[smeggy]

Thanks Dan

 

[Edwood]

Wow. nice implementation of the volume control.

And another nod to the use of 4 Pin XLR.

 

[rds]

wow, what a fantastic design!

I have to build one of these

 

[digger945]

I guess this is what you were talking about huh. Funny, I could swear I have seen those red and green schematics before *scratches head*

So just a few small changes in resistor values within the Vbe multiplier and use of IRF mosfets would be one alternative to the hard-to-come-by Renesas parts.

Quote:

Originally Posted by j4cbo /img/forum/go_quote.gif M4 uses lateral MOSFETs rather than vertical MOSFETs. The latter are cheap and robust, but designed for switching power supplies rather than operating in linear mode. Although one can build a great-sounding amplifier from vertical FETs, they require inordinately high bias currents to achieve good linearity. Although lateral FETs are significantly more expensive, they are inherently superior for audio purposes

The data sheets show somewhat different transfer characteristics for each, but I wonder exactly how much of the transfer curve is needed to satisfy headphone duty(I'm looking into it on another project myself

)

What's not to like about this amp, I want one!!!

 

[smeggy]

Indeed!

 

[bada bing]

Quote:

Originally Posted by FallenAngel /img/forum/go_quote.gif So... where do I get my hands on a board?

ditto, I'll be watching for a chance to get in on a board buy.

This looks like a very good design to integrate the differential outputs of a WM8741 DAC chip on. It seems like you could eliminate a bunch of buffering and cabling monkey motion pretty easily.

Quote:

Originally Posted by grawk /img/forum/go_quote.gif I just deleted a bunch of posts trying to get this thread back on course. Take the ground debate to the other thread.

Needs to happen more frequently IMO

 

[j4cbo]

Quote:

Originally Posted by bada bing /img/forum/go_quote.gif This looks like a very good design to integrate the differential outputs of a WM8741 DAC chip on. It seems like you could eliminate a bunch of buffering and cabling monkey motion pretty easily.

I don't want to say too much too early, but... I expect to have an all-in-one featuring asynchronous USB, a Xilinx FPGA-based 16x interpolation filter (the design goal is "as good as or better than PMD200"), 4x PCM1704 DAC, and this sucker sometime in 2010.

 

[smeggy]

heh, and all for a projected price of under $300

That's what you said right?





hehe

 

[j4cbo]

Quote:

Originally Posted by smeggy /img/forum/go_quote.gif heh, and all for a projected price of under $300 That's what you said right? hehe

Ahahahahahahahahahahahahahahahaha

 

[qusp]

Quote:

Originally Posted by j4cbo /img/forum/go_quote.gif I don't want to say too much too early, but... I expect to have an all-in-one featuring asynchronous USB, a Xilinx FPGA-based 16x interpolation filter (the design goal is "as good as or better than PMD200"), 4x PCM1704 DAC, and this sucker sometime in 2010.

awesome, something to do with the 6 1794A I have sitting right here

it'll take me that long to get around to doing something with them anyway I reckon since I have the buffalo32 here. was looking for a project to use them for, considered trying a transportable solution with them and DIR9001, which may still happen, but the async USB is making me think of either waiting, or just using 2 in the transportable. either way, i'm VERY interested in this amp, i'll put my hand up for a board/kit/whatever right now.

and yes, awesome work on clearing the 'ground' so this thread could move on.

 

[Lil' Knight]

The project looks very interesting.
Did you have a chance to compare with the M3? I really like how the M3 sounds and it'd great to have an 'upgraded' version.

 

[j4cbo]

I don't have an M3, sadly... I think I'll ship a prototype out to some folks in California with lots of things to compare with and have them do a shootout.

 

[doobooloo]

Great idea and design - amazing how many of your design goal categories match what I have in mind (i.e. my balanced TPA6120 project in my sig).

I'm really interested in how this develops, and I'd definitely be interested if there's further development that leads to a group buy of some sort.

One thing that I regret the most from my first run of boards on my balanced TPA6120 project is that I didn't make the opamps socketable... Part of the joy of DIY is the ability to tweak and try different things - and for me on my amp, it's a big itch not being able to try different buffer opamps feeding into the TPA6120 driver (I did once, but wasted two otherwise perfectly good opamps in the process)...

It looks like there may be enough room for DIP8 sockets with a bit of creative re-shuffling of the board - would this be something that you have in mind?

Anyway, stellar job and I look forward to seeing how this develops!

 

[j4cbo]

Yeah, once I'm confident enough in the state of the design (read: have properly benchmarked the regulator, and have had other people take a listen and say that it sounds good), I think I'll do a group buy.

I also have a metric assload of the smaller surface-mount components (I got a lot of reels of 1% 0805 resistors at a surplus place this summer), so my plan is to offer kits that have a board, preprogrammed microcontroller, and most of the resistors and discrete transistors. Builders will still need to get all the "expensive parts" themselves, but there should be a *lot* fewer line items in the order than there otherwise would be.

There don't seem to be that many fully differential opamps out there, so I'm not sure how useful socketing the opamps would be; I'm also wary of adding any extra lead inductance with high-speed opamps like the 4131.

 

[jcx]

I hope the R37,38 are open - the Vcom inputs of the diff op amps try to split the supply V and connecting to one rail would a bad idea

further the data sheet highly recommends bypass C on the Vcom pin - I would tie the 2 chip's Vcom together and make a AC gnd with a pair of well matched caps, one to each op amp supply rail

the fact that the ths4031 servos the output to Vcom does give you an implicit mid supply active gnd

the feedback+input resistor strings provide a DC path from the inputs to the Vcom active gnds (common mode)

recognizing the feedback R path to the Vcom active gnd then makes R39-44 largely irrelevant until they are reduced to the below the feedback R value


I also find the cascode fet ccs more headfi fashion than sound engineering - you loose too much headroom to "light up" the cascode fet ccs and in this unity gain output stage bias application you may not want to give up a few more V on the negative swing
http://www.diyaudio.com/forums/showthread.php?t=86626

bootstrap R or bootstrap the supply to a single fet ccs and you'd have more headroom and better output impedance over the full Vsupply swing, the 2 bjt feedback ccs has low headroom requirements if you don't want to use more caps for the bootstrap circuits


gate drive R should also probably be different values to give same bandwidth with the different input C of the N vs P mosfets


and pcb room and holes for heatsinking the mosfets could be good for those wanting enough bias current for Class A into Grados or K701 and enough V for 600 Ohm Beyers - there are some narrow clip-on heatsinks that may not need mounting holes
http://www.aavidthermalloy.com/cgi-b...=576802b03200g


the Q5,6,25,26 "C multiplier"/start up circuit has some logical problems as well - the fet ccs are in series - which doesn't work well: on startup when there is enough V drop across them to regulate the one with the infinitesimally higher current regulation setpoint will simply be saturated - ie a slightly nonlinear doped Si resistor

jfet tolerance also spoils the visual circuit symmetry: when operating with only 1 diode V Q5,6 jfets look like two lousy tolerance, poor linearity “resistors” shorted by C7 forming a V divider so the hoped for differential supply filtering function has some
differential-to-common mode conversion

just use 1% Resistors – if this circuit’s functionality is needed at all - what's the magnitude of the supply noise or "image" of the load current in the regulated supply and how much more filtering does the ths4031 need with ps supply rejection already >90 dB to way beyond 200KHz?