6 channel LME49830 power amp

[Zen Mod]

tnx for lbr file , cuibono ;

it's functional

[AndrewT]

the 201 & 1530 have a very low Vgs compared to other Vfets.
I think you need to change the Vbe resistors to suit the IRFs.

[cuibono]

Thanks - I'm kind of in the dark here, so any help is much appreciated.

My impression was that those resistors also effect the temp. coefficient of the tracking transistor. I'll start playing with it, but if anyone knows better,

[cuibono]

Pages 8 and 9 of the app note talk about setting Vbias, and in particular, show an equation relating Vbias, Vbe (of the tracking transtor), and the 4 resistors (circuit shown below). My current guess is to change Rb2 - - Rb3 effects the slope of the transistor's thermal response, and lowering Rb1 might cause the bias to be too high (I'm guessing). So that leaves Rb2 - I'll be increasing it from 750ohms to, I don't know, 7.5k? Supposedly, the max Vbias of the LME chip is 16V, so I suppose this will let me max it out. I just hope I don't mess up the thermal tracking...

This is the equation relating the different values:
VBIAS = (RB3 * 2mA) + VBE [1 + RB2 / (RB1 + RP1)]

[SpittinLLama]

Rb3 sets the temperature independent portion of the bias voltage while the rest set up a temperature dependent portion (based on the change in Vbe of Qvbe as you can easily see from the equation). I cannot remember what values I used for the design with IR FETs but I think I only varied the values of Rb2 and Rb3. Going to 7.5k for Rb2 will be WAY too high. I'll try to get my schematic for the IR FETs so I can tell you good starting values. I typically set the bias the same for the IR FETs as the Toshibas. I actually made 6 boards, one for each of 3 different FETs (Toshiba, IR and Renesas) then the 3 again but with an additional driver stage. I only made the note on the Toshibas because they seem to be the best and sound good too. You might also want to look up AN-1645. The LM4720 has a bias limit of about 7V but there might be some info in there for you to learn more.

-TH

[cuibono]

Thanks! I knew I shouldn't have been trying to read that stuff when tired and hungry - I kept thinking Rb3 was the temp. dependent resistor, but the equation made no sense. Now it makes way more sense..

So, I think I'll boost Rb3 some, maybe 2k, which will give about 4V, then I'll need a wider range of voltages (each module is going to have different bias levels), so use Rb2 = 1200, Rb1 = 100 and Rp1 = 1000. This should give an range of 1 to 12 volts, for a total effective range of 5 to 15 volts. (all this assumes Vbe around .5V) Sounds decent?

I'm still not sure how gate to gate voltage translates to Vgs. The LME's bias voltage goes up to 16V, so I could get up to 16V from gate to source, and double that gate to gate?

[SpittinLLama]

Looks like I used 2.4k for Rb3, 1.5K for Rb2, and 205 for Rb1. I don't remember if I spent the time to get a nice flat or slightly negative bias current over temperature but I remember the IR FETs had a 'funny' bias response. It seems the current would shift some at lower temp then stabilize out at higher temps. I think I used 22 ohms for the gate resistors, but again, not sure if I put the time in to get the design fully optimized. The compensation capacitor was 20pF. I typically tested with 4 and 8 ohm resistive loads to see how low I could push the comp. cap until I couldn't get full stability then I'd bump it up some for safety. 20pF is a pretty good number to start with. If you get too much instability then adjust it to something higher like 30pF and work with the gate resistors to get it stable. Then back the comp cap back down and continue working with the gate resistors until you get full stability across all the output power ranges (including hard clipping, say 10% THD) and the rise/fall slopes are smooth with a square wave. This is if you have the time, equipment, etc to do this.

Gate to Gate voltage is just that, measure from Source to Sink on the LME part which is the Gate to Gate voltage. This is basically the total bias voltage but not just 2xVgs because of the 0.1ohm source resistors. It is the easiest way to measure the bias voltage. This is what J9 on the reference board is for.

-TH

[Zen Mod]

pretty good work here in last pages

even if I have samples few months already , I didn't have enough time to study documentation ;

cuibono - is it too much - if I ask you to share just your sch file ?

TIA

( and - I'll stop asking there )

[cuibono]

Hello Zen,

Of course, please enjoy the files - I've included the schematic and board.

SpittinLLama,

Thanks for the good info. I'll let you know how it goes. I had bad instability with an 11 ohm load, but something wasn't working. Hope it works better. (PS, I think you meant to reference the LM4702)

[Zen Mod]

Originally posted by cuibono
Hello Zen,

Of course, please enjoy the files - I've included the schematic and board.

SpittinLLama,

Thanks for the good info. I'll let you know how it goes. I had bad instability with an 11 ohm load, but something wasn't working. Hope it works better..

respect !