XGC Board Try

Hello,

After building a LM3875 VBNIGC, i planned to build a PA100 to driver speaker based on scanspeak 15w4531g00/vifa xt25. These two in 9L box have 5ohm impedance and 87db sensitivity.
After reading GCsupersymety this option seduce me more than the PA100 since i have a built bosoz in stock. So here my latest XGC Board try. It's done under Proteus. I have too much problems with eagle....:bawling: so i follow under proteus.

[IMGDEAD]http://idefixes.phpnet.org/illu/av/xgc/lay_xgc.gif[/IMGDEAD]

The Metalman modify schématics. I intergrated the end part of the 338 regulated subbered PSU. 330nf from these will be direct soldered between lm3886 +/- pins. Electrolytics/small MKP near Lm3886 are 2200µf/50V // 100nf.

[IMGDEAD]http://idefixes.phpnet.org/illu/av/xgc/xgc1.gif[/IMGDEAD]

Marc
 
Carondimonio said:
The PCB looks really great!
I' ve been checking the layout and didn' t see anything wrong, but I' d be nice to hear some PCB guru' s opinion...

BTW, I' d be interested in having a couple of PCBs, if there will be a GB...

Thanks. It will be nice allready to have another opinion before i try to realise this board.

Marc
 
cjd said:
Last I knew, BrianGT was working on a GB.

C

Yes that's what i read in the dedicated topic, but since no news. My way is a little different : the final cut is not a GB, but my own use since à be abble to realise board at home. Of course every DIYer can have acces for his own use the board file under proteus. But for this time i need some help to check the current version and to optimize it. I have all necessary parts to built two mono blocs. Psu based on latest carlos version with 300Va/2x25v, mur1560 rectifier bridge,mallory 22.000µf/40V Cap, Lm338T regulator per bloc. I have two rubbish heatsink (200x200x40mm)

[IMGDEAD]http://idefixes.phpnet.org/illu/av/xgc/dissipxgc.jpg[/IMGDEAD]

So the board and her optimisation are the last point to progress.

Marc
 
Yeah - I'm not waiting for the GB either. In fact, I've had boards designed for a couple months now, just collecting parts at this point. :)

Of course, I want to tweak them a little now to allow for some additional testing of some of the other topology tweaks that have been posted since then (mostly, make room for larger power caps at the chip). :)

I'm assuming you're planning on soldering the feedback resistor directly to the chip pins?

I'm curious about the really long power traces (I tucked them between the chips).

And, while I'm unquestionably not an expert (or even close) on the subject, your grounding scheme seems less than ideal - particularly with the long spread-out trace.

Of course, it also really feels spread out to me, but I tend to shoot for more packed layouts to shorten traces. I kind-of do this compulsively - whether it makes sense to do so or not. :) (I think the layout I have is about 2/3 the size of yours)

I need to order a pile of resistors and a few more caps and I think I'll have everything I need.

C
 
cjd said:
Yeah - I'm not waiting for the GB either. In fact, I've had boards designed for a couple months now, just collecting parts at this point. :)

Of course, I want to tweak them a little now to allow for some additional testing of some of the other topology tweaks that have been posted since then (mostly, make room for larger power caps at the chip). :)

On your layout you have only 2 electrolitcs on board for two cheaps. I plane to followed latest Carlos recommandations : put couple of 2200µf/100nF nearest each LM pins

I'm assuming you're planning on soldering the feedback resistor directly to the chip pins?

This point is not really optimised on these version boad i will try to put R13 abd R14 nearest the cheap thean yet but R7 and R8 as they assumate negative feedback on X structure

I'm curious about the really long power traces (I tucked them between the chips).

I saw that on "GCSS boards? Brian?" topic it's a solution but i don't know how it feet with 4 2.200µF electrolytics

And, while I'm unquestionably not an expert (or even close) on the subject, your grounding scheme seems less than ideal - particularly with the long spread-out trace.

My ground scheme is a line that at first consist in Signal gnd before reaching PSU gnd. I never had pbs with this configuration. My first t on GC,, it runs without hum and offset at first try, comparating to all post "help with hum on BrianGT kits" we can see on diyaudio. Not that the kits is bad but of wiring missunderstanding. I well according that this point could be well developped in the currnet xgc version board

Of course, it also really feels spread out to me, but I tend to shoot for more packed layouts to shorten traces. I kind-of do this compulsively - whether it makes sense to do so or not. :) (I think the layout I have is about 2/3 the size of yours)

Size is not for me the first point. It cames after good placement of part and dimension size of lane.

I need to order a pile of resistors and a few more caps and I think I'll have everything I need.

C

It's good to see points to discuss. I'm beginner in this domain, it's for me a good way to learn et improuve the board.

Marc
 
That makes SO much more sense to me, with the power coming in the middle.

I also very much like the placement of the input caps. You could theoretically run the trace at the outside edges of the board further to the edge with a few more holes allowing just about any cap you please to be used here - including some of the larger film/foil "audiophile" caps if you so chose. :)

FWIW, my board was made before Carlos had decided regulated sounded best again. ;) At that time, it seemed low capacitance at the chip was ideal so a single cap per rail, or a pair - didn't seem much of an issue. My LM3886 board was a *very* quick effort on my part - I'm using paralleled LM4780's at each end of the X and that's where I've put most of my effort (it also uses two caps per rail, one at each chip).

C
 
The actual input impedance works out to be ~3Kohm per polarity, and yes it is lower than typical, but nothing a decent preamp can't drive properly. The circuit does not perform it's best with a passive volume control, where it looses it's liveliness and soundstage focus. To get the most out of the Supersymmetry the circuit as designed uses current feedback to the input differential pair. This required some optimization to find a good balance, as higher input impedance decreased the supersymmetric current feedback.

Idefixes, very nice work on your board layout.

Cheers, Terry
 
Input Z

You are right that most pre-amps can probably drive 3 K, however it does now make an amplifier that interacts more than most with the upstream component. Tube pre-amps in particular may have quite a different sound signature than a low output impedance transisitor design would using this amp.

Because of the low input impedance and feedback current right into the input termination, some of the amps feedback current will likely interact with the output stage of the pre-amp and it's feedback circuits (if there are any).

Anyone using a high Z pre-amp should carefully consider possible side effects.

One reason a passive pre has problems is because its internal parts just became part of this amplfiers feedback circuit. This changes the way the amplifier works. Worse yet, for many passive pre amps each volume setting will change the sound character slightly.

This amplifier may be more sensitive than most to cables, especially to cable capacitance.