LME49830 PCB design

[berni8k]

I quickly drew up a test PCB for the LME49830 amp chip thats going to be used with a IRFP240 mosfet output stage. The board measures only 61x47 mm

I am just asking if this board seams ok to you and if there are any improvements to be made.

[Segran]

Forgot to include the PCB file?

[berni8k]

Hmm... i did attach it why didnt it upload.

Lets give it a another try then

[Segran]

I have only glanced through the layout.

I would have added 0,1uF decoupling caps right across the supply pins of LME49830 to Pwr Gnd. Look at this: 6 channel LME49830 power amp Post 126.

I would also move the connection of the NFB resistor right to the speaker terminal.

No VBE multiplier? No pot to set bias???!! Hmmm, I wouldn't even have tried that.

Good luck!

[berni8k]

Oh yeah the feedback resistor would work better near the output.

And about the caps to ground im a bit confused on that one. He is talking about power and signal ground. The power ground is far away from the chip so i think that would render the to ground caps a bit noneffective to the LME. But dosent the main supply for the LME come from between the positive and negative supply where i have put a big film cap right next to it.

About the bias i plan to first solder a pot on wires to set the bias and replace with a fixed resistor when i know the value.

As for what i read from the documentation BJTs need a temperature compensated bias since they like to do thermal runaway while MOSFETs don't.

[Segran]

As for what i read from the documentation BJTs need a temperature compensated bias since they like to do thermal runaway while MOSFETs don't.

This is true for lateral MOSFET's but not for IRFP's which are of the vertical type. My advice is to use VBE multiplier as decribed by the AN.

dosent the main supply for the LME come from between the positive and negative supply

True, but decoupling both neg. and pos. supplies to power ground is imperative to achieve stability. The single cap you have is not likely to be enough, and it's to far away from the chip terminals. Put a ground connection path driectly from the GND connection and place the decopuling caps as close as possible to the chip terminals.

[panson_hk]

Pin 10 is not connected to -V trace coming from the bottom.

It seems you don't leave any room for LME heat sink. As Segran suggested, you should have, say, 0.1 uF decoupling cap right next to Pin 15, 10. You connect pin 2 directly to the zener diode? You should add a resistor to set the current. You should put a small cap at input to filter RF signal. It seems that you parallel NFB E-cap Ci and Ri. They should be in series! Please check your schematic against circuits shown in data sheet / app note.

You put a cap across +V and -V. Why?

[berni8k]

The more i look at the app note the less it makes sense to me, now i noticed the PCB in the app note has no connection between power ground and signal ground and has those caps on signal ground that caused that members amp to become unstable.


Also i don't get why you would use a thermal bias compensation transistor on a MOSFET since according to the datasheet of the IRFP240 as the temperature rises its resistance rises also so reducing the bias current if we use bias compensation it would additionally reduce the bias even more. So i don't really see the point. I know BJTs will start hogging even more current as they get hot so its logical to lower the bias to counter it.

[panson_hk]

Originally posted by berni8k

Also i don't get why you would use a thermal bias compensation transistor on a MOSFET since according to the datasheet of the IRFP240 as the temperature rises its resistance rises also so reducing the bias current if we use bias compensation it would additionally reduce the bias even more. So i don't really see the point. I know BJTs will start hogging even more current as they get hot so its logical to lower the bias to counter it.

I think you are referring to the on resistance of D-S. You should look at Vgs vs temperature which is bias related (Vbe vs temp for BJT). Temperature tracking not only for avoiding thermal runaway, but also for stable bias point. Please read AN-1645.

[panson_hk]

VBE multiplier also presents a lower AC resistance compared to a simple resistor. It is expressed below.

Reference: Dennis L. Feucht, Handbook of analog circuit design