For starters it would be good to know what looks like your PSU part you would like to use...like how many transformers, rectifiers (transformer with separate secondaries or not, one rectifier per channel or two rectifier per channel, each channel with its own transformer, rectifier and filtered or not, etc)
The power supply is a separate large unit consisting of:
* Power conditioning unit (AC filter).
* Soft start circuit.
* Two 500VA transformers.
* Two rectifier half-bridges (BYV32)
* Capacitor bank 40000uF per rail.
* Two regulator boards (15A per channel).
* 4700uF output capacitor per channel.
Best regards.
Last edited:
One thing we all should have learnt from the P2P prototype is that (mixed) ground planes are not a good design practice.
What are you trying to build? A nice stereo amp or the best possible sounding LM3886.
How much do you want to spend? What parts do you have all ready?
I think this would be where to start.
What are you trying to build? A nice stereo amp or the best possible sounding LM3886.
How much do you want to spend? What parts do you have all ready?
I think this would be where to start.
Last edited:
When you say mixed you mean power+signal or star ground versus ground plane?
Best regards.
Are the transformers with split secondaries or common ground wire?
Two rectifier bridges for one channel or both?
Two physically separate transformers.🙂
Half bridge per channel.
Best regards.
Nearly impossible to run each and every trace to the power ground on a single layer PCB.
sghr220 ,
C2 is a 22uF normal polar electrolytic . Positive terminal facing the load ( power amp input ) . But if you eliminate the input cap and have dc at the input of the buffer, check with your DMM what it's polarity is. Assume ( and measure if you want ) the voltage at the power amp input. May be something very close to zero. Maybe be +/-0.04 V or less. Under normal conditions the dc at the output of the buffer is at -0.65 Volts so it's more negative than the power amp.
C2 is a 22uF normal polar electrolytic . Positive terminal facing the load ( power amp input ) . But if you eliminate the input cap and have dc at the input of the buffer, check with your DMM what it's polarity is. Assume ( and measure if you want ) the voltage at the power amp input. May be something very close to zero. Maybe be +/-0.04 V or less. Under normal conditions the dc at the output of the buffer is at -0.65 Volts so it's more negative than the power amp.
Yes, I know that, but the BIG question is do they have a split secondaries (4 wires) or common ground wire (3 wires)?
Use a jumper when you really need one, it's that simple😉...Ground plane could be of use in line level circuit (read low circulating current), but for amps like this one we are contemplating about, is of no use.
Hi is talking about his buffer, I think http://www.diyaudio.com/forums/atta...41d1394623098t-inverting-lm3886-buffer-01.gif
Speaker return and C6 (zobel) should return to C3 and C5.
Feedback loops should be as short as possible.
I would use separate PSU for the LME49710. Remove R2, R3, D1, D2.
Feedback loops should be as short as possible.
I would use separate PSU for the LME49710. Remove R2, R3, D1, D2.
They do.
They are.
Why? Is it worth it or just your opinion?
PCB Review:
-You'll have trouble accessing the output connector.
-The ground return for the Zobel is really long. See if you can pick up ground closer to the decoupling caps.
- You can scoot up the VEE decoupling cap on the LM3886 and rotate the VEE cap 90 degrees. That moves those caps closer to the LM3886.
- The VEE decoupling on the LME49710 should move to the other end of the chip. You'll have lower inductance through the ground plane than through the supply trace, so moving the cap will give you lower ESL in the decoupling, hence, improve the effectiveness of the cap.
- It doesn't look like your ground plane is connected to the bypass caps. That could just be obscured by the colors chosen. Ensure that you have a good thermal relief on the connections and that the ground plane actually connects to ground.
- You'll need mounting holes. It's not a good idea to just let the PCB dangle from the LM3886.
- I'd consider moving Rmute out from behind the IC and bring the IC closer to the edge of the board. That way you can connect directly to a heat sink alongside the board.
~Tom
-You'll have trouble accessing the output connector.
-The ground return for the Zobel is really long. See if you can pick up ground closer to the decoupling caps.
- You can scoot up the VEE decoupling cap on the LM3886 and rotate the VEE cap 90 degrees. That moves those caps closer to the LM3886.
- The VEE decoupling on the LME49710 should move to the other end of the chip. You'll have lower inductance through the ground plane than through the supply trace, so moving the cap will give you lower ESL in the decoupling, hence, improve the effectiveness of the cap.
- It doesn't look like your ground plane is connected to the bypass caps. That could just be obscured by the colors chosen. Ensure that you have a good thermal relief on the connections and that the ground plane actually connects to ground.
- You'll need mounting holes. It's not a good idea to just let the PCB dangle from the LM3886.
- I'd consider moving Rmute out from behind the IC and bring the IC closer to the edge of the board. That way you can connect directly to a heat sink alongside the board.
~Tom
Why is a ground plane of no use? It minimizes the ground impedance (including inductance if my understanding is correct). Isn't that what we want?
~Tom
I'm guessing he means mixing the low current circuit ground (LME49710) with that of the high-current circuit (LM3886).
You could pour a ground plane around the LM3886 and another around it for the LM3886. Then connect the two by the ground star. The ground star would have a thermal relief to the high current ground plane and a trace from there to the low current ground plane.
You don't want the load current creating ground bounce in your low-current circuits. Without breaking the ground plane, you have no control over where in the ground plane the current actually flows.
~Tom
Thanks a lot Tom 🙂, it'll take me some time to translate your advice into the layout but here are some quick remarks...
Need not to be an output connector could just solder the speaker cable directly into the board, anything to keep traces short and the overall board compact as possible.
Can i use a jumper wire in this area or is it too sensitive?
I don't fully understand that but i'll try to get the Vee 100nF decoupling closer to the chip.
Will do. 🙂
- It's connected to the ground plane, just the color. Thermal relief on the pads connecting the two ground planes?
- Mounting holes will be added in the final version and copper standoffs will be used to provide clearance.
- Will bring the chip closer to the board edge but Rmute has very small effect on the heat sink mounting either vertically or horizontally.
You mean one around the LM3886 and a global connecting both the LM3886 and the LME49710?
Best regards.
Need not to be an output connector could just solder the speaker cable directly into the board, anything to keep traces short and the overall board compact as possible.
Can i use a jumper wire in this area or is it too sensitive?
I don't fully understand that but i'll try to get the Vee 100nF decoupling closer to the chip.
Will do. 🙂
- It's connected to the ground plane, just the color. Thermal relief on the pads connecting the two ground planes?
- Mounting holes will be added in the final version and copper standoffs will be used to provide clearance.
- Will bring the chip closer to the board edge but Rmute has very small effect on the heat sink mounting either vertically or horizontally.
You mean one around the LM3886 and a global connecting both the LM3886 and the LME49710?
Best regards.
- Status
- Not open for further replies.