BPA200 PCB Layout

[nterikun]

Hi Guys.

My name is Narong. I'm interested in BPA200 based on AN-1192 with snubber part from Carlos Filipe. Read many threads about these topics, and I decide to build one to replace existing HK3480. My speaker is EPOS M5.

The schematic is mix between AN-1192 and Carlos snubberied.
Question is

- Will this schematic work.
- Any flaw on PCB Layout.

PCB is double size with PTH and plan to put ground plane on top layer. For your review I haven't put it in.

See PDF of schematic and layout here.


schematic and pcb

[troystg]

Have you seen Zang's?

He did one and had a run of PCB's made. He may be able to send you his file, or better yet he may have a couple of boards left over you could buy and save your self the time.

But you would still need to do the pwr supply boards.

[jackinnj]

Putting in the RC network where you have drawn it defeats the purpose of the decoupling capacitor on the LM3886, LM4780, LM3875.

[sr2002]

brilliant layout man.

woooooooowwwwwwwwww.

i am no expert but dont u think the power supply tracks and output tracks should be thicker than that?

[peranders]

I haven't checked the pcb layout int detail but remember that each LM3886 is capable of 11-12 A (= 20-25 A!) each and compare this with your copper traces. Go for wider power traces, 70 um (2 oz.) copper is good.

Your 3 W snubber resistors are way to large. Better to go for 1 ohm metal film 0.6W, one to three of those. 3.3 nF won't probably do any good, no harm either. If you want to snubberize why don't you use 100 nF+ 1 ohms // 100 uF maybe also your 2200 uF.

Your MUTE connection ought to be common. Connect all IC's and use only one resistor which have to be recalulated.

If you plan to have groundplane you must separate power ground and signal ground, no current at all in the signal ground.

[nterikun]

Have you seen Zang's?

I understand that Zang's PCB had some bug in it, and that threads was just freeze no activity at all. Beside there's no part of snubberied that I want.

Putting in the RC network where you have drawn it defeats the purpose of the decoupling capacitor on the LM3886, LM4780, LM3875.

I thought that was the way to reduce the output impedance of PSU, isn't it?

i am no expert but dont u think the power supply tracks and output tracks should be thicker than that?

Thanks man.

each LM3886 is capable of 11-12 A (= 20-25 A!)

Could please recommend what track width should I use.

Your 3 W snubber resistors are way to large. Better to go for 1 ohm metal film 0.6W, one to three of those. 3.3 nF won't probably do any good, no harm either. If you want to snubberize why don't you use 100 nF+ 1 ohms // 100 uF maybe also your 2200 uF.

I just follow snubberize design on this url http://myweb.tiscali.co.uk/nuukspot/...clonesnub.html
Another reason is that I couldn't find 0R1 1W from farnell, so I go for 3W.
On MUTE connection, Is there any problem if I get rid of capacitor.

Talking about ground, ground plane and star ground which one give more efficiency ground.

Anyway, based on layout, the input signal doesn't look good to me. I will need sometime to relayout.

[peranders]

Trace width should be 100-150 mils (at least, don't forget fuses!) in 70 um copper.

The snubber document has info which can't be verified and Nuuk it self isn't behind it.

If you check my signature below about snubbers and my references in my snubber document from simulations and real world measurements you will discover that things don't match.

http://www.sjostromaudio.com/phpBB2/...52d4e116efe#28


1 What happens if you don't have a snubber?

2 What happens if you do exactly as Carlosfm suggests?

3 What happens if you do as Joseph_k and KYW suggests?

Something to think about.

The MUTE doesn't has to have a cap but it is very convenient to have a smooth startup.

[mauropenasa]

Hi nterikun,

looks the my another link of PCB dev...

http://www.diyaudio.com/forums/showt...443#post722443

Ciao

Mauro

[jackinnj]

Originally posted by peranders


1 What happens if you don't have a snubber?

2 What happens if you do exactly as Carlosfm suggests?

3 What happens if you do as Joseph_k and KYW suggests?

When you have to start worrying about milliohms of impedance at 100 MHz, you are seeing elves and fairies as well. Time to push away from the poteen and switch to coffee.

JosephK did very nice work with his network analyzer -- you have to separate the elements before you can analyze their interaction -- but all of the "snubber" discussions seem to neglect the big honking leakage inductance in the transformer secondary which is at least one order of magnitude larger than the ESL of the filter caps.

[mauropenasa]

...JosephK did very nice work with his network analyzer -- you have to separate the elements before you can analyze their interaction -- but all of the "snubber" discussions seem to neglect the big honking leakage inductance in the transformer secondary which is at least one order of magnitude larger than the ESL of the filter caps.

Absolutely, Jack.
And the "bad" "frequency-dependent modulation feedback " created by the wire (RL) GND connections among the PS and amp (if the signal/reference GND is common to power/bypass GND)...

Ciao

Mauro