[morsel]

How would people feel about replacing the TLEs with resistive dividers? Using 12K or greater resistors for a current of 1mA or less would more than cover the miniscule input signal level. The resistors are cheap, quiet, and would fit in the existing layout without having to sacrifice the separate isolated rails for each channel. We will be testing this.

(Morsel puts up a frail umbrella as protection from the incoming bunker buster bombs.)

 

[Dreamslacker]

Well... I've experimented with using resistive dividers instead of TLE2426CLP/P on my A47 variant, Proto42 & META42 amps. I can safely say that for me, I get a more open sound with the resistors than with the TI part. No idea why that's the case though.

 

[ppl]

I am adamantly against resistive Dividers. The isolation effect caused by the voltage divider effect of the Jfets and the Low output impedance of the TLE's. the isolation could go down to 10dB vs the 100 Plus we now get. other alternativs are available hear and these include keeping the present arrangement however implemented in another package

 

[AIM9x]

I wish i had some more audio circuit knowledge so I could better help you guys out.

4 weeks into an electronics course and i'm starting to understand more about what's going on. This week, we're going to do amplifier configurations and I think i'll have a heavy understanding of what's going on here,

.

 

[eric343]

Quote:

Originally posted by morsel How would people feel about replacing the TLEs with resistive dividers? Using 12K or greater resistors for a current of 1mA or less would more than cover the miniscule input signal level. The resistors are cheap, quiet, and would fit in the existing layout without having to sacrifice the separate isolated rails for each channel. We will be testing this. (Morsel puts up a frail umbrella as protection from the incoming bunker buster bombs.)

/me hops in a B-52 and heads down to California...

Weren't you the person who was adamantly defending the use of the TLE2426 in the M/ETA42?

And unless you tombstone the resistors, wouldn't the SOIC package of the TLE2426 be the same size as two resistors side by side?

Incidentally, isn't the TLE2426 availible as bare silicon? Now THAT's small

(we'd have to get aos to solder them...

)

 

[guzzler]

a bog standard resistive divider sounds like a non sense in an amp that is supposed to be as high end as solid state is going to be. It is far easy, and more space effective, to use a TLE in another package, and the matching of +/- rails is going to be far more effect with a precision trimmed TLE, than 1% matched resistors... Don't cut corners when you don't have to!

I've never heard it, but I would think that the more "open sound" dreamslacker speaks of is purely a placebo effect!

g

 

[morsel]

This is a contingency if the part is really discontinued. There is not space for 3 DIP or SOIC TLEs on the PPA. If we lose the TLE we will resort to plan B, which is still being hashed out.

 

[mekanoplastik]

edit...

never mind ..i just saw a thread discussing my question... (about Tangent stocking 2426's as he did with meta buffers)



m.

 

[morsel]

Mek, it's a nice thought, but we are not going to design the PPA around a discontinued part. Hopefully TI will get back to me within the next couple of days so we will know for sure what is going on.

 

[morsel]

Errata: I posted the wrong C1 capacitor dimensions on 4/29 which I have just corrected.

According to Tangent, the max height limit is about 28mm. 1000uF 35WVDC 12.5x25mm Panasonic FC series are the recommended parts.

 

[yejun]

if using a real ground, is the ground channel still necessary?
another question is if using multi-battery, can i just put half of batteries to v+ another half to v-?

 

[ppl]

No you can't split the Batteries in half. unlike conventional Battery Amps that use a Virtual ground driver in the Primarry power supply the PPA uses a Virtual ground driver on each Channel in the Low current only section powering the opamps.

If you delete the Ground channel you will defeat one of the real nice features of the PPA and if that is the direction you wish to go perhapps a more conventional Amp such as the META42 is what you want.

 

[morsel]

PPA v0027 layout and schematics are up for viewing at http://elvencraft.com/ppa/

Changes include a fatter power bus, added RLED to save time jumpering the Zener and putting a resistor in the CRD spot for those who will not be using a current source for the LED, added no fly zone circles around the screw holes, and landing pads for the ALPs pot prongs.

These prongs are little "feet" near the pot pins and are electrically connected to the pins. The pot actually sits on these prongs. They were shorting out against the ground plane on the prototype. There are now square landing pads where the prongs touch down. Under each silkscreen landing pad for the left and right channel prongs the ground plane is removed, so even if the prongs cut through the silkscreen and solder mask layers they will not short out. Ground plane remains under the pads for the ground prongs. It is important to make sure the pot pins are straight and perpendicular to the pot body for proper alignment of the landing pads.

These prongs are probably only an issue for the prototypes, which have no solder mask or silkscreen layers, but better safe than sorry. I would hate to have to make everyone lay down electrical tape, rubber, or cardboard to protect the board from the pot prongs.

Note that the square holes in the ground plane are not visible in a gif of the layout as the silkscreen covers them up.

Kurt and I have been busy lately. Testing will likely take many more weeks.

 

[Tim]

Looking good team ppa!

 

[yejun]

oops.
I think the ground channel's current should 1.4 times the right or left channel under the condition random L and R.

Ig=Ir+Il
<Ig^2>=<Ir^2>+<Il^2> the cross term will be zero if random.
RMS(Ig)=sqrt(2)RMS(Ir) if Ir=Il

so on random case, the ground channel current will 1.4 times individual channel.
At high frequency the current can be assumed as random, but at low frequency the R and L channel is correlated, then the ground current is twice the individual one.

So I suggest at least use 1.5 buffers at ground channel.