BF862 based SE Class A Headamp without the HEAT

Founder of XSA-Labs
Joined 2012
Paid Member
I was looking for a small portable amp for my DT880-250 headphones. I wanted class A and I wanted longer battery life and low rejected heat. These sound like requirements that are at odds with each other right?

This project started in another thread and is based on the BF862 JFET single ended White follower design suggested by EUVL here. The design has a fairly low bias current of about 50mA, but because it used qnty 8 x BF862 JFETs on the output stage in a singel-ended arrangement with about 12mA in each device, the overall heat dissipation is very low but it is guaranteed to work in pure class A (it has to because of the topology). This amp is very simple to make following this schematic:

581624d1479910835-tin-fets-headphone-amplifier-many_many_jfets.png


I did my layout using a drawing program (MS PowerPoint in this case) and came up with this:

586034d1481838813-bf862-preamp-bf862-ha-layout.png


I then made iron on transfers using laser printer and got this on the copper as the etch mask:

586104d1481882105-bf862-preamp-bf862-ha-build-03.png


Etched in HCl and peroxide (hardware store muriatic acid concrete sidewalk cleaner and hydrogen peroxide (3%) for 20 minutes:

586105d1481882105-bf862-preamp-bf862-ha-build-04.png


Etched board looks like this:

586106d1481882105-bf862-preamp-bf862-ha-build-06.png


Soldered it up (I had to switch 1R source resistors for 33R to keep my bias current low, it was pegged at 23mA ea wit 1R):

586284d1481965252-mosfet-follower-headphone-amplifier-bf862-ha-build-stereo-08.jpg


I had some difficulty getting it to make sound at first and found out that the problem was a bad 6in length of wire from the output terminal to the headphone jack. Took hours to figure that one out after debugging the board components.

Initial tests were done at 12v supply and I boosted that up to 17v to get a little more headroom and keep heat down still. I am using a small DC to DC boost converter (400kHz switcher) and added 2200uF cap to filter it some more. Ripple is about 9mV at 17v but luckily the amp has some decent PSRR and output measures 0.3mV with inputs shorted. It is silent to my ears anyhow. Absolutely no hum or hiss. Cannot tell it is on.

I spend half a day listening and decided it was worthy to be boxed. So using this niftly enclosure. After drilling some holes and adding a 50k stereo pot, and power jack, switch, LED indicator, 3.5mm jacks IN and OUT, the amp is completed:

586398d1482011463-bf862-based-se-class-headamp-without-heat-bf862-ha-build-boxed-10g.jpg


Here is the (messy) inside. I am not one to have neat wiring harnesses as my time is very limited to work on it. The wires work fine and do not seem to pickup noises. All wires are at least twisted:

586399d1482011463-bf862-based-se-class-headamp-without-heat-bf862-ha-build-boxed-internal-11.jpg


Edit: I added a cap multiplier using IRF540 MOSFET with 10mF output ca[ and circuit from Juma's Easy Peasy Cap Mx thread:

586803d1482215824-bf862-based-se-class-headamp-without-heat-bf862-ha-build-boxed-internal-cap-mx-12.jpg


Here is the amp next to DT880's that I will be using them with. I am using a 2200uF 16v electrolytic cap for C2 bypassed with 1uF 230v MKT. I initially used 330uF OSCON but found the bass lacking. With the 2200uF I can hear deep deep bass now. There is a 270R 2W load resistor, so when I plug in the 250R DT880's, overall impedance is about 125R - which I think may help to draw more current out of the amp so it gets more impact with the bass (seems so subjectively).

586400d1482011463-bf862-based-se-class-headamp-without-heat-bf862-ha-build-final-dt880-12.jpg


So in addition to sounding extremely detailed, neutral, and having excellent bass authority, they are absolutely cool. The case is ambient temperature after hours of playing. Also, since I am using a XL6009 switch mode step up regulator, the power supply input is very flexible. For desktop use I have a 12v 3A SMPS wall wart from my external disk drive. For portable use, I will use a 5v USB phone charger battery pack and let the step up take care of getting the 17v rail. I estimate a 10,000mA Li-ion battery pack will last more than 12 hours. That's a lot of pure class A play time for a small battery.

This amp sounds wonderful and I thank EUVL for a great design. It has indeed been tested and works very well. Give it a shot.

Edit: Dec. 22, 2016 - an attempt at a smaller 37mm x 26mm layout to fit in an Altoids tin with two 9v batteries.

587101d1482351920-bf862-based-se-class-headamp-without-heat-bf862-ha-small-v2.png


Edit: Jan 6, 2017 - portable battery operation using 5v USB charging battery pack. 10,000mAhr cell should last about 15 hrs.

590340d1483735403-bf862-based-se-class-headamp-without-heat-bf862-se-ha-portable-demo-v2.jpg


Sound is still great - using XL6009 DC to DC step up internally for 18v supply to amp (after cap Mx).

Edit: Jan 7, 2017 - Here is a new portable 2 FET design that sounds great - very easy to make. We will have a GB soon so stay tuned!

Schematic and THD prediction:
591187d1483977924-bf862-based-se-class-headamp-without-heat-pocket-class-bf862-zvn5306-ha-v1.png


FFT HD profile (sweet H2 and not much else!):
591189d1483977924-bf862-based-se-class-headamp-without-heat-pocket-class-bf862-zvn5306-ha-v1-fft-1khz.png


Implementation as portable mint tin pocket Class A amp a la CMOY:
591117d1483961316-mosfet-source-follower-headamp-pocket-bf862-zvn4306-ha-smt-tin-case-dtr880.jpg


Internals of above amp:
591116d1483961316-mosfet-source-follower-headamp-pocket-bf862-zvn4306-ha-smt-tin-case.jpg


590981d1483910917-bf862-based-se-class-headamp-without-heat-pocket-bf862-zvn4306-ha-top-side.-caps-mods.jpg


590982d1483910917-bf862-based-se-class-headamp-without-heat-pocket-bf862-zvn4306-ha-smt-side-mods.jpg
 

Attachments

  • BF862-HA-build-boxed-10g.jpg
    BF862-HA-build-boxed-10g.jpg
    67.9 KB · Views: 7,827
  • BF862-HA-build-boxed-internal-11.jpg
    BF862-HA-build-boxed-internal-11.jpg
    140 KB · Views: 7,796
  • BF862-HA-build-final-with-dt880-12.jpg
    BF862-HA-build-final-with-dt880-12.jpg
    109.6 KB · Views: 13,308
Last edited:
Here is the (messy) inside. I am not one to have neat wiring harnesses as my time is very limited to work on it. The wires work fine and do not seem to pickup noises. All wires are at least twisted:
.

With audio its better not to harness the wires as close wires can interfere with each other.
Its worst with valve amps where there are high VAC and heater AC mixed in with high impedance audio signals.

I just got caught out recently with a USB scope circuit.
I ran a RAM address line close to a 1meg input resistor and every time the address line changed I got a glitch on the audio signal.
 
I want to put the record straight.

I posted this schematics as a Spice file for the gentleman easphyx who started the thread " Tin of FETs headphone amplifier".
http://www.diyaudio.com/forums/solid-state/298026-tin-fets-headphone-amplifier.html
It is largely his design so all credit goes to him.
I merely suggested a variant using alternative devices.
I have not built this and have no intention to, also in the future.

For this to work as in the simulation requires matching of both Idss and Yfs.
Gettinng sound out and having the circuit functioning properly as intended are two different things.
If any of you wish to find out please google SRPP and White Follower.
Or ask the originl designer at his thread, as a respect to him.

And the circuit is capable of fitting into the original footprint by easphyx, as he intended.


Patrick
 
Founder of XSA-Labs
Joined 2012
Paid Member
Last edited:
Founder of XSA-Labs
Joined 2012
Paid Member
I have let the amp burn in overnight. Still running and sounding great. Case temp is barely warm to the touch after being on for hours. I am waiting for the 5v USB charger battery to arrive. Looking forward to seeing if it can indeed be portable battery powered. I am debating whether or not to go in and tweak DC voltage up to get more head room. I think at max volume I detect clipping with my headphones. So currently at 17v now. Maybe take it to 20v.
 
Last edited:
Founder of XSA-Labs
Joined 2012
Paid Member
I just noticed that MMBFJ310 (SMT version of J310) has 60mA current capability. Could a similar amp with just two pairs be biased at 100mA then? It would be more compact as fewer parts. Is performance of J310 not as good as BF362 musically?

http://www.arrad38.fr/datasheet/J309_J310.pdf

Also, just using a single pair through hole J310 could give similar current as four pairs of BF862. Although heat dissipation could be an issue on single TO92 at 50mA.
 
Last edited:
I just noticed that MMBFJ310 (SMT version of J310) has 60mA current capability. Could a similar amp with just two pairs be biased at 100mA then? It would be more compact as fewer parts. Is performance of J310 not as good as BF362 musically?

MMBFJ310 has 24-60mA IDSS, which is the "idle current" when Vgs=0, so I wouldn't exactly call that "current capability" for the typical MMBFJ310, which will have less than 60mA IDSS. You may have trouble finding matched pairs if IDSS does actually vary that much.

Note that forward transconductance of MMBFJ310 is only 18 maximum compared to 45 typical of BF862. I would expect you to need almost three times as many of the MMBFJ310 to maintain a similar output impedance as the BF862's, despite their higher IDSS.

Also, how do you plan to cool a tiny SMT MMBFJ310 passing 60mA at ~10V?
 
Last edited:
Good job, guy.

May you say me how did you do the transfer from the paper to the copper side of the PCB, of the toner ink? I know it can be used an iron, but I doubt about time, and quality of the transfer, and which process did you do to ensure that the transferred toner doesn't peel off.

Thanks in advance.
 
Founder of XSA-Labs
Joined 2012
Paid Member
Good job, guy.

May you say me how did you do the transfer from the paper to the copper side of the PCB, of the toner ink? I know it can be used an iron, but I doubt about time, and quality of the transfer, and which process did you do to ensure that the transferred toner doesn't peel off.

Thanks in advance.

I use a laser printer to generate the black images onto adhesive backed vinyl film (applied to paper) used to cover shelves and other surfaces. Then preheat copper with iron for 2 minutes, then lay image face down on copper and cover with another piece of paper and apply iron for 2 minutes. Remove and run under cold water, then peel vinyl off at sharp (near 180 deg angle) to leave transfer behind. Touch up holes with black Sharpie marker.

This is what I use.

https://www.dollartree.com/Con-Tact...dhesive-Shelf-Liners-18x54-/p306003/index.pro

Very crisp features can be obtained if done right. I got it after 2nd try. Start with small projects like a preamp. Take for example my BF862 preamp by Juma. Lettering and traces are very clear. This one was better than my head amp.

585916d1481784715-transistor-preamp-juma-pre-47r-mod-06.jpg
 
Founder of XSA-Labs
Joined 2012
Paid Member
Added Juma Easy Peasy Cap Mx

I added Juma's Easy Peasy Cap Multiplier (using IRF540 MOSFET and 10mF 25v output cap). I also adjusted the DC step up to produce 20.0v output after the cap Mx. The dynamic headroom really went up another notch and cannot detect clipping at highest volume setting now. Also, bass authority seems to have improved dramatically with the cap Mx. Running 20v across about 50mA is about 1w dissipation in the JFETs per channel. Cap Mx is dissipating about 200mW for 4v drop and 50mA current, per channel. So 2.4w total heat generation so should be manageable.

586803d1482215824-bf862-based-se-class-headamp-without-heat-bf862-ha-build-boxed-internal-cap-mx-12.jpg
 

Attachments

  • BF862-HA-build-boxed-internal-cap-mx-12.jpg
    BF862-HA-build-boxed-internal-cap-mx-12.jpg
    136.9 KB · Views: 10,075
Last edited: