I missed this question earlier, sorry. Sofaspud was considering the low power version that uses low power op amps. That version isn't very happy with anything below about 50 ohms although it's still acceptable at 32 ohms. Sofaspud, I believe, discussed adding resistance to raise the impedance the low power op amps see to reduce the distortion. But it sounds like he's favoring the regular "high power" version now.
HEADS UP! I was hoping I was just being paranoid, but my caution not to buy parts yet has already proven to be good advice.
Someone asked the question elsewhere if they could get rid of the power control circuit and just switch the AC power. In exploring that, I ended up doing more power on/off testing beyond what I had done before based on the simulations that agdr had run that didn't match my earlier results.
I uncovered a specific low battery condition where the existing design can put out a fairly ugly transient. It probably wouldn't hurt most headphones but I don't want to take any chances.
THE BEST FIX: The right way to fix the problem is I'm simply adding a new capacitor value to the BOM and changing one other value. C16 and C21 drop from 0.22 uf to 0.022 uF (which is the new line item). And C22 increases from 0.47 uF to 1.0 uF. I'll be updating the docs shortly. The net change in cost is +$0.12.
THE WORKAROUND: If some of you have already bought parts, don't panic. If you just leave out C16 and C21 completely, and keep C22 at 0.47 uF, it solves the low battery problem. The downside is the turn on "click" is significantly worse but not dangerously so (it's about 0.8 V RMS for half of a 400 hz cycle). You can also add the correct cap values later if you want next time you order parts for another project. Or use "junk box" parts as they're not critical capacitors. Any value from 0.01 uF to 0.027 uF would be OK at 50 volts or higher.
Sorry about this, but I'm glad I caught it now before hopefully most of you have ordered parts. I've been working on the O2 in my spare time and it's required some late hours, etc. Don't be too shocked if something else crops up. There's little point in ordering parts until the first batch of boards are ordered at (I think?) the end of the month. The boards will have the longer lead time. So I encourage everyone to please be patient in case other changes need to be made.
Someone asked the question elsewhere if they could get rid of the power control circuit and just switch the AC power. In exploring that, I ended up doing more power on/off testing beyond what I had done before based on the simulations that agdr had run that didn't match my earlier results.
I uncovered a specific low battery condition where the existing design can put out a fairly ugly transient. It probably wouldn't hurt most headphones but I don't want to take any chances.
THE BEST FIX: The right way to fix the problem is I'm simply adding a new capacitor value to the BOM and changing one other value. C16 and C21 drop from 0.22 uf to 0.022 uF (which is the new line item). And C22 increases from 0.47 uF to 1.0 uF. I'll be updating the docs shortly. The net change in cost is +$0.12.
THE WORKAROUND: If some of you have already bought parts, don't panic. If you just leave out C16 and C21 completely, and keep C22 at 0.47 uF, it solves the low battery problem. The downside is the turn on "click" is significantly worse but not dangerously so (it's about 0.8 V RMS for half of a 400 hz cycle). You can also add the correct cap values later if you want next time you order parts for another project. Or use "junk box" parts as they're not critical capacitors. Any value from 0.01 uF to 0.027 uF would be OK at 50 volts or higher.
Sorry about this, but I'm glad I caught it now before hopefully most of you have ordered parts. I've been working on the O2 in my spare time and it's required some late hours, etc. Don't be too shocked if something else crops up. There's little point in ordering parts until the first batch of boards are ordered at (I think?) the end of the month. The boards will have the longer lead time. So I encourage everyone to please be patient in case other changes need to be made.
I'm guilty. It was I. And yes I'm going for the high power version. I'm not as decided about springing for the TLE2062s, but my current plan is to investigate just what kind of portable ps performance I can get with my headphones in my normal use. Then proceed from there. A little 5 or 12 volt DC-AC inverter for charging might be a handy accessory.
Playing around importing the BOM into Mouser, I see that 512-KA7912ATU_Q is now discontinued. Although a 512-KA7912ATU is the direct replacement, I'm guessing a cheaper 863-MC7912ACTG would be suitable (as you used the +ve version).
Mouser USA has 500+ of the Fairchild regulator in stock but they've gone from 0.19 to 0.67. The $0.19 part was not RoHS compliant so that's why they were cheap and are discontinued. I was just trying save everyone $0.30 
The best part is the $0.47 On Semi MC7912ACTG. The On Semi regulators have a lower noise spec and the "A" versions have tighter tolerances.
I'll revise the BOM when I'm fairly sure I don't have any other changes. As I said, please don't order parts yet. There's little reason when the boards are still many weeks from being delivered.
The best part is the $0.47 On Semi MC7912ACTG. The On Semi regulators have a lower noise spec and the "A" versions have tighter tolerances.
I'll revise the BOM when I'm fairly sure I don't have any other changes. As I said, please don't order parts yet. There's little reason when the boards are still many weeks from being delivered.
Mouser USA has 500+ of the Fairchild regulator in stock but they've gone from 0.19 to 0.67. The $0.19 part was not RoHS compliant so that's why they were cheap and are discontinued. I was just trying save everyone $0.30
The best part is the $0.47 On Semi MC7912ACTG. The On Semi regulators have a lower noise spec and the "A" versions have tighter tolerances.
I'll revise the BOM when I'm fairly sure I don't have any other changes. As I said, please don't order parts yet. There's little reason when the boards are still many weeks from being delivered.
I figured that was the case
I'm waiting until the boards are ordered before I order any parts. I was just playing to see what the approx cost of parts was in GBP. They come in about £19 for the basic parts, so would need to get 3 sets (or add other stuff) to get it over the £50 threshold for free shipping.
I have a dilemma. Given the large number of boards already on order, and the fact nobody has really "beta tested" the O2 yet, I've been doing more testing. I've uncovered an issue that's unlikely to be a problem for the vast majority of users but it's still bothersome.
The half wave supply in the O2 generates some fairly significant peak charging currents in the primary diodes, filter caps, PCB traces, etc. That's because the caps "droop" a lot further between charging pulses than in a full wave supply. While the regulators do a great job of removing the resulting ripple from the rails they do nothing for the electromagnetic and E-fields (EMI) the charging pulses generate.
The wall transformer already has a 5 - 10 ohm effective impedance limiting the peak current of the spikes. Trying to limit it further isn't practical but I'm open to suggestions if I'm missing something.
In the tight confines of the O2's packaging, if the input circuit is connected to a relatively high impedance source, it picks up some of the EMI produced by the high charging currents. The result can be some faint hum in the headphones when nothing is playing. It's typically around -90 dB referenced to 1 volt worst case at full volume and 2.5X gain. That's still better some headphone amps manage but nothing like the -106 dB the O2 is capable of. With sensitive headphones, and nothing playing, it's audible (referenced to 400 mV it's -82 dB).
To put this in perspective, the LOD and headphone outputs on my portable players are fine. As are the line outs on the Benchmark DAC1. And the O2 is dead quiet as seen in my measurements with the dScope as the source. And it's always dead quiet running from battery with any source.
I've reviewed the PCB layout and, if I had more room, I could certainly make some improvements, but given the constraints of 80mm x 100mm I'm not sure I can do more than I've already done.
All that said, I have found an interesting solution: If the case is grounded to the input ground the problem goes away. The big question is:
How do I reliably connect the input ground to the anodized (insulated) case in a DIY friendly way? One possibility is to change the board layout to accept an input jack with a metal barrel and a screw on nut. But those look seriously cheesy to me. It's a shame the ALPS pot doesn't have a ground pin for its metal shank but it doesn't.
Bare aluminum is famous for quickly developing an aluminum oxide layer that serves as an insulator. So you can't just have something in contact with the case, like a trace on the edge of the PCB as it won't make reliable contact over time. It has to be some sort of screw, nut, etc.
The only method I can think of so far is to trap a thin (like 30 gauge wire wrap wire) wire under one corner of the front or back panel where the screw would hold it in tight contact. Or a small thin "eye" terminal. But that's not exactly elegant.
I realize without an amp sitting in front of you it's not an easy question to answer, but I figured it doesn't hurt to ask?
As a side note, if anyone knows of any high impedance sources (beyond tube outputs) that might be used with the O2, it would be good to know what they are (and ideally their output impedance). Thanks in advance for any help on this.
The half wave supply in the O2 generates some fairly significant peak charging currents in the primary diodes, filter caps, PCB traces, etc. That's because the caps "droop" a lot further between charging pulses than in a full wave supply. While the regulators do a great job of removing the resulting ripple from the rails they do nothing for the electromagnetic and E-fields (EMI) the charging pulses generate.
The wall transformer already has a 5 - 10 ohm effective impedance limiting the peak current of the spikes. Trying to limit it further isn't practical but I'm open to suggestions if I'm missing something.
In the tight confines of the O2's packaging, if the input circuit is connected to a relatively high impedance source, it picks up some of the EMI produced by the high charging currents. The result can be some faint hum in the headphones when nothing is playing. It's typically around -90 dB referenced to 1 volt worst case at full volume and 2.5X gain. That's still better some headphone amps manage but nothing like the -106 dB the O2 is capable of. With sensitive headphones, and nothing playing, it's audible (referenced to 400 mV it's -82 dB).
To put this in perspective, the LOD and headphone outputs on my portable players are fine. As are the line outs on the Benchmark DAC1. And the O2 is dead quiet as seen in my measurements with the dScope as the source. And it's always dead quiet running from battery with any source.
I've reviewed the PCB layout and, if I had more room, I could certainly make some improvements, but given the constraints of 80mm x 100mm I'm not sure I can do more than I've already done.
All that said, I have found an interesting solution: If the case is grounded to the input ground the problem goes away. The big question is:
How do I reliably connect the input ground to the anodized (insulated) case in a DIY friendly way? One possibility is to change the board layout to accept an input jack with a metal barrel and a screw on nut. But those look seriously cheesy to me. It's a shame the ALPS pot doesn't have a ground pin for its metal shank but it doesn't.
Bare aluminum is famous for quickly developing an aluminum oxide layer that serves as an insulator. So you can't just have something in contact with the case, like a trace on the edge of the PCB as it won't make reliable contact over time. It has to be some sort of screw, nut, etc.
The only method I can think of so far is to trap a thin (like 30 gauge wire wrap wire) wire under one corner of the front or back panel where the screw would hold it in tight contact. Or a small thin "eye" terminal. But that's not exactly elegant.
I realize without an amp sitting in front of you it's not an easy question to answer, but I figured it doesn't hurt to ask?
As a side note, if anyone knows of any high impedance sources (beyond tube outputs) that might be used with the O2, it would be good to know what they are (and ideally their output impedance). Thanks in advance for any help on this.
A wire to a front plate screw is fine. It can easily be hidden. Gold plated loop connector and everything is groovy. Maybe a little sanding needed to ensure good contact but certainly not a huge issue for me at least.
Oh yea.. high impedance source. How about a guitar pickup or a microphone? Should be like worst case scenario there.
You could also run a 100k pot in front.. More hair brained ideas brought to you by me. A guy who spelled his handle wrong and sometimes resorts to fixing things with hammers and tape and also glue.
Oh yea.. high impedance source. How about a guitar pickup or a microphone? Should be like worst case scenario there.
You could also run a 100k pot in front.. More hair brained ideas brought to you by me. A guy who spelled his handle wrong and sometimes resorts to fixing things with hammers and tape and also glue.
Last edited:
Good advice for ordering parts. You could also pick up "generic" devices that are commonly used: 1k & 10k resistors, 5532 op amps, 1N400x diodes, 2N3904 & 2N3906 transistors, TRS and RCA jacks, etc. You could also buy parts to make your own connecting cables. I think it's easier to diy my own source-to-O2 cable than finding a good, short, factory-made job.
Alexium, the amp runs from +/- 9 V on battery and +/- 12 V on AC power. That's how it gets 7+ V RMS of output (20 volts peak-to-peak) to drive difficult headphones like the 600 ohm Beyerdynamics, etc. A +/- 5V amp is limited to about 2.3 - 3 V RMS output.
@Rembrant, thanks. Glad at least one other person finds that acceptable.
Also, for anyone making a desktop amp with RCA jacks, it's easy to just use one (or both) of the RCA jacks as the ground point as long as that's the only chassis ground.
@Rembrant, thanks. Glad at least one other person finds that acceptable.
Also, for anyone making a desktop amp with RCA jacks, it's easy to just use one (or both) of the RCA jacks as the ground point as long as that's the only chassis ground.
Last edited:
How high of an impedance for the source are we talking about? 50 ohm? 250 ohm? 500 ohm? I'm guessing that the interference ramps up as the source impedance goes up too.
Trapping a thin wire between the back plate and the case is also acceptable to me. otherwise you get into the problem of soldering or welding two different types of metal together.though i wonder if covering up the area where your ground wire touches the case with glue would keep it from oxidizing as fast.
Suggestion: I typically grind off a spot of the anodization with a dremel tool - or just de-ox the surface with it if not anodized - then use NoAlOx under the screw head or standoff:
Noalox Anti-Oxidant Compound
Noalox 0.5 oz. Anti-Oxidant Joint Compound-30-024 at The Home Depot
The stuff if great. It is made for use with aluminum electrical power wiring, 12 gauge on down. It has suspended metal particles to cut through oxide along with a carrier that keeps future oxidation at bay. I've seen ends of 2 -6 gauge wire used on electrical power panels (aluminum, copper, and silvered copper wire, I've seen it used on all) that were dipped in the stuff 10+ years ago and still looked like new when removed.
I haven't seen any data on the stuff vs. frequency though.
Last edited:
hmm, A wire to a front plate screw seems logic. but then i cannot make the front panel with laser etched acrylic which pretty much i was intended to make once my 02 finished
No worries, you can use acrylic. The wire can just as easily go to the back panel and it still seems to work just fine. For the front panel you can just just use a lead clipping from one of the 1/8 watt resistors as the input jack ground pin is very close to the lower left corner of the front panel. For the rear panel you need to run an actual wire.
And it doesn't need the shielding of the front panel. I'm doing all my testing with no panel at all (just the screw) and the noise reduction on the dScope is dramatic and sufficient just grounding the "shell".