TI E2E Community
Audio Amplifiers Forum
We have a design with a LM4861 that becomes unstable when it is driven to the clipping limit. When the about 6 Hz self oscillation has started it can be stopped by disconnecting the loudspeaker. During the oscillation the amplifier output DC-voltage drops from the normal 2.5 V to about 1 V and it gets very hot.Has anybody seen a similar behavior and found out how to prevent it?
Best regardsKlaus Mølholm | Senior HW Engineer TAA8email firstname.lastname@example.org | dir +45 72 11 24 44 | mobile +45 31 64 20 44Anritsu A/S | Kay Fiskers Plads 9, 2300 Copenhagen S | Denmarkweb www.anritsu.com | tel +45 72 11 22 00 | fax +45 72 11 22 10 | Co.reg.no.: 28 31 01 61
I've seen a similar behaviour in a circuit using LM386 but the problem was a missing 0.1uF supply bypass capicitor, other problems could occur with unity gains (-1), because of the phase margin (if the phase reaches 2pi*n) meeting the Barkhausen criterion you may have oscillation, but I think this amplifier is unity gain compensated (the phase is kept far from 2pi*n). The 0.1 uF capacitor should be placed near the supply pin in the PCB layout and ground paths must be short and wide to prevent signal injection to ground and ground loops (because real conductors are not ideal so when current passes thru it the voltage is above 0), the input ground must be connected on the ground path near the IC ground.
Thank you for answering. We already have a 0.22 uF sitting right next to the amplifier, then we have 52 uF a few inches away and then we have 86 uF on the other side of a 13 mOhm FET. The LM4861 is unity gain stable, but we run it with a gain of 80, so this should not cause the oscillation either.
Do you have tried to reduce the gain to 20 like the circuit in figure 3, using the same values of the components?
You should also use a voltage regulator with the desired voltage (always between the operational limits).
You can then add a 4 V/V preamplifier like the LME49720 if you don't have specific PCB space requirements.
How do you do to control the volume of the amplifier, sometimes if you use a pot at the input oscillation could occur, so it's better to use a voltage follower (voltage buffer).
If you don't want to take this drastic measures you could try to add an audio buffer at the input, a preamplifier with vg = 1 V/V, also with LME49720 or a cheaper multipurpose opamp like TL082, I've solved one of my oscilation problems, driving an ST amp with a TI opamp TL082.
A sugestion of a very versatile regulator if you already don't have one is the LM338T or LM338K, with adjustable output voltage and a large output current, I've one and it's one of the best linear regulators I've used.
Can you post the schematic of your circuit?
I hope that this message helped you,
The best luck for your project,
Is the fet fully turned on? Is your supply capable of providing enough power?
Can you capture a scope shot of the voltage at the VDD pin to make sure it's stable?
Audio Applications Engineering Manager
Dallas, TX USA
We already use a IR3473 voltage regulator.
We don't need a preamplifier, the current gain is set to high.
The volume is controlled digitally in the SGTL5000 codec that is driving the LM4861.
You can se the schematic here: 0820.LM4861.PDF
Yes the FET is turned fully on, it is turned on as a part of the power on sequence. The supply can provide 6A.
The voltage at the VDD pin is normally flat, during oscillation a small variation (about 20mV) coinciding with the oscillation can be seen.
OK, sorry, we always have to rule out the most obvious things first...
What about the freq of oscillation? Can you post some scope shots?
The oscillation frequency is about 6 Hz. When seen on the scope it looks like the amplifier oscillates between shutdown mode and active mode because the DC-voltage on both outputs drops from the normal 2.5V to about 1V. I am sorry but I am not able to recreate the error at the moment, so I can't post scope shots of the oscillation.
Thank you very much for sending your schematic, it seems a very interesting project, the amp is usb powered?
In this post I've wrote some sugestions for your design:
You should try to monitor the voltage on the shutdown pin during the oscillations, the oscillation could be caused by signal injection on ground plane influencing the shutdown circuits, the ground plane should be wide and short and have a large aluminium electrolytic capacitor to lower the impedance of this tracks during high loading conditions.
Other question you are using the amplifier as an inverting adder amplifier?
Atention for this circuit the expressions are: Vo1 = -Rf*((Vin1/Rin1)+(Vin2/Rin2)+(Vin3/Rin3)+(Vin4/Rin4)) + Vos [V]
Vo2 = Rf*((Vin1/Rin1)+(Vin2/Rin2)+(Vin3/Rin3)+(Vin4/Rin4)) + Vos [V]
Vo1 and Vo2 in respect to common (gnd), Vo differential
You should connect Vo2 at plus terminal of the speaker, and Vo1 at the minus speaker terminal, to avoid phase shift 180º or pi at the output sound.
Vo = Vo1 - Vo2 = -2*Rf*((Vin1/Rin1)+(Vin2/Rin2)+(Vin3/Rin3)+(Vin4/Rin4))
Vos aproximately Vcc/2, to maximise the output excursion.
This expression are only valid for AC signals.
The expression has the 2 times, because the amplifier is differential.
Vin1 = BEEP_CPU
Vin2 = BEEP_FPGA
Vin3 = LINE_OUT_L
Vin4 = LINE_OUT_R
Rin[1...4] - input resistors for the above signals.
I think that Vin3 and Vin4 should have a different capacitor and not the same capacitor, and the resistors should be added after the capacitor (figure attached), but this aspects are only sugestions.
fL = 1/(2*pi*Cin*Rin), for each input
Notes: All these expressions are ideal, in real devices the expressions can vary. The ground of the signal generating devices must be connected to the same common (gnd) as the amplifier.
Thank you very much for your atention,
The amp is powered by a IR3473 6A switch mode power supply.
At the moment we are not able to repeat the error, but if we see it again i will check the shutdown pin. The ground plane consist of 4 layers of copper . According to the data sheet the shutdown pin is a logic input and should therefore be fairly insensitive to noise on the pin.
Yes I am using the amplifier as an inverting amplifier with 4 signal sources.
Thank you for your comments.
Personally I think that switched mode power supplies are not very good to use with audio applications, but if there isn't another option.
There are no other options.
Nearly every single customer application I work on uses an SMPS for power, as you stated, realistically, there are no other options for a mass produced product, especially if you want to sell it world-wide...
6 Hz... That almost sounds like maybe it was the thermal protection kicking in.
If you run across it again, please try to capture some scope shots. I'd like to see the power supply (at the pin of the LM4861), the outputs (OUT1-OUT2 differentially), and the input signal.
I also think that it is the thermal protection circuit that somehow self-oscillates.
I promise that if we see it again, we will get the documentation while we can.
OK, great! I'm going to close out this thread, then.
All content and materials on this site are provided "as is". TI and its respective suppliers and providers of content make no representations about the suitability of these materials for any purpose and disclaim all warranties and conditions with regard to these materials, including but not limited to all implied warranties and conditions of merchantability, fitness for a particular purpose, title and non-infringement of any third party intellectual property right. TI and its respective suppliers and providers of content make no representations about the suitability of these materials for any purpose and disclaim all warranties and conditions with respect to these materials. No license, either express or implied, by estoppel or otherwise, is granted by TI. Use of the information on this site may require a license from a third party, or a license from TI.
TI is a global semiconductor design and manufacturing company. Innovate with 100,000+ analog ICs andembedded processors, along with software, tools and the industry’s largest sales/support staff.