Lm1875 - need some help..

[bamera]

Hi, I´d just like to start with a big thank you for this wonderful forum. Four months ago I didn´t know what a capacitor was, today I´ve finished my first power amp project.   

My problem is simple I hope.

I´m using a +/- 18V 3A transformer with a full wave rectifier bridge (1N5408 diodes) and 10000uf smoothing capacitors as my power supply. I tested it with the TDA2050 PDF split supply application circuit through an 8ohm 200W speaker and it sounded great, no problems at all.


I decided to try the Lm1875 split supply circiut from the PDF. I put it together exactly like the schematic and tested it, it worked perfectly (excellent quality sound) for about 10 seconds and shut down.
I tried again and the same thing happened.

The problem is that my heatsink is quite large and gets barely warm so I can´t see this as temperature issue. I used arctic silver thermal paste for further testing and still the same problem. I had the same exact problem with the LM3886, after two circuit builds of that one, I gave up.

Any tips on what the problem might be besides the heatsinking.....   

Thanks, Daniel C

[PerroGrande]

Is it correct to assume that your transformer is a 36V Center Tapped transformer?

Have you measured the output from your power supply when it is loaded down by the components?   I'm just trying to eliminate any possible sagging -- especially since the TDA2050 will work with a lower minimum voltage (+/- 4.5v) than will the LM1875.   So one thing to look at is the power supply's behavior.  It is the one item in common with your LM1875 and 3886 builds.

[km-r]

i didnt have problems with my LM1875 with split 15V DC...

[bamera]

Is it correct to assume that your transformer is a 36V Center Tapped transformer?


-Yes 36V center tapped


Have you measured the output from your power supply when it is loaded down by the components?   I'm just trying to eliminate any possible sagging -- especially since the TDA2050 will work with a lower minimum voltage (+/- 4.5v) than will the LM1875.   So one thing to look at is the power supply's behavior.  It is the one item in common with your LM1875 and 3886 builds.

-No sagging at all... 56V... it shuts off at low volumes too for some reason.. tested it this morning with the heatsink nice and cold..... shutdown after about 10 seconds.. heatsink was still cold.

[Joecool85]

I'm not sure exactly, but it sounds like you might want to check out ssguitar.com, it's a forum that I built that only focuses on solid state guitar amps.  We have a handful of ss gurus over there, and a bunch of amateurs like myself.  It'd be worth a post to see if someone there can help.  Good luck!

[PerroGrande]

Good -- we've eliminated the power supply as an issue.

Both chips have shutdown for short-circuit conditions and thermal conditions.  I find it a little unlikely that thermal is the issue -- especially with a cold heat sink and only 10 seconds of running.

In the 1875 application notes, it did make a special point to stress certain design criteria to avoid oscillation and other issues.  Have you taken these into account on your setup?

[R.G.]

On all of the power amp chips, the heat sink tab is common with the V- supply pin. If you attach these to a bipolar supply and also to a grounded heatsink, you must electrically insulate the power amp chip from the heat sink. Heat sink goo, especially the silver-powder-bearing stuff is not an insulator. If you fail to insulate the chip from the heat sink, or the heatsink from ground, you wind up shorting the power supply. What happens then depends on the details of how you built your power supply and amp wiring.

[Mark Hammer]

This would be where mica insulators come into the picture?

[PerroGrande]

Correct.  They provide electrical insulation between the tab and whatever the heatsink might be connected to.  So, for example, if your heatsink is attached to the chassis which is grounded, you'll want to isolate the tab using one of these insulating strips.  Alternatively, isolate the heat sink. 

[Mark Hammer]

Just out of curiosity, assuming this interpretation (a mistakenly grounded heatsink tab) is correct, what is accounting for the 10 seconds of functionality?  Is that a result of the size/value of the power supply caps?

[PerroGrande]

Yeah Mark, the 10 seconds of functionality is quite puzzling.   I've had things work for a few seconds and then fail, but they don't usually work again until the real problem is fixed. 

Shorted power supplies usually let you know that they're shorted...  Fuses, sparks, toasted stuff, smoke...even flames... 

From what I gather of the power supply in this case, regulators aren't in the equation, so that eliminates any possible short-circuit or heat-related "graceful" shutdown on their end.  Plus, the problem happens at low volume levels... kinda the opposite one might expect.

I'm still very curious about the shutdown mechanisms within the 1875 and what they would do in the case of oscillation at frequencies above the audible level.  If the amp was oscillating at a very high frequency, could that trip one of the shutdown mechanisms?

[PerroGrande]

Daniel -- just to clarify:

With the 1875, you power up the amplifier and play it, even at low volume levels, and get only 10 seconds of sound.  The sound then stops. 

If you shut the unit off and immediately turn it on again, it exhibits the same behavior.

During the time when the amp is playing *and* when the sound has shut down, if you measure the power supply voltage(s), they are at the expected level.

There is no significant heating of the amp, surrounding components, or the power supply.

Additional questions:

Does leaving the amp on, even after the sound has shut off, produce any significant heating from any component or connection?

If you leave the amp on for some longer period of time, does it try to come on for another 10 seconds and shut down?

Is the tab electrically isolated from the heat sink, or is the heat sink electrically isolated from ground?

What impedance is your speaker?

[bamera]

Wow, lots of responses...

Joecool85 : Since I had a similar issue with my 3886 project I asked this same question at the dyiaudio forums, only one answer which addressed oscillations possibly causing thermal shutdown - problem is there was no heat!!! I´m registered at ssguitar.com and will ask there too.


PerroGrande : Thanks for taking an interest  To address the heatsink grounding - I tested the heatsink connected to the chassis and isolated from the chassis, same results for both cases. The circuit itself is star grounded on a corner of the PCB, isolated from the chassis. This is the same method I used for my other circuits and all have worked fine with no noise issues. The only connection to the chassis is the AC Earth ground. I didn´t notice any difference in sound whether the heatsink was isolated or not. The heatsink does not make any contact with any of the circuit components other than the 1875.


Leaving the amp on does not cause any heating either.

I wonder if this has anything to do with the giant capacitors I used for the PSU. 2 x 10000uf. The amp takes a while to discharge the large caps and will only start again once they are discharged  (I use the on/off LED on the secondaries to montior this).  have you ever heard of this phenomenon?

Thanks for all the kind responses, Daniel C

[bamera]

...sorry I forgot to answer your question regarding my speakers.

I tested on a 12" 8ohm 200W selenium and an 8" 4ohm bravox guitar speaker (don´t know the wattage but it can get very very loud)......  My shutdown issue was identical on both but the darn thing sounds wonderful for a few seconds until it does!!!

Daniel C

[R.G.]

OK, good going with the thermal tab not grounded. That trips up a lot of people.

If it takes 10 seconds to quit and you are not getting heating, that makes capacitor charging a likely suspect. If this were the LM3886, I'd say to look at the mute pin connection, but the 1875 doesn't have one.

Hook up your DC meter across your loudspeaker, set to a voltage range of 20V or more. Turn on the amp and see if there is DC across the speaker, and more importantly, does it slowly increase across 10 seconds.

[PerroGrande]

Well, if you can just learn to enjoy music in 10 second intervals...   

Yeah, R.G. -- I went to the data sheet hoping for something like a mute pin...  Oh well.

One other question:

How much gain do you have in your circuit?  (From the data sheet, it looks like this is set using a simple voltage divider in the feedback loop back to the inverting input).

[R.G.]

Which does bring up an interesting question - Daniel ... exactly... what schematic are you using? "from the PDF" is fairly non-specific.

Put another way, let's go do the same old same old - what to do when it doesn't work. Nail down the exact schematic, the exact pin voltages, and in this case the pin voltages with time.

[bamera]

Hook up your DC meter across your loudspeaker, set to a voltage range of 20V or more. Turn on the amp and see if there is DC across the speaker, and more importantly, does it slowly increase across 10 seconds.

Thanks R.G I will try this tomorrow (all this fussing around broke a pin off the 1875, gotta buy a couple more)

Well, if you can just learn to enjoy music in 10 second intervals...  

Yeah, i can use the amp to play one liners in HiFi

Well, if you can just learn to enjoy music in 10 second intervals...  

Yeah, R.G. -- I went to the data sheet hoping for something like a mute pin...  Oh well.

One other question:

How much gain do you have in your circuit?  (From the data sheet, it looks like this is set using a simple voltage divider in the feedback loop back to the inverting input).

I used the exact same components as the data sheet,  part by part my project is identical to the PDF so it should have the same gain. I´m not sure what that is though

[bamera]

Which does bring up an interesting question - Daniel ... exactly... what schematic are you using? "from the PDF" is fairly non-specific.

Put another way, let's go do the same old same old - what to do when it doesn't work. Nail down the exact schematic, the exact pin voltages, and in this case the pin voltages with time.

R.G I appoligize, you posted as i was writing my last reply.

The shematic I used was the one labeled "Typical Applications" on page 1 of the PDF. It´s the split power supply schematic.

Tomorrow when I get a replacement for my busted chip (previous post) I will post all the pin voltages (with time). 

Thanks for giving me a direction into solving this

[PerroGrande]

I was curious if you had accidentally or intentionally reduced the gain to a point below where the unit is stable.  I think both schematics have a voltage gain of 21, so you're okay there...