Amplifier IC issue -- resurrecting an old project

Whilst cleaning off the workbench and surrounding area, I came across an old project that got shelved.  It was an attempt to build a little amplifier using an NTE1293 5.8W audio power amp IC as the power amp stage.  (Usually these things are used in car stereos, or so I've read...)

Anyway, the NTE1293 is identical to the UPC1185 amp chip.  I built the amp stage based on a schematic I found for the UPC chip.  (Attached...)

The problem I ran into back then, and still have now, is that connecting a speaker produces a very loud howling sound that has no bearing on what's connected to the input.  With nothing connected, there's no howl.  This time, I decided to do some actual troubleshooting.  Here are my findings so far....

I disconnected the 15VDC power supply that was built for it and connected a 15VDC bench supply to eliminate the possibility of a supply issue.  Same result.

Leaving the speaker aside, before I kill it, I looked at the output on a scope while feeding the input a -26dBV sine wave, I see a perfect square wave that, according to the scope settings, measures about 15V.  -26dB doesn't seem like it should be sufficient to overdrive a power amplifier.  Looking at the input, I see a nice clean sine wave of much smaller amplitude, as one would expect, so I know I'm feeding it a clean signal.  I measured the output (of one channel) with a voltmeter.  I measure 7VDC(!).  Switching the meter to AC, I measure around 7VAC.  Is the DC just the meter's interpretation of a square wave?

I'm inclined to think that the chip is shot, unless there's something else I should check.

It might be wiring/layout if the current from the speaker induces positive feedback.

Does the chip data sheet provide a proforma layout?

JR

JohnRoberts said:

It might be wiring/layout if the current from the speaker induces positive feedback.

Does the chip data sheet provide a proforma layout?

JR

Click to expand...

The only documents I could find were that schematic (and a few others that were almost identical) and the spec sheet from NTE.
http://www.nteinc.com/specs/1200to1299/pdf/nte1293.pdf

Funny thing...  When I did a search for UPC1185, I found a dealer with a link to a data sheet.  Clicking on the link brought up the data sheet for the NTE1293!

Looking a little more carefully, I noticed one particular deviation from the schematic I posted.  The resistors in series with the 100nF caps on the outputs, the ones that are supposed to be 1 Ohm (why bother??) -- I guess at the time I didn't have any that small, and I installed 12 Ohm instead.

How much gain does it have? Usually there's quite  a bit of gain. So you have a high gain power amp in which case of course it's gonna howl.

The most important thing is probably ground layout. You want the ground end of the speakers to connect together at some point where all of the other stuff that is connected to ground converges at. In other words, if you have a long ground wire with the supply ground at one and and then a bunch of stuff connected to ground and then, at the other end you have the speaker grounds, that would be bad. A simple fix might be to just connect the supply ground right next to where the speaker grounds are and then all of the other grounded components are connected to a ground that branches off from there.

As for the 7V, that is perfectly normal since this looks like it's single supply. So internally everything inside the caps is going to be biased to Vcc/2.

Depending on what you want to do, a little negative feedback would go a long way. If you're doing a guitar amp, I would leave it maxed out like it is. But if you want a little more fidelity at the expense of power, you could add a 1k in series with the caps on pins 5 and 8 for more feedback. That might help stabilize it as well.

Another thing you can do is make the caps on 5 and 8 smaller. That will reduce the gain at low frequencies. You'll loose some bass though.

CurtZHP said:

The only documents I could find were that schematic (and a few others that were almost identical) and the spec sheet from NTE.
http://www.nteinc.com/specs/1200to1299/pdf/nte1293.pdf

Funny thing...  When I did a search for UPC1185, I found a dealer with a link to a data sheet.  Clicking on the link brought up the data sheet for the NTE1293!

Looking a little more carefully, I noticed one particular deviation from the schematic I posted.  The resistors in series with the 100nF caps on the outputs, the ones that are supposed to be 1 Ohm (why bother??) -- I guess at the time I didn't have any that small, and I installed 12 Ohm instead.

Click to expand...

The 1 ohm R is part of an RC pole that helps stabilize the amplifier... the 12 ohm where a 1 ohm should be is the real "why bother". Try again with correct values. 

Good Luck

JR

CurtZHP said:

.  I measured the output (of one channel) with a voltmeter.  I measure 7VDC(!).  Switching the meter to AC, I measure around 7VAC.  Is the DC just the meter's interpretation of a square wave?

Click to expand...

A squarewave of 50% cyclic ratio will be seen by a DC meter as half the peak voltage.
It looks like there is way too much gain. Datasheet says gain 54dB. Feeding -26dBV at the input would result in +28dBV output or 25Vrms, which largely exceeds the output capability of a pice that's powered with 12V. You need some attenuation at the input.

squarewave said:

The most important thing is probably ground layout. You want the ground end of the speakers to connect together at some point where all of the other stuff that is connected to ground converges at. In other words, if you have a long ground wire with the supply ground at one and and then a bunch of stuff connected to ground and then, at the other end you have the speaker grounds, that would be bad. A simple fix might be to just connect the supply ground right next to where the speaker grounds are and then all of the other grounded components are connected to a ground that branches off from there.

Click to expand...

Both speaker grounds, input grounds, and supply ground are all pretty close together, within a couple inches.  Maybe I should just solder them all right to the ground pin of the chip.  Can't hurt.

abbey road d enfer said:

It looks like there is way too much gain. Datasheet says gain 54dB. Feeding -26dBV at the input would result in +28dBV output or 25Vrms, which largely exceeds the output capability of a pice that's powered with 12V. You need some attenuation at the input.

Click to expand...

Or reduce the gain by adding resistors in series with caps on 5 and 8. But you have to reduce the caps accordingly. For example, if the resistors are 1K, the caps would then have to be more like 4.7u. Otherwise gain will be high at lower frequencies and that could contribute to instability.

Well, Squarewave wins!

Which is ironic, given what this amp was putting out up until a few minutes ago.

I redid all the grounding and basically just attached input, speaker, and power ground to the exact same spot on the perf board. 
Also, dropped the output of the tone generator to -60dBV to keep the speaker from flying across the room.  Sounds nice and clean now.  I will definitely need to place some form of attenuation on the inputs to tame whatever is going in.

CurtZHP said:

dropped the output of the tone generator to -60dBV to keep the speaker from flying across the room.  Sounds nice and clean now.  I will definitely need to place some form of attenuation on the inputs to tame whatever is going in.

Click to expand...

What did I tell you?
"It looks like there is way too much gain. Datasheet says gain 54dB. Feeding -26dBV at the input would result in +28dBV output or 25Vrms, which largely exceeds the output capability of a pice that's powered with 12V. You need some attenuation at the input."
Sometimes I wonder why I care to answer when no one seems to read...

abbey road d enfer said:

What did I tell you?
"It looks like there is way too much gain. Datasheet says gain 54dB. Feeding -26dBV at the input would result in +28dBV output or 25Vrms, which largely exceeds the output capability of a pice that's powered with 12V. You need some attenuation at the input."
Sometimes I wonder why I care to answer when no one seems to read...

Click to expand...

I did read.  That's why I turned down the tone generator.  (Quicker than installing an actual attenuator...)  And, not surprisingly, given the quality of advice I regularly receive here, it helped. 

abbey road d enfer said:

What did I tell you?
"It looks like there is way too much gain. Datasheet says gain 54dB. Feeding -26dBV at the input would result in +28dBV output or 25Vrms, which largely exceeds the output capability of a pice that's powered with 12V. You need some attenuation at the input."
Sometimes I wonder why I care to answer when no one seems to read...

Click to expand...

Then you would never get to repeat yourself.

JR

CurtZHP said:

I will definitely need to place some form of attenuation on the inputs to tame whatever is going in.

Click to expand...

Why attenuate when you can just reduce the gain? You're just amplifying noise.

squarewave said:

Why attenuate when you can just reduce the gain? You're just amplifying noise.

Click to expand...

Some amplifier chips are only stable when operated at higher gain** (check the data sheet).

JR

*** Higher closed loop gain means the NF is attenuated more providing more margin against instability as lag in the forward path approaches 180' phase shift at very high frequency.  NF mut be attenuated below unity before this phase shift equals 180' turning negative feedback positive. 

squarewave said:

Why attenuate when you can just reduce the gain? You're just amplifying noise.

Click to expand...

Well, the more I thought about it, the preamp stage I was planning on connecting to the input will likely take care of that for me.  I won't get to that until next week anyway.  I'm just glad the amp actually works.