This is my take on a small head amp with diamond buffer (BUF634T), below distortion into 50 ohm at 950mV
If you want to use an IC for an output buffer, the TPA6120A2 also has very impressive specs with 700mA max current, 1300V/usec slew rate, and similar 36V supply. It is used on some very high performing headphone amps from Topping as a nested composite amp. I use it in my HyperDrive-2 headphone amp. But I find the harmonic signature of Jhofland’s Diamond Buffer to be more pleasing with similar low distortion figures.
The final production data files for the Desktop Diamond Buffer headphone amp have just been finalized and they boards are starting to get made, finally. The board house had so many questions for me this time - back and forth 5 times at least.
But here is what the parts placement (SMT) is going to look like.
But here is what the parts placement (SMT) is going to look like.
Just got my board, but don't see anywhere to mount snubbers. Are they needed? If so, where do they mount?
I used 0.1uF 100v Wima film caps and a small SMT resistor (circa 5ohms to 10ohms) to make the snubber. I mounted it like this:
If you are ok with X7R 100nF caps as your snubber you can make it much more compact:
If you are ok with X7R 100nF caps as your snubber you can make it much more compact:
Not quite. Must be outside of feedback loop.Good point. Is presence of emitter resistors enough?
Here is the BOM. Bold items are the through hole parts that need to be purchased and installed by the user. Feel free to boost 47uF caps up to 100uF.
Note that you will need to add a snubber at the output TRS jack solder pins as shown above. A 4.7/6.8/8.2/10R are all values that should work. Plus a 100nF film cap. The little Wima 100v MKS ones are nice and small. A 100nF X7R ceramic cap can work too and is much more compact but some people don’t like ceramic caps in the signal path.
For C101/102 the input coupling caps, I used Elna Silmic 10uF 35v electrolytic caps. I faced the +ve end into the opamp - although I don’t think it matters. The DC levels are so small that the caps won’t have any issue.
You might have a hard time finding the Alps RK09 pot with on off switch. A good alternative although not Alps for sure (I have tested and works) is on Aliexpress:
https://a.aliexpress.com/_mPDoSFO
Note that you will need to add a snubber at the output TRS jack solder pins as shown above. A 4.7/6.8/8.2/10R are all values that should work. Plus a 100nF film cap. The little Wima 100v MKS ones are nice and small. A 100nF X7R ceramic cap can work too and is much more compact but some people don’t like ceramic caps in the signal path.
For C101/102 the input coupling caps, I used Elna Silmic 10uF 35v electrolytic caps. I faced the +ve end into the opamp - although I don’t think it matters. The DC levels are so small that the caps won’t have any issue.
You might have a hard time finding the Alps RK09 pot with on off switch. A good alternative although not Alps for sure (I have tested and works) is on Aliexpress:
https://a.aliexpress.com/_mPDoSFO
|
Last edited:
Note that the equivalent TRS jacks made by SwitchCraft (used on my PCA amp) also work and have same footprint. Although the Altoids tin cutout holes have changed somewhat with this amp so you need to redo the holes manually (the cutout template for PCA doesn’t work).
I ordered these a while back and they're the real deal.
https://www.lcsc.com/product-detail...iometers_ALPSALPINE-RK0971221Z05_C336003.html
https://www.lcsc.com/product-detail...iometers_ALPSALPINE-RK0971221Z05_C336003.html
I was thinking that these might be the perfect stick on heatsinks.
https://a.aliexpress.com/_mrbHm2k
https://a.aliexpress.com/_mrbHm2k
After trying a bunch of things to tame the instability in the Desktop Diamond Buffer without success, I looked the circuit on LTspice and did an .AC simulation to get insight on the gain and phase margin. I first had to find the latest NE5532 model for the TI variant. The simulation showed that the default 470pF feedback resistor is too large. It causes the gain to flatten out at 0dB while the phase margin falls precipitously past 3MHz. The simulation showed that changing that cap to 47pF produces a gain that varies monotonically rather than flattens out above 300kHz and the phase actually flattens vs falling off. Swapping out the 470pF NP0 to 47pF NP0 fixed the instability right away - as predicted by LTspice.
In the LTspice schematic, C3 is the cap that was swapped from 470pF to 47pF.
Here is the gain and phase plot (closed loop) - cursor is at 0dB and we see that phase is -70deg:
In the actual circuit the cap is labeled C105 and C106 - located on the end of row so easy to change out:
It sounds super quiet now - and the overall sound quality is really nice.
I think I am ready to release these Desktop Diamond Buffer headphone amp boards for sale now.
This also implies that a change to C105 and C106 should be made on the Pocket Diamond Buffer headphone amps to improve the sound quality and eliminate oscillation even some more:
Make sure you use NP0 or C0G high quality MLCC caps for this application.
In the LTspice schematic, C3 is the cap that was swapped from 470pF to 47pF.
Here is the gain and phase plot (closed loop) - cursor is at 0dB and we see that phase is -70deg:
In the actual circuit the cap is labeled C105 and C106 - located on the end of row so easy to change out:
It sounds super quiet now - and the overall sound quality is really nice.
I think I am ready to release these Desktop Diamond Buffer headphone amp boards for sale now.
This also implies that a change to C105 and C106 should be made on the Pocket Diamond Buffer headphone amps to improve the sound quality and eliminate oscillation even some more:
Make sure you use NP0 or C0G high quality MLCC caps for this application.
I made up another board with increased bias current of 78mA for lots of Class A depth of operation. Taller heatsinks. I am using Silmic 10uF input electrolytic caps bypassed with 0.47uF MPT film caps to keep it compact.
The 47pF cap modification was applied to C105 and C106. No oscillation and letting it break in overnight. Sounds great.
The 47pF cap modification was applied to C105 and C106. No oscillation and letting it break in overnight. Sounds great.
The higher bias current of 78mA gives a nice harmonic profile with only H2 and H3 present. Here is 1.0Vrms into 33ohms, with a +/-9.9v PSU (DC/DC followed by CLCRC filter:
Here is the profile for the same condition but with the other amp running at 18mA bias current, the overall THD has increased and there are a few more orders present H2, H3, H4, H5:
Here is the profile for the same condition but with the other amp running at 18mA bias current, the overall THD has increased and there are a few more orders present H2, H3, H4, H5:
Got my PDB amp done.
I replaced C105/106 with 47pF, installed the snubbers, and am using ..... HEY! Stop ogling the last pic! Focus!! .... the OPA 1642 opamp.
The tin is one I've had for years. I also cheated with the rail cap's by using Elna Silmic II 100uF/16V that I still had in my stash (unobtainium now).
I replaced C105/106 with 47pF, installed the snubbers, and am using ..... HEY! Stop ogling the last pic! Focus!! .... the OPA 1642 opamp.
The tin is one I've had for years. I also cheated with the rail cap's by using Elna Silmic II 100uF/16V that I still had in my stash (unobtainium now).
Sorry I'm so late getting back with impressions. I've been comparing both the Pocket Class A (PCA) to the Pocket Diamond Buffer (PDB).
My chain is: Laptop - Creative Labs Super XFi dongle (as DAC) - amp - Fiio FH7 IEM's.
The PDB is definitely a step up from the PCA. The bass is deeper, more space between instruments, better transient attacks, better soundstage. If I had to choose between the two, I would pick the PDB. Fortunately, I don't have to choose.
The differences, though, are not night and day. The PCA is a great amp. It just comes up a tad short when compared to the PDB. It is maybe more like a nice tube amp and the PDB is more solid state. If you're in the mood for smooth, the PCA is your ticket; like a more dynamic sound, then the PDB is it.
When the PCA first came out, I built three of them. Sadly, they were all sold as I didn't really have a decent pair of IEM's to use with them. I just like building stuff. The one I have now is going nowhere; neither is the PDB. It's great to have both.
As an aside, we're having a meet up with several Head-Fi members on February 17th. I'm going to take both amps to get others' opinions, so stay tuned.
My chain is: Laptop - Creative Labs Super XFi dongle (as DAC) - amp - Fiio FH7 IEM's.
The PDB is definitely a step up from the PCA. The bass is deeper, more space between instruments, better transient attacks, better soundstage. If I had to choose between the two, I would pick the PDB. Fortunately, I don't have to choose.
The differences, though, are not night and day. The PCA is a great amp. It just comes up a tad short when compared to the PDB. It is maybe more like a nice tube amp and the PDB is more solid state. If you're in the mood for smooth, the PCA is your ticket; like a more dynamic sound, then the PDB is it.
When the PCA first came out, I built three of them. Sadly, they were all sold as I didn't really have a decent pair of IEM's to use with them. I just like building stuff. The one I have now is going nowhere; neither is the PDB. It's great to have both.
As an aside, we're having a meet up with several Head-Fi members on February 17th. I'm going to take both amps to get others' opinions, so stay tuned.