I’m working with Jhofland to develop a new Class AB version of the Pocket Class A (PCA) headphone amp in an Altoids tin called the Pocket Diamond Buffer (PDB). It will use the Diamond buffer discussed in another thread combined with an OPA1642 for the voltage gain. The buffer will be inside the feedback loop of the opamp for a composite amp. This will be a state of the art headphone amp in a nostalgic DIY package. The power on/off will be controlled by the click on the volume pot (same as PCA) but a MOSFET had to be added to turn on and off the negative rail. Power will be two 9v (rechargeable or alkaline) batteries. Being Class AB, the power consumption should be way less than the PCA.
Here is first draft of layout. It’s very dense with 16 actives and an IC:
Here is top level schematic:
Individual buffers look like this (values TBD for operation at lower impedances):
For those unfamiliar with the PCA, here is what I am thinking it will resemble:
Stay tuned as this is still early stages. We need to make a prototype and test and verify still.
Update - PCB and component location stuffing guide.
Bottom:
Top:
* Important* please change C105/106 to 47pF NP0/C0G. 470pF causes oscillation at 6MHz.
Kyocera KGM15ACG2A470FT is a good replacement.
Here is first draft of layout. It’s very dense with 16 actives and an IC:
Here is top level schematic:
Individual buffers look like this (values TBD for operation at lower impedances):
For those unfamiliar with the PCA, here is what I am thinking it will resemble:
Stay tuned as this is still early stages. We need to make a prototype and test and verify still.
Update - PCB and component location stuffing guide.
Bottom:
Top:
* Important* please change C105/106 to 47pF NP0/C0G. 470pF causes oscillation at 6MHz.
Kyocera KGM15ACG2A470FT is a good replacement.
Last edited:
The assembled PCBs just arrived. They look great. I need to install the through hole parts and test it.
The Pocket Diamond Buffer headphone amp sings! Running about 80mA class A bias current and BjT temps around 70C. The sound is lovely on my OB-1 headphones. Using an iPhone dongle DAC at the moment.
Nice to see something work the first time. No smoke. 🙂
A big thanks to @jhofland for the circuit design!
I am very much enjoying how powerful and clean and natural sounding this amp is.
Nice to see something work the first time. No smoke. 🙂
A big thanks to @jhofland for the circuit design!
I am very much enjoying how powerful and clean and natural sounding this amp is.
Fitted into an Altoids tin finally.
With silicone pads conducting heat to the lid, the outputs are about 45C.
With silicone pads conducting heat to the lid, the outputs are about 45C.
Here is the schematic of the diamond buffer output stage that is then nested in the above main schematic. Output of the buffer has feedback to the input of the main voltage gain opamp (set for 12dB).
This is predicted 1Vpp into 32ohms (4mW) and gets 0.0007% THD:
This is predicted 1Vpp into 32ohms (4mW) and gets 0.0007% THD:
There is a GB for this board over in the PCA GB thread. $62ea fully assembled with all SMT parts. You need to provide and install the TH parts yourself.
https://www.diyaudio.com/community/...ass-a-headamp-gb.302859/page-112#post-7507213
https://www.diyaudio.com/community/...ass-a-headamp-gb.302859/page-112#post-7507213
How are you handling possible asymmetries in battery voltage? It looks like the ground is taken where the two batteries are connected in series, but that means the rail voltages may differ over time as the batteries drain.
It’s not specifically handled. You could add the DC offset protection SSR circuit I have for headphones. But then it won’t fit in an Altoids tin anymore. We are trying to keep this simple for a 9 active component per channel discrete output buffer with 2 active components for power switching and an opamp for voltage gain. As Jhofland called this circuit: “ten pounds of cr@p in a five pound bag”. 🙂
But in use so far, I have not noticed an offset problem in my headphones. I will measure it with a DMM as battery drains and see if its an issue.
But in use so far, I have not noticed an offset problem in my headphones. I will measure it with a DMM as battery drains and see if its an issue.
With the batteries in a partially depleted state, I measured 7.11v and -7.39v rails. The output DC offset was -8mV on one channel and -11mV on the other. These are rather low DC offset levels given that no active DC offset is being implemented.
It’s now installed in a chassis as a Desktop Diamond Buffer headphone amp. I am using a Recom +/-15v 340mA AC/DC converter with a CLC and CMC filter. Followed by a linear regulator to drop the voltage to +/-9v. The amp has a heatsink bonded to it using RTV.
Initially I tried to use a light pipe (PLA 3D printer filament) to bring the PDB green LED to the front panel but that didn’t work since the LEDs are side emitting models. I ended using the Molex 2 pin power indicator on the PSU board.
Initial listening tests are very very good. Absolutely silent - no hum or noise with music not playing. Very powerful and dynamic yet natural sounding. Super bass on hard to drive HE-400i 32ohm planar magnetic headphones.
Jhofland is designing a new Desktop Diamond Buffer headphone amp board using large TO-220 output devices with PCB mounted heatsinks.
Initially I tried to use a light pipe (PLA 3D printer filament) to bring the PDB green LED to the front panel but that didn’t work since the LEDs are side emitting models. I ended using the Molex 2 pin power indicator on the PSU board.
Initial listening tests are very very good. Absolutely silent - no hum or noise with music not playing. Very powerful and dynamic yet natural sounding. Super bass on hard to drive HE-400i 32ohm planar magnetic headphones.
Jhofland is designing a new Desktop Diamond Buffer headphone amp board using large TO-220 output devices with PCB mounted heatsinks.
I added the snubber to the Pocket DB amp. 0.1uF 100v X7R 1210 cap and 33ohm 1210 resistor. The revised board will use a through hole film cap instead.
Revised PCB with TH snubber:
Revised PCB with TH snubber:
I just checked the PDB for oscillation with an Oscope and the 33R and 0.1uF X7R cap was not sufficient. I replaced the 33R with 8.25R and the MLCC cap with a Wima 0.1uF 100v MKS. The oscillation is now gone.
Hi Folks,
You can buy the PDB in my shop now for $69. All SMT parts fully populated. You only need to install through hole parts. Fully tested and verified working by me - hence I had to install a 4 pin JST connector.
https://xrkaudio.etsy.com/listing/1606375374
You can buy the PDB in my shop now for $69. All SMT parts fully populated. You only need to install through hole parts. Fully tested and verified working by me - hence I had to install a 4 pin JST connector.
https://xrkaudio.etsy.com/listing/1606375374
I finally got a chance to put the little PDB in Altoids tin on the measurement rig. I’m using a Focusrite 4i4 3rd gen in ASIO mode at 48kHz. Victors 1kHz low THD oscillator, 33ohm metal thin film dummy load. My usual balanced input AC coupled voltage divider rig and 100kHz brick wall noise filter. The amp is powered by “9v” EBL rechargeable batteries probably circa 8v each.
Here’s the setup showing 3.0Vrms and O-scope showing it is clean with no clipping yet. Distortion is extremely low. We will step through various power settings next:
Here is closeup of O-scope display at 3.0Vrms:
That’s about 273mW into 33ohms with 0.0021% THD (clean), 3rd harmonic dominant at high power since in Class AB push pull:
Starting in pure Class A at 4mW, we have 0.0025% THD with dominant 2nd harmonic and a bit of third harmonic. This is a beautiful and perfect distortion profile. I usually don’t listen much louder than this as my headphones are 96dB/mW so 4mW is quite loud. But on 83dB planars this is 89dB - or same as dance floor loud:
Here is 10mW into 33ohms we have 0.0026% THD and still dominant 2nd harmonic:
Here is 25mW into 33ohms and still looking good at 0.0021% THD and still wonderful sounding harmonic profile:
At 50mW into 33ohms we still have 0.0018% THD and dominant 2nd harmonic and lower 3rd, indicating it is still in Class A:
At 125mW into 33ohms we are starting to cross onto Class AB with 0.0014% THD and harmonic profile is equal parts 2nd and 3rd, still considered very nice sounding and sweet but with some bite:
If we go all the way to max output at the onset of visible clipping we get 378mW and 0.014% THD (still very respectable THD figure but the harmonic profile is all order present at same level so it would sound congested. However, the volume is so high here, your eardrums would be blown out and not be able to tell:
For reference, the 60Hz noise peak and some small blips at 5kHz and 8kHz and 16kHz are all picked up in the dirty lab RF air as this is with the amp turned off so you can see the residual noise background:
Anyhow, I think the main takeaway is that this is a beautifully performing headphone amp and subjectively it sounds great and this is clearly backed up with measurements showing a wonderful 2nd harmonic dominant distortion profile with some 3rd and nothing else. The noise floor is absolute silence at about -120dB. At typical listening volumes, THD is around 0.002% and max power without distortion is about 350mW into 33ohm headphones. These are really remarkable figures for such a small and compact portable pocket amp, or even large desktop amps. The sound really is just natural, non fatiguing, yet resolving and carries hefty bass authority with the ability to drive so much power.
Here’s the setup showing 3.0Vrms and O-scope showing it is clean with no clipping yet. Distortion is extremely low. We will step through various power settings next:
Here is closeup of O-scope display at 3.0Vrms:
That’s about 273mW into 33ohms with 0.0021% THD (clean), 3rd harmonic dominant at high power since in Class AB push pull:
Starting in pure Class A at 4mW, we have 0.0025% THD with dominant 2nd harmonic and a bit of third harmonic. This is a beautiful and perfect distortion profile. I usually don’t listen much louder than this as my headphones are 96dB/mW so 4mW is quite loud. But on 83dB planars this is 89dB - or same as dance floor loud:
Here is 10mW into 33ohms we have 0.0026% THD and still dominant 2nd harmonic:
Here is 25mW into 33ohms and still looking good at 0.0021% THD and still wonderful sounding harmonic profile:
At 50mW into 33ohms we still have 0.0018% THD and dominant 2nd harmonic and lower 3rd, indicating it is still in Class A:
At 125mW into 33ohms we are starting to cross onto Class AB with 0.0014% THD and harmonic profile is equal parts 2nd and 3rd, still considered very nice sounding and sweet but with some bite:
If we go all the way to max output at the onset of visible clipping we get 378mW and 0.014% THD (still very respectable THD figure but the harmonic profile is all order present at same level so it would sound congested. However, the volume is so high here, your eardrums would be blown out and not be able to tell:
For reference, the 60Hz noise peak and some small blips at 5kHz and 8kHz and 16kHz are all picked up in the dirty lab RF air as this is with the amp turned off so you can see the residual noise background:
Anyhow, I think the main takeaway is that this is a beautifully performing headphone amp and subjectively it sounds great and this is clearly backed up with measurements showing a wonderful 2nd harmonic dominant distortion profile with some 3rd and nothing else. The noise floor is absolute silence at about -120dB. At typical listening volumes, THD is around 0.002% and max power without distortion is about 350mW into 33ohm headphones. These are really remarkable figures for such a small and compact portable pocket amp, or even large desktop amps. The sound really is just natural, non fatiguing, yet resolving and carries hefty bass authority with the ability to drive so much power.
Last edited:
Nice measurements.
You have no resistor from the output.
Have you tested with a cap load?
Say 100n and 220n capacitor across the load.
You have no resistor from the output.
Have you tested with a cap load?
Say 100n and 220n capacitor across the load.
No, I have not put a cap load - but the output does have 100nF in series with 8.25ohm to GND as snubber for preventing oscillation.
What do you mean by “no resistor from the input”?
What do you mean by “no resistor from the input”?
Usually we put a resistor from the output to the load to make capacitive stable.
Such capacitance can come from the cable etc.
Such capacitance can come from the cable etc.