Eliminate turn-off thomp on LM3886 - How?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi,

Having browsed through the forum I am yet to find an easy way to eliminate thomp or pop at power off of my LM3886 amp. I am using the standard stereo kit from chipamp.com, and when I turn on the amp there's nothing, but when i cut power there's a (huge) thomp in both speakers. Looking at the bass/midrange speaker I feel like i never want to turn the amp off, the coil is moving quite a distance on turn off.

On a side note: My current setup is using a 2x24VAC 120VA transformer, as I am waiting for the correct one, but I cannot see that it would make any difference when I receive the correct transformer (220 VA)

Any ideas how to move forward?

Cheers,
Fredrik
 
Last edited:
I built a delay ON relay. so the relay will connect the speaker after few seconds. also the speaker will be disconnected immediately when the power switch is OFF. thus eliminating the pop sound. email me if you want the circuit. I have to snap the picture before sending it to you.
 
the 3886 has "mute" pin.
Have you used it?
The "mute" can be instant or it can be delayed.
The 3886 also has low voltage cutout. Are your voltages getting down to that level before the mute has silenced the chip?

Have you read the datasheet?
 
Last edited:
Thank you all for your replies. I'm not an electrical engineer, so the technical discussion may be a bit over my head, so be gentle :rolleyes:

I've got a 100 uF connected to pin 8 of the LM3886, so yes the mute function should work, and it looks like it does as the amp is dead silence on start-up.

AndrewT: As I am currently using a 2x24VAC 120 VA transformer (I'm waiting for the correct one), could this cause a low voltage cut out -> producing thomp?

LAJ: I do not really follow you; By collapsing together do you mean draining the PSU caps " at the same time" or? What would I typically use as bleeder resistor for the PSU caps, and how are they connected?

Thanks,
Fredrik
 
I've got a 100 uF connected to pin 8 of the LM3886, so yes the mute function should work, and it looks like it does as the amp is dead silence on start-up.

AndrewT: As I am currently using a 2x24VAC 120 VA transformer (I'm waiting for the correct one), could this cause a low voltage cut out -> producing thomp?
the 24Vac is not the cause of the OFF thump.

What resistor are you using with the 100uF on pin8? look at the schematic on page 1 of the datasheet. are your cap and resistor placed like that and connected to -ve supply?
look at the mute graph on p16. check your current against that graph.
You may be holding the mute ON for too long by using too much current and then the cap extends the time it takes before it allows the mute to go OFF.
 
Apologies for waking up an old thread, but I have a peculiar problem .
I built a Lm3886 based on Bob cordell's super gain clone but without the DC servo .
There is around a 4v turn on DC spike and and -4V turn off spike, I have narrowed it down to the buffer circuit causing this when the rails collapse. Even though the 3886 mute timing is correct the buffer has a direct connection to output via R7 which probably causes an offset when the rails collapse and LM3886 being in mute mode has no control on this.I think the -ve rail collapsing faster than the +ve rail might be causing this.
Cordell's design with the servo might not have this as the servo would be counter acting any DC but I still doubt as to how effective the servo would be when the power supply is collapsing.
Would bleeders on the main filter/reservoir caps help ?

Any thoughts on this would be helpful.

Cordell-Schematic.jpg
 
Last edited:
1) are you measuring those spikes with the speaker connected?
If so, there is nothing the inverter can do to directly cause such a spike on the output.
Hint: 20k in series with 8 ohms.

2) If you are reproducing a published circuit, in principle copy it in full or not at all.

3) what´s wrong with the datasheet example ? :confused:

I bet it works properly and sounds very good :)
 
1) are you measuring those spikes with the speaker connected?
If so, there is nothing the inverter can do to directly cause such a spike on the output.
Hint: 20k in series with 8 ohms.

2) If you are reproducing a published circuit, in principle copy it in full or not at all.

3) what´s wrong with the datasheet example ? :confused:

I bet it works properly and sounds very good :)

I measured it without the speaker and with the speaker I hear a distinct pop.
The diagram I reproduced is actually the full diagram in Cordell's book.

The datasheet is fine but I wanted to use the inverting configuration and try out the technique for removing the large cap from the feedback path in the normal non inverting configuration.
Except for the startup thump this configuration sounds very very good, no complaints on sound quality.

I have got a little hint though why it might be occurring, its on page2, http://www.analog.com/media/en/technical-documentation/application-notes/AN-202.pdf.
 
Moderator
Joined 2002
Paid Member
I don't see an option except to use muting relays on amp output.

The mute control of the 3886 is not turned on until the current through the pin drops below 1ma. The low voltage turn off threshold is 12V, so it's possible that the mute control is not activating (depends on your selection of Rmute) before the amp has reached that voltage.

In the meanwhile, the inverter power supply might be collapsing faster and lose control over its output. I see similar issues with active crossovers and buffers, and the only two options are to mute the speaker output or amplifier input, or use power timing control to ensure the small signal stages always get power, at least until the 3886 shuts down.

You can monitor shut down status through the current in the mute resistor.
 
Apologies for waking up an old thread, but I have a peculiar problem .
I built a Lm3886 based on Bob cordell's super gain clone but without the DC servo .
There is around a 4v turn on DC spike and and -4V turn off spike, I have narrowed it down to the buffer circuit causing this when the rails collapse. Even though the 3886 mute timing is correct the buffer has a direct connection to output via R7 which probably causes an offset when the rails collapse and LM3886 being in mute mode has no control on this.I think the -ve rail collapsing faster than the +ve rail might be causing this.
Cordell's design with the servo might not have this as the servo would be counter acting any DC but I still doubt as to how effective the servo would be when the power supply is collapsing.
Would bleeders on the main filter/reservoir caps help ?

Any thoughts on this would be helpful.

View attachment 571882

Are there any single supply line level circuits before the amp, or is the amp single supply? These will often use an AC coupling cap (a first order RC HP filter). The output of the filter will have a step response consisting of a large pulse at turn on and turn off. This is when the halfway biased voltage steps up from 0V (when off) to half of supply voltage, and then drops down again to 0V when the power is turned off. The step response pulse can be quite large. You will need a relay to isolate the input (or the amp's output) until the pulse has subsided.

The same pulse can occur if there is large DC offset coming from upstream circuits to the amp. Check for DC at the amp input during operation using a multimeter.

Keep in mind that AC coupled equipment has start-up and shut down behavior that is different than the steady-state behavior.
 
Neurochrome.com
Joined 2009
Paid Member
Make sure you + and - rails are collapsing together. You may need some bleeder resistors across your filter caps to accomplish this.

+1

Also make sure that all loads, other than the amp, are on the negative rail. This includes the various sundry items like power-on LEDs and such.
The MUTE circuit in the LM3886 is driven by the negative rail. If the negative rail collapses first, you have no thump. If the positive rail collapses first, you get a thump. So make sure the rails either go together or that the negative rail is guaranteed to collapse first.

Tom
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.