My Hi-End/Current LME498xx Amplifier

[panson_hk]

This is my latest power board design for LME498xx plug-in amplifier. The board is designed for four-par power device. They can be ThermalTrak or 3-lead TO-264. Pre-driver is included for output triple configuration.

The driver stage can be TO-220 transistor or ThermalTrak. The later is chosen for tight thermal tracking. The schematic is shown here for reference.

[panson_hk]

Before I execute systematic test. I briefly test the first assembled unit. No pre-driver is employed in this unit. Four NJL3281D/1302D pairs are used. The driver is MJE15030/15031. LME49811 is assembled in the driver board.

Four TT diodes together with a 500R trim pot are used to generate bias. The power supply is +/- 40 V no load (330VA transformer). The initial result is very promising.

THD vs power for 1kHz and 20 kHz for 8.3, 4.1 and 2.8 Ohms are illustrated in the figure.

[Johno]

Great stuff Panson. i thought everyone had forgotten about these little drivers.

I am doing a similar 4 diode bias, on bread board so far, but mixing 2 thermal track and 3 standard parts in the same output array (5 pairs). Saves on realestate and seems to work but not sure if I will need to do any matching. Look forward more of your magic measurements.

[panson_hk]

Originally Posted by Johno

Great stuff Panson. i thought everyone had forgotten about these little drivers.

I am doing a similar 4 diode bias, on bread board so far, but mixing 2 thermal track and 3 standard parts in the same output array (5 pairs). Saves on realestate and seems to work but not sure if I will need to do any matching. Look forward more of your magic measurements.

Thank you!

I also wonder why they are not frequently discussed in the forum. They really deserve much more attention. Probably beginners prefer single chip like LM3886 and experienced diyers prefer fully discrete.

A very high performance power amplifier in fully discrete is too much work to me. With the LME driver chips, I can focus on the output stage, bias, thermal, etc.

Which standard pairs do you use? I think MJW3281/1302 and NJL3281/1302 should work seamlessly.

[jackinnj]

Nice work Panson

[Johno]

Originally Posted by panson_hk

Which standard pairs do you use? I think MJW3281/1302 and NJL3281/1302 should work seamlessly.

Hi Panson,
The MJW have the TO247 case. The NJL and MJL both have TO264. I understand the die for them are the same for all the different cases but....

Also, I have gone for the higher voltage spec 4281/4302 for both the 3 pin and 5 pin transistors although in hindsight it probably was not the wisest choice. NPN and PNP gain does not match/track so well although thermal resistance said to be lower.

The other thing is I think I need to go to a 5 diode bias string with the triple darlington, feel it is slightly under compensated - but am I correct thinking you are happy with just 4 diodes? (sorry, I know the bias to and fro has been interminable)

John

[panson_hk]

Originally Posted by Johno

The other thing is I think I need to go to a 5 diode bias string with the triple darlington, feel it is slightly under compensated - but am I correct thinking you are happy with just 4 diodes? (sorry, I know the bias to and fro has been interminable)

John

Hi John,

The bias tracking by four diodes is pretty good in my test unit. It actually provides very slightly overcompensation.

Panson

[klewis]

Hi Panson,

I've been using 15034/15035 for the drivers. The HFE measurements are closer for the N channel vs P channel. Like you I plug in the front end to the current amplification section of the layout. This way I have been able to try lots of variations of different front ends and current amp designs. I went down the regulated front end route with a Salas V1.2 shunt, it helped a bit, but, not much. My current front end is a descrete design made by Paul bySouth. It has lower noise and THD at low power levels than the LME49811. Not to knock the LME, it works quite well.

[panson_hk]

Here are measured data for thermal tracking. One of the Re voltage drop was measured together with main heat sink, driver heat sink, TT case and LME heat sink temperatures. The ambient temperature was about 25 C.

The test unit used four-pair TT where four of the diodes were used for biase/thermal tracking.

The test unit was in idle mode for an hour. The main heat sink/case temperature moved from 30 C to 45.4 C/57.7 C. The bias tracked the temperature variation pretty well as shown by the Re drop plot.

The test unit then delivered 47 W into 2.8 Ohms for about 40 min. When the signal was removed, the bias was lower than the target point 24 mA/Re (Re = 0.22 | 0.22). It then gradually moved up. Hence, it is a slightly overcompensated tracking system (good to have).

[panson_hk]

Originally Posted by Johno

The other thing is I think I need to go to a 5 diode bias string with the triple darlington, feel it is slightly under compensated - but am I correct thinking you are happy with just 4 diodes? (sorry, I know the bias to and fro has been interminable)

John

Hi John,

Are you using output double or triple? Five diodes may be too high for double unless you want a heavily biased Class AB.

Panson