I've been doing quite a lot of testing lately on LME49810...
Anyone else found that it does not like when the power rails are not turned on simultaneously? Particularly if the -VE rail is connected before the +VE. The output can "lock" at the -VE rail voltage, even after the +VE rail is connected.
When the +VE rail is turned on before the -VE rail the output rises to about 7.5V.
There is obviously a horrible thump from the speaker when this happens. Hardly click or pop free.
Circuit under test is a standard app note implementation but using thermal-track output devices. Power rails for testing are from a benchtop lab supply, +/- 35V DC.
After multiple connections and disconnections of the power rails (in no particular order), either the LME49810 blows up (exploded with sparks even once), or the GND PCB trace from pin 3 to star GND burns out, or the traces between sink/source and output devices can burn out.
Circuit and implementation are fine. Perfect operation if powered up both rails simultaneously, or with toroidal based AC supply (where obviously both power rails rise simultaneously by default).
Hoping someone else has found similar issues?
Anyone else found that it does not like when the power rails are not turned on simultaneously? Particularly if the -VE rail is connected before the +VE. The output can "lock" at the -VE rail voltage, even after the +VE rail is connected.
When the +VE rail is turned on before the -VE rail the output rises to about 7.5V.
There is obviously a horrible thump from the speaker when this happens. Hardly click or pop free.
Circuit under test is a standard app note implementation but using thermal-track output devices. Power rails for testing are from a benchtop lab supply, +/- 35V DC.
After multiple connections and disconnections of the power rails (in no particular order), either the LME49810 blows up (exploded with sparks even once), or the GND PCB trace from pin 3 to star GND burns out, or the traces between sink/source and output devices can burn out.
Circuit and implementation are fine. Perfect operation if powered up both rails simultaneously, or with toroidal based AC supply (where obviously both power rails rise simultaneously by default).
Hoping someone else has found similar issues?
Thanks for the suggestion, I didn't think of that one
In this particular application I cannot gaurantee that they turn on and off at the same time. And I would like to avoid an extra relay and control circuitry.
Thanks for the suggestion, I didn't think of that one
In this particular application I cannot gaurantee that they turn on and off at the same time. And I would like to avoid an extra relay and control circuitry.
Sure, no problem.
In what application would they not be turning on and off simultaneously?May I suggest you to make some crude tests with resistors in the mute, clip ground pins. Perhaps the logic part of the chip ( baker clamp, mute ) is sensitive to the power supply. Because ground pin is mainly for these functions, I would put a resistor to see what happens. The mute command is also sensitive. Could you try to drive it with a proper low voltage source and redo your experiment.
Anyway good findings
JPV
Further LME49810 findings...
In mute mode, if you heat the chip up the output locks at negative rail. Stays locked even when the chip has cooled down.
LME49810 apparently has thermal protection that cuts in at 150 degrees C, but could there be a design problem?
Has anyone else noticed this?
Anyone willing to test and report back? I simply monitored the output with a multimeter (no driver or load connected), and put a hot air gun fairly close to the chip. Output locked at negative rail within 5 seconds. Be careful not to melt surrounding capacitors or other components of course!
In mute mode, if you heat the chip up the output locks at negative rail. Stays locked even when the chip has cooled down.
LME49810 apparently has thermal protection that cuts in at 150 degrees C, but could there be a design problem?
Has anyone else noticed this?
Anyone willing to test and report back? I simply monitored the output with a multimeter (no driver or load connected), and put a hot air gun fairly close to the chip. Output locked at negative rail within 5 seconds. Be careful not to melt surrounding capacitors or other components of course!
I suggest tossing TI an email or posting in the E2E forum on TI.com. The applications guys usually respond pretty fast.
If you really need the hot-plugable feature, I suggest making a input circuit with a couple of MOSFETs that will keep the power off until both supplies are present. Throw in a little RC time constant to prevent it from chattering as the connectors are plugged in.
I have not seen any issues with my LME49811+STD03-based amp when using a regular unregulated supply.
Here's a thought: Do you have any largish reservoir cap (say 1000 uF) on the supplies by the IC itself?
~Tom
If you really need the hot-plugable feature, I suggest making a input circuit with a couple of MOSFETs that will keep the power off until both supplies are present. Throw in a little RC time constant to prevent it from chattering as the connectors are plugged in.
I have not seen any issues with my LME49811+STD03-based amp when using a regular unregulated supply.
Here's a thought: Do you have any largish reservoir cap (say 1000 uF) on the supplies by the IC itself?
~Tom
Hi Tom,
I sorted my power supply rail issue a few years ago. I ended up just switching both rails at the same time with big power relays, after the connector was plugged in and stabilised.
I still believe there is a design problem with the chip though in this regard. If you blow a mains fuse, and only on one rail, then the same problem occurs. So you need to make sure you power the LME before any HV rail fuses.
This recent temperature issue is additional to my first issue. Do you have an LME49810 amp? Perhaps you could try heating the LME up and see if it locks on the negative rail?
-S
Oh, I didn't notice that this thread is rather aged...
My amp is based on an LME49811. As long as the die temperature is within the specified limits, the IC should perform as listed in the data sheet. If you are having a different experience, that definitely warrants a support call/email to TI.
When you say "locks on the negative rail" do you mean that Vout = VEE and that it does not recover once the IC has cooled down?
~Tom
My amp is based on an LME49811. As long as the die temperature is within the specified limits, the IC should perform as listed in the data sheet. If you are having a different experience, that definitely warrants a support call/email to TI.
When you say "locks on the negative rail" do you mean that Vout = VEE and that it does not recover once the IC has cooled down?
~Tom
Yes, Vout = VEE and it does not recover once the IC has cooled down.
I now have two amps on the testbench here side by side. One based on LME49810 and one using an LM4765. According to the datasheet the LM4765 thermal shutdown acts in a similar way but at a slightly higher temperature. I can heat the LM4765 up and it mutes the output at about 170 degrees C and it resumes playback as per normal when the temp cools down about 10 degrees C. The LME49810 does not do this.
So it seems LM4765 works correctly, but LME49810 does not.
I have just posted on the TI forum as well, hopefully someone there can chime in.
Tom, is your LME49811 easily accessible in your amp? Perhaps you could put a hot air gun on it and report back?
I now have two amps on the testbench here side by side. One based on LME49810 and one using an LM4765. According to the datasheet the LM4765 thermal shutdown acts in a similar way but at a slightly higher temperature. I can heat the LM4765 up and it mutes the output at about 170 degrees C and it resumes playback as per normal when the temp cools down about 10 degrees C. The LME49810 does not do this.
So it seems LM4765 works correctly, but LME49810 does not.
I have just posted on the TI forum as well, hopefully someone there can chime in.
Tom, is your LME49811 easily accessible in your amp? Perhaps you could put a hot air gun on it and report back?
Hi!
I have no experience about the temp issue.
But some findings about the start up procedure:
I have noticed that the chip likes its both rairs to be precisely on the same potential refered to ground. So if i want to make a very hi-endish amplifier from it, i usually put a active virtual ground for it to force the both rails at exactly the same potential at all times.
Also i have a voltage monitor to switch off the mute (or actually SHUT DOWN, a small but important difference) when all rail voltages have come up above a certain level (usually 25V/power rail, with a couple of volts hysteresis to guarantee a clean shut down at power off).
All the above for the chip itself. To the output stage i can pretty much feed all sorts of rubbish or cut one rail completely (like blow a fuse for example).
Your VEE rail lock at start up (if no VCC precent) may be closely related to a situation when no GND reference is available. If no GND reference, then too the chip tries its best to find it somewhere and finally locks to the VEE untill it selfdestruckts.
A nice way to protect the chip from this china syndrome is to give it a current limited supply. No need to feed it with 1000W. Feed it with 10W and limit the current.
I have no experience about the temp issue.
But some findings about the start up procedure:
I have noticed that the chip likes its both rairs to be precisely on the same potential refered to ground. So if i want to make a very hi-endish amplifier from it, i usually put a active virtual ground for it to force the both rails at exactly the same potential at all times.
Also i have a voltage monitor to switch off the mute (or actually SHUT DOWN, a small but important difference) when all rail voltages have come up above a certain level (usually 25V/power rail, with a couple of volts hysteresis to guarantee a clean shut down at power off).
All the above for the chip itself. To the output stage i can pretty much feed all sorts of rubbish or cut one rail completely (like blow a fuse for example).
Your VEE rail lock at start up (if no VCC precent) may be closely related to a situation when no GND reference is available. If no GND reference, then too the chip tries its best to find it somewhere and finally locks to the VEE untill it selfdestruckts.
A nice way to protect the chip from this china syndrome is to give it a current limited supply. No need to feed it with 1000W. Feed it with 10W and limit the current.
Hi Pal,
Thanks for chiming in.
The virtual ground is a good idea to solve the power up issue.
It's interesting that LM4702 does not seem to suffer from the same power up issues. I suspect there is a design flaw in LME49810, or they left out a part of the protection circuit that they included in 4702?
Do you happen to have any LME49810 amps accessible? I'd greatly appreciate if you could test out the temperature shutdown issue, and let me know if your amp also locks
to the negative rail? At least if someone can confirm the same results then I know that it's not an issue with some other part of my circuit.
Thanks!
Thanks for chiming in.
The virtual ground is a good idea to solve the power up issue.
It's interesting that LM4702 does not seem to suffer from the same power up issues. I suspect there is a design flaw in LME49810, or they left out a part of the protection circuit that they included in 4702?
Do you happen to have any LME49810 amps accessible? I'd greatly appreciate if you could test out the temperature shutdown issue, and let me know if your amp also locks
to the negative rail? At least if someone can confirm the same results then I know that it's not an issue with some other part of my circuit.
Thanks!
I do not have a test platform for a 810.
I have a test rig for a 811 only. Will check during the weekend if I can find a way to heat the IC without heating everything else on the pcb also...
Hmm??? A 100W soldering iron on the sink maybe?????
It is of course nice to know what exactly happens during a thermal shut down. Although that will never happen in real life if the chip only has an adequate heatsink.
I have a test rig for a 811 only. Will check during the weekend if I can find a way to heat the IC without heating everything else on the pcb also...
Hmm??? A 100W soldering iron on the sink maybe?????
It is of course nice to know what exactly happens during a thermal shut down. Although that will never happen in real life if the chip only has an adequate heatsink.
Perhaps a hot air solder blower with the little "snoot", if you have one? Take the heatsink off to make it easier to heat the chip up.
Agreed... should not normally happen in real life if the chip is heatsinked. But I recently found one chip that it DID happen on in real life and after testing it appears the thermal shutdown threshold of the IC is wrong and it shuts down at a lower temperature. I have sent it to TI for analysis.
Thanks for testing over the weekend, look forward to your results.
Agreed... should not normally happen in real life if the chip is heatsinked. But I recently found one chip that it DID happen on in real life and after testing it appears the thermal shutdown threshold of the IC is wrong and it shuts down at a lower temperature. I have sent it to TI for analysis.
Thanks for testing over the weekend, look forward to your results.
First of all: read the datasheet.
Secondly: This is a common phenomenon for chips which has plus, minus and ground
Thirdly: The chip is designed to have a normal power supply.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.