Current Buffer Choices?
Jul 20, 2009 at 4:36 AM Thread Starter Post #1 of 12

hiker101

New Head-Fier
Joined
Jul 3, 2009
Posts
25
Likes
10
I am working my way up the DIY amplifier food chain.
I built a CMOY, then a Mini^3, and an Apheared 47.

I think the next step is to incorporate a multiloop topology- which requires a buffer. I have Grado 225i's, so I'm looking for a high current solution.

LMH6321 is used in the Pimeta 2, and seems to be available at digikey. I don't like the SOP package, but I can deal with it. Is this buffer stackable?

BUF634 is currently available at digikey in any quantity I need. How does this compare to the 6321? This is available in DIP8 so that is a plus.

AD8397 puts out plenty of current. Could it be used as a buffer?

Are there any other useful buffer ICs out there?
 
Jul 20, 2009 at 4:57 AM Post #2 of 12
You don't need to stack LMH6321, it can already output a LOT of current. It's a considerably nicer sounding buffer than BUF634.

AD8397 is an opamp, you don't need to use it as purely a buffer.

Look into building a discrete buffer instead, perhaps one of the very nice little ones by Sijosae? I really like his buffers and somebody has now made a PCB for them, pin compatible with BUF634. You can buy PCBs here.
 
Jul 20, 2009 at 6:34 AM Post #3 of 12
The LMH6321 is a *very* nice buffer, and if you can live with the SMD form-factor, hard to beat. Better (IMO, and that of many others) than the BUF634, and (again, IMO) probably any other IC buffer out there. That said, there's no reason to skip the Sijosae buffers, which could be a good start on the discrete buffer road, which eventually leads to... JISBOS?
 
Jul 20, 2009 at 8:34 AM Post #5 of 12
How about making a simple Class A single ended MOSFET or BJT output stage with an opamp gain stage?

You don't get to use jung multiloop but it's super cheap and sounds good with proper settings.

Search for "Tori Amp PRR MOSFET" or just run through FallenAngel's topics he created on head-fi. One of it is his build of a single ended BJT buffer for the opamp.
 
Jul 20, 2009 at 8:58 AM Post #6 of 12
Oh yeah, that one.
smily_headphones1.gif
Simple 1 transistor (BD139 used), deep class-A at about 100mA biased by resistor. Very simple amp, just throw in a decent PSU that can supply 300mA current and you're good to go.
 
Jul 20, 2009 at 1:32 PM Post #7 of 12
Quote:

Originally Posted by FallenAngel /img/forum/go_quote.gif
Look into building a discrete buffer instead, perhaps one of the very nice little ones by Sijosae? I really like his buffers and somebody has now made a PCB for them, pin compatible with BUF634. You can buy PCBs here.


Interesting little buffer. Is there any more details on it? Sound quality? Voltage requirements? Doesn't seem to have any adjustments for bias or dc offset...
 
Jul 20, 2009 at 4:09 PM Post #8 of 12
Quote:

Originally Posted by Mr.Duck /img/forum/go_quote.gif
Interesting little buffer. Is there any more details on it? Sound quality? Voltage requirements? Doesn't seem to have any adjustments for bias or dc offset...


Search for "monofied sijosae buffer" and you should find many details. How does it sound? Better than a BUF634 (IMO); different than an LMH6321.

Quote:

Originally Posted by FallenAngel
Oh yeah, that one. Simple 1 transistor (BD139 used), deep class-A at about 100mA biased by resistor. Very simple amp, just throw in a decent PSU that can supply 300mA current and you're good to go.


A BD139? I was looking at the Tori Amp just the other day (after I'd spend an hour almost completely reinventing the wheel, FWIW) and I'm pretty sure it was an IRF510/IRF610 MOSFET as the output. Something very much along these lines, where the buffer aspect is concerned.

In the interests of equality, I think the Head-Fi community needs a "Kari Amp", as well. Something small, sleep, sophisticated, and nice to look at... probably with a red PCB, as well.
wink.gif
</MythBusters fan>
 
Jul 20, 2009 at 5:38 PM Post #9 of 12
Jung shows CFA op amps with gain as the high current output stage in some of his multiloop circuits - I've used the TPA6120 - not many "debugged" diy amp projects use this but the specs for audio are great

using AD8397 with gain as a multiloop output current buffer/amp could be fine although its not as fast as most CFA op amps and should be run with its local gain set to >2 for greater loop stability and because of the input common mode range limit

frequency compensation can be an issue - not every audiophile op amp rolling suggestion would work as a input op amp in such a multiloop

loop stability in this style multiloop amplifier is easiest to obtain by setting the output op amp gain to ~sqrt(Av)
where Av=overall gain of the composite amplifier=outer loop feedback gain setting
then only use unity gain stable input op amp with GBW < AD8397 GBW

these op amps have a "power pad" - the lead frame that the chip is mounted on is exposed on the bottom of the package so direct soldering to the PCB can provide good heatsinking - its not obvious how well a hobbist can hand solder this
 
Jul 20, 2009 at 6:07 PM Post #10 of 12
Quote:

Originally Posted by Mr.Duck /img/forum/go_quote.gif
Interesting little buffer. Is there any more details on it? Sound quality? Voltage requirements? Doesn't seem to have any adjustments for bias or dc offset...


It's called a Sijosae Class-A buffer. Very simple and sounds quite nice. Voltage swing isn't huge, but can run off 5V to above 30V, I've run it at both (AD823 at low voltage, OPA2134 at high-voltage). Wrap it in the feedback loop of an opamp and you don't need to correct DC.

Quote:

Originally Posted by Nemo de Monet /img/forum/go_quote.gif
A BD139? I was looking at the Tori Amp just the other day (after I'd spend an hour almost completely reinventing the wheel, FWIW) and I'm pretty sure it was an IRF510/IRF610 MOSFET as the output. Something very much along these lines, where the buffer aspect is concerned.

In the interests of equality, I think the Head-Fi community needs a "Kari Amp", as well. Something small, sleep, sophisticated, and nice to look at... probably with a red PCB, as well.
wink.gif
</MythBusters fan>



http://www.head-fi.org/forums/f6/mak...e-help-277069/

Latest schematic is NOT on first page, somewhere on 3rd.
 
Jul 21, 2009 at 5:06 AM Post #11 of 12
Wow, I've just been sucked into hours of reading about discrete buffers. Just the kind of suggestions I was hoping for.

Quote:

Originally Posted by FallenAngel /img/forum/go_quote.gif
You don't need to stack LMH6321, it can already output a LOT of current. It's a considerably nicer sounding buffer than BUF634.

AD8397 is an opamp, you don't need to use it as purely a buffer.

Look into building a discrete buffer instead, perhaps one of the very nice little ones by Sijosae? I really like his buffers and somebody has now made a PCB for them, pin compatible with BUF634. You can buy PCBs here.



I've always wondered why buffers were stacked in the Pimeta and PPAv1. BUF634 also puts out lots of current. I assumed that the stacking addressed some other short-coming of the buffer. It sounds like that doesn't apply to the LMH6321. I'm glad to see that there is some concensus as to the best opamp solution.

I think I'll start with LMH6321 opamps and leave the option to upgrade to discrete buffers. I intend to solder my own DIP adapters for the LMH6321, I'll just use the same pins as the BUF634.

Quote:

Originally Posted by Nemo de Monet
... That said, there's no reason to skip the Sijosae buffers, which could be a good start on the discrete buffer road, which eventually leads to... JISBOS?


Quote:

Originally Posted by JamesL
The lmh6321 is quite nice imo. If you want a discrete buffer, might as well make the PPAv2


Ha ha. PPAv2, or M^3. I certainly expect to head down that road eventually. I want to try a few different opamps first and see if I can actually hear any differences. Too many variables ...
darthsmile.gif


Quote:

Originally Posted by TzeYang
How about making a simple Class A single ended MOSFET or BJT output stage with an opamp gain stage?

You don't get to use jung multiloop but it's super cheap and sounds good with proper settings.

Search for "Tori Amp PRR MOSFET" or just run through FallenAngel's topics he created on head-fi. One of it is his build of a single ended BJT buffer for the opamp.



Reading back through FallenAngel's posts led me to all sorts of fun stuff, a lot of digital sound issues... I never found the amp you mentioned. Good thing FallenAngel linked it below. It sounds like a room heater. I need to build a power supply first. Maybe Sigma-11.
 
Jul 21, 2009 at 5:22 AM Post #12 of 12
Quote:

Originally Posted by jcx /img/forum/go_quote.gif
Jung shows CFA op amps with gain as the high current output stage in some of his multiloop circuits - I've used the TPA6120 - not many "debugged" diy amp projects use this but the specs for audio are great

using AD8397 with gain as a multiloop output current buffer/amp could be fine although its not as fast as most CFA op amps and should be run with its local gain set to >2 for greater loop stability and because of the input common mode range limit

frequency compensation can be an issue - not every audiophile op amp rolling suggestion would work as a input op amp in such a multiloop

loop stability in this style multiloop amplifier is easiest to obtain by setting the output op amp gain to ~sqrt(Av)
where Av=overall gain of the composite amplifier=outer loop feedback gain setting
then only use unity gain stable input op amp with GBW < AD8397 GBW

these op amps have a "power pad" - the lead frame that the chip is mounted on is exposed on the bottom of the package so direct soldering to the PCB can provide good heatsinking - its not obvious how well a hobbist can hand solder this



I was a little worried that the 8397 would be more work than a buffer since it seems to be a bit finicky. The nice thing is that it's dual channel and so is about half the price of a real buffer. Hmm, I think I like the LM6321 solution better- that will allow me to use trickier opamps in the gain stage.

I just built the MINI^3 and used heat-sink paste to thermally couple the chip to the circuit board. Soldering it would probably be beyond me. At my normal listening levels I don't think I'm stressing the chip much, so I bet the paste isn't even necessary.
 

Users who are viewing this thread

Back
Top