Chip Amp Oscillations!!!

[john65b]

Hey all,

I am stumped on this one - tried searching the forum and believe it or not, could not find anything...I have been scratching my head for months...

I built a point to point GC with the 9 components per the original. Originally had the input cap removed and works perfectly with my 12B4 preamp. But once I put my passive pre in front of the amp, Oscillates like crazy. Tried my Lightspeed pre and again oscillates like crazy. Tried my Foreplay, and another tubed pre and works great.

I put in the Input caps on the amp, and it still oscillates like nuts with the passive pre, LS, but not with the tubed preamps....

I do not have bypass caps on the 2200uF caps on rails at chip. I basically have it as shown on Mick's url:

http://dogbreath.de/Chipamps/GainCardCopy/GainCardCopy.html

Identical, but with 30V rails, and a Preamp where the front end Pot is. And I do have 6800uf caps/Rail on the supply end too.

I put a voltmeter across the dummy resistors while it was oscillating and read 20VAC (no DC on outputs that I could measure) before the 2A fuse blew.

Any advice???

 

[john65b]

One of the cool things that came out of this "problem" is I built an oscillation detector...

I am one of those poor slobs in this hobby without a scope (well I have one - bought it on ebay for $20 - 'nuf said)

Anyway on another thread, Unclejed613 suggested building a cheap and easy oscillation detector with a couple of components...

http://www.diyaudio.com/forums/showthread.php?s=&threadid=112878&perpage=25&pagenumber=3


"you might want to try a "quick and dirty" oscillation detector using an LED. you want to take an LED, put a 100 ohm resistor in series with it, then put a 4.7nf cap in series with the 100 ohm resistor. the "unused" leg of the LED becomes ground (polarity of the LED doesn't matter). the open leg of the cap becomes the input. then put a 1k resistor from the junction of the cap and series resistor to ground. the circuit is connected across the speaker terminals. any oscillation above 1 volt and above 20khz should light the LED, especially if the oscillation is continuous.

LED's are such useful devices......"


Very Cool! Pic attached with my 25W 8.2 load resistors - I just attach to the amp outputs and led light during oscillations. I built it in 25 minutes and will plan on using it on all my amps to spare possible damage to speakers...

Anyway, the detector ights up like a Christmas tree with the passive pre or lightspeed preamp hooked up to the Gainclone...

 

[kaos]

First thing I’d try is placing a resistor in series with Cin (about 1K), and a 220pF cap across Rg.

 

[Stormrider]

Does it oscillate with a load, such as a speaker or power resistor? Perhaps try adding a zobel?

 

[john65b]

Only oscillated with passive preamp attached...

There is a zobel at end of the amp - see the url.

100nF and 2.7 ohm.

I actually added a 1K resistor in series with Cin (between Cin and amp input) - no more oscilations, but low volume on passive pre.

 

[Duo]

It sounds as if your amplifier oscillates when driven from a possibly low impedance source. Most generally, your amplifier chip should have some resistance in series with its input, note that many audio amps like this use something in the area of 1k, and also something between 10K and 100K to ground to bias the amplifier to zero volts DC offset.

I fixed a gainclone for a friend recently which had the input series resistors, but no grounding resistors, and would float around with DC offset notably just pulling right to one rail. This is bad for your speakers! haha.

EDIT: Bypass capacitors are also a good idea to keep things quiet and/or stable.

 

[gootee]

You should probably have a small cap across the feedback resistor, Rf, too. But you could try lower values, first, and use the lowest value that works. I'd start with at least 15 pF, and increase it if you have to.

And you should _definitely_ have 0.1uF caps directly from (right at) each power pin to power ground.

After adding a resistance, as already suggested, in series with the + input, I would add a small capacitor to ground, from the + input pin, to act as a (high-frequency) lowpass RF filter. If you use a 1K resistor, there, then 330pF to ground would give a lowpass corner frequency of f=1/(2*Pi*R*C) = 482 kHz.

If you haven't already done so, you will probably want to download and study the datasheet, from http://www.national.com .

A photo of your layout might be helpful.

 

[Duo]

Datasheets are good! Some application notes and data will talk about stability as well, and good ideas and methods of properly using the chip.

 

[peranders]

I do not have bypass caps on the 2200uF caps on rails at chip. I basically have it as shown on Mick's url:

How much do you have in total? A good idea is the read the datasheet and the AN-1192.

 

[john65b]

I have 2200uF at the chips/rail and 6800uF at the PS.

I have removed the 4.7uF cap and just placed the 1k resistor in series on the inputs. All is well. No more oscillations. My passive pre is a 24 step attenuator that the highest volume setting is 30k ohm attenuation - I forgot that full clockwise is not 0 ohm on this attenuator - just checked with my lightspeed and all works fine.

I do not want to add all the other residual bypass/resistors/components and want to stay minimalistic like the original. I knew something simple was keeping it from working properly. I wonder if a real Gain Card world also oscillate in this setup without that 1k series resistor - I guess I can pay $5000 to purchase one to find out...nah.

Thanks for the help...

 

[AndrewT]

Hi,
if the only modification to prevent oscillation is the 1k0 input, then I suspect your phase margin is very low.
I recommend adoption of National's standard build components.

If your minimalist amp bursts into infrequent oscillation then some day something is going to break.