These are my two cents ( or a little bit more)
The 49811 ( formally 4702) is the basic front end. The 49810 is the front end plus two high voltage transistors to be used as driver or pre driver.
I recommend the triple T or Leach/Locanthi topology which is simply three push pull common collectors in cascade. The two first ones beeing biased in class A with enough current to be able to drive the output transistors.
See Leach site for thorough analysis.
Advantage: very low output impedance of the drivers of the output transistors, no cut off of the drivers, very low loading on the Vas.
Perhaps some headroom lost before clipping ( the Vbe's of the driver) but this is small penalty if any.
The two output transistors of the 49810 are the first stage of the Locanthi triple topology.
I am using thermaltracks as output transistors ( 3 in //) but also the same type as driver. This could be overkill but I don't think so. The thermaltracks are fast and have constant beta in the range of currents of interest.
The advantage is twofolds:
from a mounting point of view you can mount them (output and driver ) on the same heatsink with a pressing bare ( U profile) common to all of them. Using Belleville washer allows you to have constant pressure with temp.
I know that mounting the drivers on the same heatsing as the ouput transistors is risky because thermal pumping but if the cooling is very efficient there is no risk; by doing so, you can have a very compact pcb on top of the heatsink with safe connections to the diodes ( reliability and susceptibility).
Second advantage is biasing.
The 49810 has unfortunately two internal diodes to compensate for the Vbe of the internal transistors. Therefore the Vbias to generate will be 4 x Vbe + 2x Iq Re where Iq is the bias of the output transistor and Re is the resistance in the emitter of the same transistors ( 0.22ohms or less normally) . You may then use two diodes of the drivers and two diodes of the output transistors
If you put them in serie ( first the 2 driver ones then the two output ones) and if you put in serie with the string a potentiometer you have your adjustable Vbias, no need of a Vbe multiplier because the bias pins are a current source of 2.8mA.
And now the final touch:
The objective of the diode string is to track in temperature the output Vbe's.
If each twin diodes in the string is //ed with a resistor, adjusting the resistor adjusts the current in the diodes (they are driven by a current source). Because the drift of a Vbe junction is current dependant, this is a way to adjust in the Vbias generator the drift of the diodes to be equal to the drift of the drivers and the drift of the output transistors. All this presuppose that the output transistors are sufficiently matched.
Fine tuning of potentiometer and two resistors can be done at normal and high operating temperatures while monitoring minimum distortion with a spectrum analyzer.
The Lme49811 would allow to use the more diodes of two output transistors in serie with the diode of the first driver. You must also use an external predriver pushpull to make the triple. The Vbias drop is now 6 x Vbe + 2 Iq Re.
It has been said that the Lme49811 is another beast requiring other compensation. Looking at the open loop in the spec, it is indeed different but I suspect the open loop gain of the LME49810 as documented to be a bit compensated because in looks single pole. This is for further study
Let me know if you like it and if you need more info
JPV
Thanks JPV...
I will definatly going with the 49811 and will be using 4 pairs (8 thermal tracks) of output devices for each channel for added stability even at high loads and low impedance.
My objective is to build a simple "module" that can be cloned as many times as needed times to build a multi-channel HT amp.
As it sits right now, each pair of channels will be mounted on a forced air sink and fed by a 1KVA toroid and "dampened" by 40000uF
I will definatly going with the 49811 and will be using 4 pairs (8 thermal tracks) of output devices for each channel for added stability even at high loads and low impedance.
My objective is to build a simple "module" that can be cloned as many times as needed times to build a multi-channel HT amp.
As it sits right now, each pair of channels will be mounted on a forced air sink and fed by a 1KVA toroid and "dampened" by 40000uF
I have been testing these driver chips. 49810 with ThermalTrak, 49830 with IRFP240/9240 and 49811 with Lead output. Without using Darlington devices, 49810 + ThermalTrak is attractive combination for compact size.
Panson,
Is there a minimum voltage that is specified for adequate performance. If I choose to connect this amp to a 2 ohm speaker load I would assume a lower voltage and multiple output devices would be the correct arrangement for long term success.
When using the Leach output configuration is the second stage needed or is it already a part of the LMExxx?
Tad
Is there a minimum voltage that is specified for adequate performance. If I choose to connect this amp to a 2 ohm speaker load I would assume a lower voltage and multiple output devices would be the correct arrangement for long term success.
When using the Leach output configuration is the second stage needed or is it already a part of the LMExxx?
Tad
download the LME498xx datasheets.
They specify the maximum output current.
Apply the best gain at high current output and determine the highest peak current that the amp will be able to deliver.
The 811 will require a three stage output, the 810 may work with very high gain devices in both driver and outputs, but I suspect that driving 2ohm will ask too much and you will need three stages for this chip as well.
A 2ohm severe reactance speaker could draw a peak transient current of a 0r7 resistor as your load.
Vpk =20Vpk
Ipk 2r0 = 10Apk
Ipk 2ohm <= 30Apk.
What is the gain of the output devices when passing 30A? hFE=10 or 12 or 15?
They specify the maximum output current.
Apply the best gain at high current output and determine the highest peak current that the amp will be able to deliver.
The 811 will require a three stage output, the 810 may work with very high gain devices in both driver and outputs, but I suspect that driving 2ohm will ask too much and you will need three stages for this chip as well.
A 2ohm severe reactance speaker could draw a peak transient current of a 0r7 resistor as your load.
Vpk =20Vpk
Ipk 2r0 = 10Apk
Ipk 2ohm <= 30Apk.
What is the gain of the output devices when passing 30A? hFE=10 or 12 or 15?
Tad,
I don't think there is a specific min voltage as long as we meet the LME voltage range.
Is 2 Ohm really your design parameter? Is your speaker nominal impedance 2 Ohm? Using multiple output devices is a must for high output current, to meet SOA, to minimize beta droop. As Andrew mentioned, beta droop is sever at high current (if still within SOA). If we have multiple output devices, current of each device is lowered. Beta droop of each device is lesser. Can you access to Self's Audio Power Amplifier Design Handbook? You will get valuable information for dealing heavy loads.
Run 4 or 5 pairs of output transistors and use the proper LME498XX front end to simplify the modules, but to run 2Ohm loads you will need a beefy power supply that can meet the current demand. Also for running low impedance loads i would suggest on having some extra "padding" in the power supply with more capacitance to ease the load off your transformer and reduce fluctuations in voltage.
I have a diagram for mine as of right now ( on a piece of paper still, but will have it in cad latter on tonight )...
The Module will be using 5 pairs of NJL3281/1302's driven by LME49811.
Biasing will be done through the thermal track internal diodes in a leach style configuration (Thanks again JPV for the advice )
You may want to think this one through. Extra cap in the supply does not ease the tranformer load. It increases the transformer load: charging current pulse amplitude increases, and they get narrower. Not Nice for a xformer...
jd
An excellent, very easy to use and free piece of software to explore your crazy supply ideas can be found here:
http://www.duncanamps.com/psud2/index.html
jd
http://www.duncanamps.com/psud2/index.html
jd
I use 1kVA 40+40+40+40Vac transformer feeding two PSUs each with +-45mF to drive a pair of 170W into 8ohm amplifiers. These amps will drive less than severe reactance 4ohm speakers.What i meant is to use a lil bit more than the usual capacitance for the supply side.
I will be using 60000uF and 1KVA toroid for 2 channels
Your 60mF is a lot less than usual.
No. 60.000 is a lot more than usual. It is a lot less than usual in diy amps from builders that may not have in-depth knowledge of all the issues involved.....
jd
I think it is a lot. I can't see any advantage over 20k or 30k. The ripple starts to look more and more like a sawtooth, especially with higher loads, and we all know that saw tooth waves have a very rich set of higher harmonics.
100 (or 120) Hz ripple isn't so bad for the amp, because the power supply rejection at low frequencies is almost always quite good. But at higher frequencies it is much less so these higher harmonics of the saw tooth can get to the output easier than the basic ripple.
So my point is really why pay double for the caps unless it brings you something in terms of sound quality. I don't see that.
PSUD2 is an excellent program to get a feel for it.
And, Andrew, I always try to be nice; I sometimes don't succeed but I hope that in that case you can excuse me...
jd
PSUD2 is an excellent program to get a feel for it.
And, Andrew, I always try to be nice; I sometimes don't succeed but I hope that in that case you can excuse me...
jd
Actually, i used that program before and its missing a lot of options like allowing one to set up the supply to their spec not by a set list of "options"
That's why i have uninstalled it.
Andrew I should know, I have been on the wrong side of Jan a few times too...............but then again where would we be if we did not have him to keep us on the straight and narrow.
Even the force is no match for him................. and it usually works................or should I say Jan is like the Japanese police.http://video.google.com/videoplay?docid=5490020651450496505#docid=3311044402532116301
Jam
Attachments
Last edited:
Do you have a reference for this, or is it your own observation? I never thought one could do harm by adding filtering caps. Thanks.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.