It seems Mouser has sold out of pots 
I've revised the board (not published yet) to also use 15 mm shaft length pots if you cut off the tiny 0.8mm "nub" on the front (which is easy as it's a brittle casting). I avoided the 15mm pot initially because it does require cutting off the "nub" to work with the 2mm panel and push-on knobs. But now the PCB has both footprints (15mm and 20mm). So this allows use of:
Alps RK09712200MY (15mm long "D" shaft 3B taper)
Alps RK09712200MC (15mm long shaft 15A taper)
Bourns PTD902-2015F-A103 (15mm long "D" shaft)
Bourns PTD902-2015K-A103 (15mm long "star" shaft)
And the 20mm shaft pots:
Alps RK097122008T (20mm "D" shaft 3B taper)
Alps RK09712200HA (20mm "D" shaft 15A taper)
Bourns PTD902-2020F-A103 (20mm long "D" shaft)
Bourns PTD902-2020K-A103 (20mm long "star" shaft)
And, if all else fails, you can use a 20K version of the above but the noise performance will suffer a bit:
Alps RK09712200MT (20mm "D" shaft 15A taper 20K)
I've revised the board (not published yet) to also use 15 mm shaft length pots if you cut off the tiny 0.8mm "nub" on the front (which is easy as it's a brittle casting). I avoided the 15mm pot initially because it does require cutting off the "nub" to work with the 2mm panel and push-on knobs. But now the PCB has both footprints (15mm and 20mm). So this allows use of:
Alps RK09712200MY (15mm long "D" shaft 3B taper)
Alps RK09712200MC (15mm long shaft 15A taper)
Bourns PTD902-2015F-A103 (15mm long "D" shaft)
Bourns PTD902-2015K-A103 (15mm long "star" shaft)
And the 20mm shaft pots:
Alps RK097122008T (20mm "D" shaft 3B taper)
Alps RK09712200HA (20mm "D" shaft 15A taper)
Bourns PTD902-2020F-A103 (20mm long "D" shaft)
Bourns PTD902-2020K-A103 (20mm long "star" shaft)
And, if all else fails, you can use a 20K version of the above but the noise performance will suffer a bit:
Alps RK09712200MT (20mm "D" shaft 15A taper 20K)
do it while off the board, pretty sure i just killed some relays dremelling 5mm off a pot shaft while mounted, so the knob mounted flush. used a diamond cutoff wheel so took no time, but just the relays dont work properly, well one, they are wired in parallel so it stopped the other one working too. pita i even thought twice about it and did it anyway.
IMPORTANT! This has unexpectedly killed most of my day, but I have revised the PC board to accommodate a wider variety of volume controls and that necessitated other changes. All the design files are now dated August 30th and available in O2 Resources.
I would very much appreciate another set of eyes or two to look things over, especially the Gerbers, and see if everything looks OK?
Here are the changes:
- I've added a section on "pot tapers". The Alps volume controls are available with two different tapers and you might want to decide which you would prefer. See the Obtaining The Components section.
- I've added a section with all the volume control part numbers explained.
- I've revised the BOM to reflect some of the latest changes at Mouser, new pots, and some other changes suggested by others.
- The optional R26 & R27 output resistors have been eliminated. You can accomplish the same thing by increasing R10, R11, R15 and R18. I needed the board space for the pot change.
- The off board output jack header, P2, has been moved to the other side of the output jack. This was to make room for R12 which was displaced with the new volume control option.
- The channels have been swapped at the volume control and output jack. This enables much cleaner high-current wiring of the output jack and doesn't change the function of the amp (tip in is still tip out) but might confuse someone familiar with the old schematic as U3 and U4 now serve the opposite channels.
- Minor improvements were made to the ground routing and ground fills made possible by the other changes.
- The gain stage components were tightened up and the inputs to U3 and U4 re-routed to further reduce EMI susceptibility.
- I've added a tools section.
OLDER CHANGES: Just to put everything in one place, the previous major changes were:
- C16 and C21 reduced to 0.022 uF
- C22 (now C1) increased to 1.0 uF
- C1 (AC input filter) eliminated
- Footprint revised for Eagle battery clips
- Larger hole size on U5, U6, Q1 and Q2 to accommodate more parts
- Gain change to 2.5X and 6.5X
I would very much appreciate another set of eyes or two to look things over, especially the Gerbers, and see if everything looks OK?
Here are the changes:
- I've added a section on "pot tapers". The Alps volume controls are available with two different tapers and you might want to decide which you would prefer. See the Obtaining The Components section.
- I've added a section with all the volume control part numbers explained.
- I've revised the BOM to reflect some of the latest changes at Mouser, new pots, and some other changes suggested by others.
- The optional R26 & R27 output resistors have been eliminated. You can accomplish the same thing by increasing R10, R11, R15 and R18. I needed the board space for the pot change.
- The off board output jack header, P2, has been moved to the other side of the output jack. This was to make room for R12 which was displaced with the new volume control option.
- The channels have been swapped at the volume control and output jack. This enables much cleaner high-current wiring of the output jack and doesn't change the function of the amp (tip in is still tip out) but might confuse someone familiar with the old schematic as U3 and U4 now serve the opposite channels.
- Minor improvements were made to the ground routing and ground fills made possible by the other changes.
- The gain stage components were tightened up and the inputs to U3 and U4 re-routed to further reduce EMI susceptibility.
- I've added a tools section.
OLDER CHANGES: Just to put everything in one place, the previous major changes were:
- C16 and C21 reduced to 0.022 uF
- C22 (now C1) increased to 1.0 uF
- C1 (AC input filter) eliminated
- Footprint revised for Eagle battery clips
- Larger hole size on U5, U6, Q1 and Q2 to accommodate more parts
- Gain change to 2.5X and 6.5X
Google Docs works perfectly for me and I'm using a Mac.When I use the Google Docs viewer it takes a second or two after you zoom in to download the hi-res version. Perhaps Australia is connected to the USA via dialup?
Based on all that's changed (see post 263 above), the group buy being well past 500 boards, PayPal delays on Oliver's side, and nobody else volunteering to get a small run of boards made, I went ahead and ordered a 4th round of proto boards. I paid more for quick-turn boards from a US supplier so I can get them them next week and hopefully not delay things too much past the existing PayPal delay.
At least it should help me sleep better not having to worry if there's a significant problem with the PCB files when the stakes are relatively high. I've removed the old PCB files from Google Docs for now pending approval of the new prototypes.
As I've said many times, this hasn't played out as I was expecting. I never imagined the very first group PC board order would be for more than 50 boards let alone more than 500. I also didn't imagine Mouser would sell out of substantial inventories of key parts due to the high demand necessitating a board re-design. I just wanted to design an amp that raised the bar and demonstrate what's possible on a modest budget with modest parts.
At least it should help me sleep better not having to worry if there's a significant problem with the PCB files when the stakes are relatively high. I've removed the old PCB files from Google Docs for now pending approval of the new prototypes.
As I've said many times, this hasn't played out as I was expecting. I never imagined the very first group PC board order would be for more than 50 boards let alone more than 500. I also didn't imagine Mouser would sell out of substantial inventories of key parts due to the high demand necessitating a board re-design. I just wanted to design an amp that raised the bar and demonstrate what's possible on a modest budget with modest parts.
Hmm, doesn't really look like there are going to be pots available anyway once I get around to ordering my parts. I don't suppose the identical Alps with the switch can be used? Those are what I have on hand. That is why I always preferred wired pots and jacks. It simplifies the enclosure also. Drilling precise holes in a front panel is not one of my strong suits... My payment for the GB has been sent the old fashioned way. So I apologize in advance if I'm holding anything up.
My payment for the GB has been sent the old fashioned way. So I apologize in advance if I'm holding anything up.
What part numbers do you recommend for lemos for headphone amps? Do you use the Lemo for interconnects, and if you do, do you do two cables for stereo, or do you use just one for both channels? I've been thinking about getting some Lemos to try, I think they have the best design, and since we're building things ourselves, we might as well have the best connector.
I think that there are cable mount mini xlr's, I'll have to search around. Neutrik makes some, but I'm not certain they have male and female.
Ah, ok.
Yeah, the Redels are a nice connector, though the one drawback I see for a small portable amp is the 11/16" diameter taken up by the "nuts."
Personally I think a 4 pin mini-XLR would be a good choice.
se
FYI, it appears Mouser is also out of FK28X7R1H223K MLCC 0.22uF capacitors. The information I have says they expect them mid-November. I believe DigiKey part #445-5254-ND is the same cap, and they have 888 available. With 500 boards on order, the group buy may sell out DigiKey too!
[EDIT: Never mind. That's the substitute capacitor. The listed one is in stock.]
[EDIT: Never mind. That's the substitute capacitor. The listed one is in stock.]
Last edited:
O2 signal path sim using real NJM models!
Well it turns out that New Japan Radio actually does have Spice models for the O2 parts! Fairly good models too. Each model comes with a verification/comparison document of the model vs. the actual part measurements in a test jig for various parameters. The NJR Spice models work just fine in (freeware) LT spice when new symbols are created and linked to the NJM model subcircuits.
NJM4556A (macromodel in lower right corner, site registration is required)
NJM2068
The Spice simulation really doesn't add anything at this point to the O2 results since RocketScientist has made actual measurements with his equipment. But for any electrical engineering students out there who may be buying an O2 and wanting to try their hand at inserting the models in LT Spice, this is proof it can be done! Hint: the subcircuits are for dual chips, as shown in the schematic below, so you have to supply 4 inputs, 2 outputs, Vcc and Vee and the symbol then has to have 8 pins in the appropriate order to match up with the subcircuit arguments.
The legend on the plots are at the top of each and match up with the labels in the circuit. The top plot is the top channel and the bottom plot is the bottom channel. The input signal is 2Vrms for each, with the top running at 2kHz and the bottom at 4.2kHz just to be different. The pot is "set" at 80% rotation for each channel with the 2k/8k resistive dividers. The load is 44R resistive, Shure SRH840 or 940. Clicking on the arrows in the lower left of the plot with blow it up to full size.
Something fun to try: slap some capacitance (150pF, 250pF, 500pF) across the 44R load to simulate headphone cable capacitance and then check to see if the amp simulates as still stable with no ringing (it does).
Notice that right now the input signal and the output signals of the two stages are all in phase since the audio band frequencies are significantly less than the Fc of the input RF filter, the RC stage going from the pot into the output stage, and the op amp feedback loop compensation RC. Another fun thing to try is increase the input frequency closer to the Fc of each, say 200kHz, and then check phase and amplitude vs. audio band signal at 25kHz. Everything works the way it should. With proper part selection like this the audio band passes through unfazed but RF and gain at RF frequencies (which can produce oscillation) gets reduced toward unity.
Well it turns out that New Japan Radio actually does have Spice models for the O2 parts! Fairly good models too. Each model comes with a verification/comparison document of the model vs. the actual part measurements in a test jig for various parameters. The NJR Spice models work just fine in (freeware) LT spice when new symbols are created and linked to the NJM model subcircuits.
NJM4556A (macromodel in lower right corner, site registration is required)
NJM2068
The Spice simulation really doesn't add anything at this point to the O2 results since RocketScientist has made actual measurements with his equipment. But for any electrical engineering students out there who may be buying an O2 and wanting to try their hand at inserting the models in LT Spice, this is proof it can be done!
Hint: the subcircuits are for dual chips, as shown in the schematic below, so you have to supply 4 inputs, 2 outputs, Vcc and Vee and the symbol then has to have 8 pins in the appropriate order to match up with the subcircuit arguments.The legend on the plots are at the top of each and match up with the labels in the circuit. The top plot is the top channel and the bottom plot is the bottom channel. The input signal is 2Vrms for each, with the top running at 2kHz and the bottom at 4.2kHz just to be different. The pot is "set" at 80% rotation for each channel with the 2k/8k resistive dividers. The load is 44R resistive, Shure SRH840 or 940. Clicking on the arrows in the lower left of the plot with blow it up to full size.
Something fun to try: slap some capacitance (150pF, 250pF, 500pF) across the 44R load to simulate headphone cable capacitance and then check to see if the amp simulates as still stable with no ringing (it does).
Notice that right now the input signal and the output signals of the two stages are all in phase since the audio band frequencies are significantly less than the Fc of the input RF filter, the RC stage going from the pot into the output stage, and the op amp feedback loop compensation RC. Another fun thing to try is increase the input frequency closer to the Fc of each, say 200kHz, and then check phase and amplitude vs. audio band signal at 25kHz. Everything works the way it should.
With proper part selection like this the audio band passes through unfazed but RF and gain at RF frequencies (which can produce oscillation) gets reduced toward unity.Attachments
Last edited:
@agdr, thanks for the sim notes. I didn't know about the models for the NJM parts. That's a good find.
I'm curious what the sim does with a 32 ohm load (I used the Sennheiser CX300 headphones) and even 0.1 uF (100,000 pF!) in parallel? In real life the O2 still doesn't become unstable with 10 Khz square wave even at that high capacitance.
Also, did you use 1/2 ohm output resistors (rather than 1 ohm) on purpose to compensate for the 4556 "dual" model in some way?
I'm curious what the sim does with a 32 ohm load (I used the Sennheiser CX300 headphones) and even 0.1 uF (100,000 pF!) in parallel? In real life the O2 still doesn't become unstable with 10 Khz square wave even at that high capacitance.
Also, did you use 1/2 ohm output resistors (rather than 1 ohm) on purpose to compensate for the 4556 "dual" model in some way?
RocketScientist - those NJM4556 chips do have an amazing ability to handle load capacitance! Headphone cable capacitance and parasitic headphone capacitance shouldn't be a problem with the O2. I've run sims below with the 32R and 0.1uF on one channel, and again with 32R and 0.001uF (1000pF).
The output resistors actually are 1R. The labels for the 4556's (NJM4556- 1/2) are just sitting right on top of the "1R" labels. Thanks for the heads up on that, I've moved the resistance label above the resistors here.
The first plot is 32R with 0.1uF across it at 2kHz sine, and the next two are the same at 10kHz sine and 25kHz sine. No oscillation, just a small amount of waveform distortion on the 4556 output at the two higher frequencies. But good grief, that is something like 500 times the capacitance of a headphone cable! Pretty impressive. The next plot is 10kHz sine with 32R load and 0.001uF across it. Also just a reminder to anyone reading this - it is just a sim - the actual circuit measurements are the truth of course.
The last plot is 32R load with 0.001uF and with 10khz square waves. I'm clearly doing something wrong here to have the output just going between the positive rail half, and I'm not seeing the problem right at the moment. But the waveshape looks good given that high capacitance and high current load. No significant undershoot, overshoot or ringing.
The output resistors actually are 1R. The labels for the 4556's (NJM4556- 1/2) are just sitting right on top of the "1R" labels.
Thanks for the heads up on that, I've moved the resistance label above the resistors here.The first plot is 32R with 0.1uF across it at 2kHz sine, and the next two are the same at 10kHz sine and 25kHz sine. No oscillation, just a small amount of waveform distortion on the 4556 output at the two higher frequencies. But good grief, that is something like 500 times the capacitance of a headphone cable! Pretty impressive. The next plot is 10kHz sine with 32R load and 0.001uF across it. Also just a reminder to anyone reading this - it is just a sim - the actual circuit measurements are the truth of course.
The last plot is 32R load with 0.001uF and with 10khz square waves. I'm clearly doing something wrong here to have the output just going between the positive rail half, and I'm not seeing the problem right at the moment. But the waveshape looks good given that high capacitance and high current load. No significant undershoot, overshoot or ringing.
Attachments
-
O2 signal path with NJM chips - circuit2.png -
O2 signal path with NJM chips - 32R 0.1uF 2khz sine.png -
O2 signal path with NJM chips - 32R 0.1uF 10khz sine.png -
O2 signal path with NJM chips - 32R 0.1uF 25khz sine.png -
O2 signal path with NJM chips - 32R 0.001uF 10khz sine.png -
O2 signal path with NJM chips - 32R 0.001uF 10khz square.png
Last edited:
Another day, another review of the O2. TLDR: very low noise, on/off transients sound like "small click" on and "brief thump" off with IEMs, very neutral sound.
Lord Voldemort eh? That's some rap just from finding issues with the AMB amp.
Or have I missed something?Lord Voldemort eh? That's some rap just from finding issues with the AMB amp.
I was wondering about this too, but I think I get it now. In the Potter books, Lord Voldemort is also know as "He-who-shall-not-be-named". The review author uses it to allude to the fact that you aren't allowed to name banned members on head-fi.