Borbely headphone amp

[Jupiter]

yhmzzz,

Do you know if anyone have compared the Borbely with other well-known designs?

 

[yhmzzz]

Hmm..

Many Korean DIYers compared them.
A few said the Borbely amp was good but others said it was not good.
I've experienced many DIY amps but it's too hard to describe the
details of comparisons because of my poor English..

Just enjoy your amp!! I'd liked the Borbely amp. But now I'm tube-lover.

 

[was ist los?]

So Jupiter, what do you think?

 

[Jupiter]

I haven't listened to it yet. I need a few more parts before I can put it in an enclosure.

 

[yhmzzz]

I'm expecting your amp..

 

[caffeineconsumer]

Hi there,

I'm a bit new to this forum, 1st post actually. But i've done a few diy audio thingys before - lm3886 stereo amp, and an all jfet onboard preamp for my bass guitar (designed by myself, still a work in progress).

Well i was just wondering if any of you could give me a parts list for the borbely all fet headphone amp. I'm real keen to build it but not cashed up enough to fork out 150 euros for the kit (being in australia i have to source the kit from germany).

I've sourced all the hard to get components for a reasonable price, and will embark on a pcb making project as Borbely no longer sells just pcbs.

I'm understanding this thread is a bit old now, but hopefully someone will come through with the goods.

Thanks, Luke.

ps. You might like to know that i've used my LM3886 chip amp with headphones through a resistive divider network, both to get the amp working in its sweet spot and also as protection for the headphones. I've used it with my brothers Allessandro Music Pros and it does pretty good, almost as good a job as his Talisman T-3H amp, they both sound incredibly similar.

 

[evo_lution]

Quote:

Originally Posted by caffeineconsumer You might like to know that i've used my LM3886 chip amp with headphones through a resistive divider network, both to get the amp working in its sweet spot and also as protection for the headphones. I've used it with my brothers Allessandro Music Pros and it does pretty good, almost as good a job as his Talisman T-3H amp, they both sound incredibly similar.

Hi Luke,
I am very interested in your last commments as i have some spare LM3875s.
Could i build a HeadAmp with these?-and how do i make a resistive divider network?

evo

 

[caffeineconsumer]

Quote:

I am very interested in your last commments as i have some spare LM3875s. Could i build a HeadAmp with these?-and how do i make a resistive divider network?

Well to build a dedicated headphone amp with your LM3875s is possible, you'd have to lower the supply rails, run it with lower gain, and probably put some sort of current limiter in the output stage. Exactly how to do this, with optimum results I am not sure, I could probably have a stab at it. But if you'd like me to tweak a circuit diagram around the kit I have built just ask. If you have a kit, or pcb of a current design it will just be a matter of component value changes and maybe a resistor or two on the output, if this is the case just post your circuit on the forum and myself and others will be able to suggest changes. Any information about sources, and the impedance headphones you wish to drive would be helpful. What the end result would be I'm not quite sure, maybe someone else has done this.


But what I did using a network of resistors on the output of my power amp can be used on almost any amplifier to drive headphones.

Rod Elliott shows this technique on his ESP site:
http://sound.westhost.com/project100.htm

The resistor network I'm using isn't quite the same as his, for a couple of reasons:
1. I wanted to drive my amp into a lower impedance as I wasn't sure how it would behave driving a high impedance load.
2. I wanted to provide protection in case of incorrect connection to the amplifier - no chance of shorting the output.

Here is my version:

With my design R(z) will be the load presented on the amplifier (given R(z) << R(s) ).
Making R(s) larger will reduce the amount of power going to your headphones, reducing it will increase the amount of power reaching your phones. Keep it larger than 20 Ohms though.

CAUTION: If you are going to build either of these adaptors (Rod Elliotts or mine) you must be 100% sure your amplifiers outputs share a common ground, otherwise you risk frying your amp

.

Anyway if anything i've posted is unclear, ask about it rather than risking it, we cool

.

Signing off, Luke

 

[rickcr42]

Quote:

Well i was just wondering if any of you could give me a parts list for the borbely all fet headphone amp. I'm real keen to build it but not cashed up enough to fork out 150 euros for the kit (being in australia i have to source the kit from germany).

the closest you will come is this :

http://www.borbelyaudio.com/adobe/hybrid.pdf

which is the hybrid version but if you study the hybrid headphone amp :

http://www.borbelyaudio.com/pics/EB-...20final%20.pdf

..they are the same.

Now compare it to this,the all jfet/mosfet version :

http://www.borbelyaudio.com/pics/EB-602-210%20final.pdf

and you will see the output is identical and the front end is very close,just jfets replacing the 6GM8 tubes.

By studying the three files plus getting an understanding of jfet operating points in these files :

http://www.borbelyaudio.com/adobe/ae599bor.pdf

http://www.borbelyaudio.com/adobe/ae699bor.pdf

you should be able to "tough out" the parts.

And yes I have the list of the parts but no I will not share commercial designs or intellectual property with those who would clone the product but if you build your own version with the information freely available then........

 

[caffeineconsumer]

thanks rickcr42.

the first of your links you provided is most helpfull to me. the others i already have printed out from b4, i found them a good read, yes, thats why i wish to construct, mmmn.

In requesting a parts list I wasn't intending to steal interlectual property i just didn't realise thats what i was asking of ppl. sorry to all head-fiers. i be more thoughtful in the future.

So its time for me to whip out the graphics calculator and nut out some component values. now i'm wondering about any tweaks to the circuit - this is gonna be fun(?)......

 

[rickcr42]

BTW-the borbely hybrid is actually in the Headwize Project library and has been for many years (it was the schematic I used five or so years ago) :

this is the page,a really good read for all ;

http://headwize.com/projects/showfil...=opamp_prj.htm

I will leave the actual schematic up 'til tomorow because I am not sure if it is cool to have it up but the link will stay.

good luck and hoping for a successful build


Rickeraptor

 

[getfi]

Luke
How are those calculations going?
How about a quick design tutorial?
That way we are just looking at a basic circuit topology as a starting point ie the Borbley Headphone amp
Lets start with the output current and voltage relationships assuming a 32 ohm load
I2 R gives us the load power as does V2/R which leads to:
RMS output current of 80mA or 110mApk
RMS output voltage of 6.4 V or 9Vpk
Access to the device data sheets is required from here.
Draw a load line for Q3 (a low current low input capacitance MOSFET) starting to the right of the Vds axis at 24V and Ids=0 (MOSFET off) then draw a line up to a point on the Ids axis between 250 to 350 mA which will give an operating bias point, Q point, of around Vds=-8V and Vgs=-2V at a drain current of 150mA for Class A operation. This is a little to the left of centre, but that’s OK as the drain of Q3 will be at 0V (q point) and extending to –9vpk and at this stage will be nearly off say conducting 40mA as Q4 sinks 150mA ->110mA peak from the load and 40mA from Q3 -> which will have Vds = -34V at 40mA ( no problems there) So back to the Q point->> Vds= -8V, Id = 150mA Therefore
R10 =V/I=16/.15 =110 ohms

Moving down: R14 and C5 a zobel type correction for inductive type headphones say:32 ohms and 0.22uF for 32ohm 0.2mH phones to 0.47uF and 32 ohms for 0.4mH phones
R13 is a current sense for the constant current source R13 = 0.6/.15 = 3.9 ohms.
Q4 is the current source 2sk216 for symmetry? Low noise, but to my mind could be NPN transistor BD139 as could Q5 but either way Low noise is the key
R11 not critical say 5k6 to 10k C8 simple rc filter in conjuction with R11
R12 same as R7 100 ohms is a common value but could be eliminated altogether for more attack ( higher slew rate possibly at the expense of ringing) or increased by a factor of 2 or 3 to tame the attack
R8 and R9 form the feedback divider and I would choose 100k and 10k to match what would apply to the other differential input R1 10k and R2 100k
C4 would be there for high frequency stability and appropriate positioning of the poles and zeros in the splane, but that’s another story>> best dealt with empirically because the overall stability at frequency will be very much a function of careful layout >> watch location of inputs and outputs, keep inter capacitances down and lead lengths short etc
Remember the key to a great differential pair is balance and speaking of balance the best way to drive this baby is balanced via R1 and R9 anything less would be like putting a Vdub engine in a Ferrari.
This brings us to the input 2sk389 >>well balanced

Forget P1 I don’t like pots R5 and R6 should be in the range 47 to 220 ohms Refer to the memory distortion man for a good explanation of emitter (read source0 resistor regeneration.
http://peufeu.free.fr/audio/memory/memory-6-test.html
Final tuning of zero output volts can be achieved more elegantly by using a resistor decade box in parallel with either R5,R6 to achieve balance value >> then whack it in parallel
Look at the sk389 data sheet and note source resistance must be 10k to 100k for best noise result. Study Vds vs Id for linearity and best region of operation also study noise figure versus Id and if your like me I would be choosing operation in the 4 to 8 mA range which definitely means using the bl version or the v version of the Toshiba 2sk389

The constant current source then becomes 8 mA to 10 mA
Choose Nchannel jfet with Idss = 8mA low noise sk170 bl version
Were almost there

C1 is almost redundant with increase in R1 to 10k ohms and 25pf input capacitance of Q1 giving a corner freq of 1.5MHz
R1 100 C1 100pf could be placed across XLR connector to balance input Z but??
This just leaves the current mirror:
Low noise 2SA1015 I believe is a better choice than 2SA872 refer to data sheets
Or why not use BD140 we already know they work well in the input stages of high quality audio amps don’t we luke
Just leaves the current mirror D2 small signal diode load line with R4, R3=R4,
Remember that we have up’d the constant current by a factor of 2
A more elegant solution would be to use 2sa1349 again for balance and low noise
Have a great day
R u right with power dissapation and heatsink calcs? watch power diss in R10
and power diss in Q4

 

[caffeineconsumer]

thanks getfi.

give me a couple of days to work through your post and get my head around it all - it's all good stuff, and trust me i'll get there in the end just gotta get the brain cranking over. any developments worth mentioning i'll throw onto head-fi, actually i might get a website up, we'll see.

anyway gotta go.

btw. you're post is sounding a lot like my father, very much so in fact, and your first post too, wow, you signed up this month also, you can't fool me.
he he, i see you've been thinking about this circuit a little bit, it will be good when those discrete op-amp pcbs arrive.

 

[Jupiter]

Here's some scope pics from one of my Borbely amp boards. The top signal is the input, bottom is output into a 44 ohm dummy load.

50 Hz

100 Hz

1 KHz

10 KHz

100 KHz

250 KHz

 

[Jupiter]

One more scope pic.