasing voltage drop will lead you to the short, and beyond it, the voltage drop will be constant. It is probably a good idea to pull the opamps and the comparator out of their sockets for this exercise.
For starters, I would recommend looking carefully at the top of the board around where the negative rail is located. They are not too many: pin 4 of the opamps. pin 12 of the comparator and C52 nearby, diodes D2 and D22 near the output connector, collectors of the PNP output transistors (those facing opamps), small ceramic caps and 10 kOhm biasing network resistors that are right in front of those transistors, and Q52, C53 and R60 near the power connectors.
If visual inspection doesn't reveal the short, the easiest way is to connect a current limited 5-15V power supply to the negative rail board, set the limit at say 1 amp, switch it on, and trace the voltage drop along the negative rail traces on the bottom with a DMM. Use the power connector's pin 4 (negative rail) as the reference point for the measurements. Increasing voltage drop will lead you to the short, and beyond it, the voltage drop will be constant. It is probably a good idea to pull the opamps and the comparator out of their sockets for this exercise.
Last edited:
There are 4 small holes (vias?) around C33. I got solder in one of these holes. Could that be causing the short?
Picture removed as requested
Those vias are the negative rail, too. Solder in the vias by itself should not be causing a short - unless it somehow connects to the ground pour around, or to the grounded lead of C33.
Thanks Alex. I checked the points you suggested and they all look good. I was able to remove the solder from the via but it had no continuity to C33. I will have to build myself a power limited supply and test as you suggested. Thanks again for your help.
Dan
Dan
The trace resistance is about 3.5 mOhm per cm of length (9 mOhm per inch), so you'd need at least a couple of hundreds of milliamps to flow in order to see millivolts of difference along the trace. Just use a suitable resistor in series with a suitable DC supply - note the resistor will need to dissipate quite some heat. You want the correct polarity and voltage not exceeding 18V to keep the circuit alive if/when the short is gone.
Another Omicron is born. I had made a bonehead mistake. The negative rail was shorting to ground because I did not use the recomended wire headers and wired the power input V+,g,g,V-,Pgood, rather than Pgood, V-,g,g,V+. I was trying to find the mistake with a current limited DC supply with no luck when the lightbulb moment happened. Checked the continuity at the power input on my spare unpopulated board and realized my mistake. Thanks again Alex for monitoring this thread and being so quick to respond to questions.
Dan
Dan
This little amp is something else. Listening last night it really satrted to open up after a couple of hours. Bass goes deeper than any of my other headamps. Very black background. Smooth yet detailed. I ended up listening much later than I had planned. Very musical. Definitely a keeper. Thanks for making this available Alex.
Dan
Dan
This sounds familiar - we have similar experiences ourselves and hear this sort of comment quite often. The music simply takes you away...
This looks really interesting! How does the sound compare with headphone amp that use a large number of NE5532 in parallel?
Paralleled 5532s don't come close.
Actually, with high-impedance headphones (which become rare as low-voltage portable electronics takes over) you may not even need to parallel; a single LM4562 will drive high-impedance cans, CMoy-style. But the sound is also CMoy-style - not bad but totally unremarkable.
Actually, with high-impedance headphones (which become rare as low-voltage portable electronics takes over) you may not even need to parallel; a single LM4562 will drive high-impedance cans, CMoy-style. But the sound is also CMoy-style - not bad but totally unremarkable.
Last edited:
Stanisław Lem said in His Master's Voice: "Information at second hand always gives an impression of tidiness, in contrast with the data at the scientist’s disposal, full of gaps and uncertainties."
Have you tried paralleling 5532s yourself? What are your measurements and/or listening impressions?
Have you tried paralleling 5532s yourself? What are your measurements and/or listening impressions?
Last edited:
You don't need to. The power rail voltage is limited by the opamp, and the output current is limited by that voltage and load impedance. One pair of BD139/140 is beefy enough - that is, the loadline is well withing its SOA for all reasonable headphone loads.
With +/-17V rails, Omicron drives its output to +/-15V, which gives peak output current of 300mA with a 50 ohm load, corresponding to 4.5W peak power - which should be enough. I believe @Will2226 is driving Hifimans with his Omicron may he can share his experience.
If, for some reason, you want more power, opamps with buufers won't provide it. You need to look into discrete designs, such the KSA-5clone with my mods, posted elswhere on this forum, or to chipamps like @EUVL suggested to you in the other thread.
With +/-17V rails, Omicron drives its output to +/-15V, which gives peak output current of 300mA with a 50 ohm load, corresponding to 4.5W peak power - which should be enough. I believe @Will2226 is driving Hifimans with his Omicron may he can share his experience.
If, for some reason, you want more power, opamps with buufers won't provide it. You need to look into discrete designs, such the KSA-5clone with my mods, posted elswhere on this forum, or to chipamps like @EUVL suggested to you in the other thread.
D
Deleted member 375592
Alex, have you tried to bias 5532 into class A by loading them with a current source of ~2mA (+ or -)? As much as I played with nested feedback amplifiers in SPICE, it made some improvements for low-impedance loads and allowed for lower Iq.
I've standardized on Omicron for the majority of headphone use cases that I'm expecting to encounter.
I also intend on building a version of the Aegis (hand wiring it for the standard 6L6 class tubes, plus 7591 and KT8c) for higher power situations and just because it's supposed to be another great sounding cans amp. That ain't cheap though - just the iron is around $1000.
Once I have that happening, I might try driving the Aegis output tubes with the Omicron (bypassing the 6SL7s) and see what transpires. It's way outside of the Omicron footprint and power requirements to use tubes, but hmm...cathode driven OTL? Could be fun!
I also intend on building a version of the Aegis (hand wiring it for the standard 6L6 class tubes, plus 7591 and KT8c) for higher power situations and just because it's supposed to be another great sounding cans amp. That ain't cheap though - just the iron is around $1000.
Once I have that happening, I might try driving the Aegis output tubes with the Omicron (bypassing the 6SL7s) and see what transpires. It's way outside of the Omicron footprint and power requirements to use tubes, but hmm...cathode driven OTL? Could be fun!
With suitably gargantuan power supply components, and suitably oversized heatsinks, you could easily DIY something that pumps 1 ampere (RMS) into 8 ohm headphones. A textbook "Blameless" solid state amplifier with +/- 36 volt supply rails, whose output stage uses two pairs of output transistors, biased in Class-A at 0.5 amperes DC per pair, would do the job. It'll drive 14 watts into 32 ohms single ended, or 28 watts into 32 ohms balanced. More than the Hifiman headphone amp w/DAC. Just as importantly: you'll experience the joy of DIY circuit design and the further joy of DIY PCB layout. If you also want vanishingly low distortion, to go along with ridiculous high power, add an opamp to the front and create a composite amplifier for error correction. As mentioned in the Buffer thread .
_
_
Attachments
Last edited:
Not sure I understand. The opamps in the Omicron do not drive low impedance loads. For this and some other reasons, I don't think loading opamps in the Omicron with resistors or current sources would bring any benefits. Also, note that many modern opamps with so-called Monticelli rail-to-rail output stage do not benefit from such biasing at all.
Measurements of assembled boards often diminish the joy of designing them.