LME49810 - a new cousin for LM4702

if the renesas Laterals are biased @ >=100mA/device pair then the negative temp co ensures that the bias is stable without requiring Bias temp compensation.

I have resistor biassed @ 75mA/device pair and even that lower bias shows a reduction in bias current as ambient temperature rises.
 
Hi Andrew and the others.

I've reading all topics dealing on lme49810 since i have on and i planed to use them for HT power blocs. I ve read all datasheet and application notes on LM4702, LME49810, and LME49830 (I will attends buiding something with these i have too later with IRFP240/9240 or 9140). Beside these i have a good stock of MJL3281/1302 and MJE1530/15031 (Andrew T you known this i mentioned in brother of quasi thread). The idea was to put them together in power channel for home theater. Here begin diffulties since i could'nt found à schematique cooresponding to what i was searching.
Ma dedicated room has 15m² surface. The speakers i planned to use have following specifications :

center : 91db/3,5 ohms min impedance
front lateral : 90db/6,0 ohms min impedance
suround : 88db/ 6,5 ohms min impedance

So i don't think I will need huge power amount (50/60w @ 8ohms will be sufficiant) but confortable current disponibility. I can have to a good price 330V 2x30 trafos.

I am be able to calculated the amount of Base current will be need for fixer bias current on output stage, but i don't really understood how we could set these base current through ajusting level voltage between Mbias/Pbias. I'am able to applicate ohms law with the 2,8mA Ibias but i don't find the relation between the voltage level at M/Pbais and Ib from outputstage. I have somme difficulties to manage the schematic with 2 ou 3 (if it's requierd) output bipolar and the "buffer" taht could provide the necessary amount of current requiered to drive multiple output transistors.

I found on diyaudio to schematic that could be a begin to my project :
 

Attachments

second :
 

Attachments

  • lme48910avecdriver.jpg
    lme48910avecdriver.jpg
    60.2 KB · Views: 1,391
Re: Specs on circuit parts

[QUOTEI'd like recommendations for:
Qmult; Q1, 2, 3, 4; heat sink for the LME49810.
Assume output of 150W into an 8 ohm load.

As a learning tool: how would I determine this? [/B][/QUOTE]

Would you like to try OnSemi ThermalTrak for output BJT? http://www.onsemi.com/PowerSolutions/product.do?id=NJL4281D.

I recommend you visit http://sound.westhost.com/articles.htm to find what you are looking for. I also highly recommend Douglas Self "The Audio Power Amplifier Design Handbook" http://www.dself.dsl.pipex.com/ampins/ampins.htm.

Have fun!
 
Re: Specs on circuit parts

marthaman said:
As a "newbie" to amp construction, could Forum members recommend specific parts for the simple amp design of the LME49810 shown on the vendors website?
http://www.national.com/pf/LM/LME49810.html

I'd like recommendations for:
Qmult; Q1, 2, 3, 4; heat sink for the LME49810.
Assume output of 150W into an 8 ohm load.

As a learning tool: how would I determine this?


I have used the inexpensive TIP142/147 for the output transistors -- a good resource is to look at Mark Brasfield's article on a high quality LM4702 design: http://www.national.com/an/AN/AN-1490.pdf He used MN2488 and MP1620 -- these you can get from MCM (Newark Electronics). He also used a TIP31 for the VBE Multiplier.

The ThermalTrak transistors aren't Darlingtons.

The LM4702 was the first of the high power driver chips and it now has a couple of brothers and sisters.

The heat sink calculation is a function of the worst-case load you will be running, supply rail voltages etc. This is the case where BIB. There's a good explanation of determining the correct heat sink value on the product folder for the LM3886, page 18 -- don't assume that your load is going to be as high as 8 ohms, however. http://www.national.com/ds/LM/LM3886.pdf The LME49810, LME4702 like to be kept cool as well, but in the amplifiers I have built with them I just use some spare PCB material attached to the chip.
 
panson_hk said:
Why use Darlington in audio amp? Space limitation? Cost?

Simplicity, fewer components.

Disadvantages:
- No control over internal topology (there are usually internal resistors in the Darlington).
- Non linear gain (beta) at low currents (=higher distortion compared with discrete). The Darlington doesn't turn on until a milliamp or so is fed into the base, and then a small incremental current turns it on much harder. ie. Hfe jumps from a couple of hundred to a couple of thousand.
- Can't use the EF Type II topology, with switch-off speed up capacitor (see fig 33 at http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm#5 )

Still, a credible amp can be built from them. I have designed and constructed a successful LM4702 + TIP142/147 amp.

There's also my 100W Class A effort of many years ago ( http://www.diyaudio.com/forums/showthread.php?threadid=32285 ) which uses 4 pairs of TIP142/147.