Here's a tentative layout for a beefed up "supergainclone". Not exactly original, just trying to get it right.
The input buffer, the inverting stage and the servo are provided by an opa4134.
The sharing resistors are 3W 2512 smd resistors in case you wonder.
What do you think ?
The input buffer, the inverting stage and the servo are provided by an opa4134.
The sharing resistors are 3W 2512 smd resistors in case you wonder.
What do you think ?
Attachments
IC4a is seeing 10k||1k||1k as it's load.
That is very low for exemplary performance from any 134 opamp.
You need a high current buffer at that location or at least double the impedance being fed, from 476r to >900r.
Designing for 2k loading would be my recommendation. But that makes R7 & 8 =5k1. Risk of too much noise.
That is very low for exemplary performance from any 134 opamp.
You need a high current buffer at that location or at least double the impedance being fed, from 476r to >900r.
Designing for 2k loading would be my recommendation. But that makes R7 & 8 =5k1. Risk of too much noise.
Last edited:
I decided to drop the project in favor of the class-D Liteamp.
Here are the eagle files if it can help someone.
It hasn't been tested.
Here are the eagle files if it can help someone.
It hasn't been tested.
In case anybody else decides to pick it up:
I suggest flooding the ground plane. You can leave the input section on its own little island. The rest should be on a plane. At the absolute very least fill the trench in the ground plane next to the supply caps.
You'll need more decoupling capacitance by the LM3886. National recommends 20 uF + 100 nF by the chip. 470 uF where the power enters the board. With two LM3886es, you'll probably want 1000 uF. If you're using ceramic decoupling caps, beware of the voltage coefficient. You can see my thoughts on the LM3886 decoupling here: Taming the LM3886 Chip Amplifier
Tom
I suggest flooding the ground plane. You can leave the input section on its own little island. The rest should be on a plane. At the absolute very least fill the trench in the ground plane next to the supply caps.
You'll need more decoupling capacitance by the LM3886. National recommends 20 uF + 100 nF by the chip. 470 uF where the power enters the board. With two LM3886es, you'll probably want 1000 uF. If you're using ceramic decoupling caps, beware of the voltage coefficient. You can see my thoughts on the LM3886 decoupling here: Taming the LM3886 Chip Amplifier
Tom
I had five minutes of free time, so I did some quick changes to the board, to follow Tom's suggestions.
- replaced a pair of 4.7u/50V x7r caps by some 22uF/50V low esr ones (pana FR is what I'd use).
- extended the groundplane.
- the caps bank was 3*330uF per rail; now 4* 330uF.
I didn't change the schematic but raising the gain setting resistors on the lm3886 to 1k5 and 30k might help a bit the poor opa134, as pointed out by AndrewT. Still, at 1Vrms, a 600ohms loading should be ok. There's room enough to add a small class a buffer in the feedback loop of the opamp if someone wanted to push thd down.
- replaced a pair of 4.7u/50V x7r caps by some 22uF/50V low esr ones (pana FR is what I'd use).
- extended the groundplane.
- the caps bank was 3*330uF per rail; now 4* 330uF.
I didn't change the schematic but raising the gain setting resistors on the lm3886 to 1k5 and 30k might help a bit the poor opa134, as pointed out by AndrewT. Still, at 1Vrms, a 600ohms loading should be ok. There's room enough to add a small class a buffer in the feedback loop of the opamp if someone wanted to push thd down.
Attachments
- Status
- Not open for further replies.