killing op amp osclation

I have been experimenting with some video op amp with good result. Often they have oscilation as i soldered those SOIC to a DIP sockets for easy op amp rolling. In the case of single op amp, a lytic cap across the +/- ve ps is sufficient. With Dual op amp, I add a 12pf across the +/- ve input, plus a lytic cap across the +/- ve PS.
I have seen some pre with the b pass caps across the +/- ve input,. but could not find any discussion on the bet.
Has any body tried it, and what value do you use?
I don't really like to use by-pass cap at the feedback Resistor, that connect output to -ve input. I suspect with phase shift this willl cause instability at high freq. But in most amp, the designers actually said they used feed back by caps to corrct the phase. Well, wouldn't it actyally add phase shift??

rgds

William Lee
 
Note that proper power supply bypassing is required. The opamp manufacturer's datasheet should have the recommended power supply bypass capacitor values. Try those values first, in the recommended connection configuration. If you don't have the datasheet's bypass C values, try something like 0.1 or 0.01 uF X7R (or maybe even NPO/C0G?) ceramic in parallel with a small (say 10 uF) electrolytic (aluminum; or maybe a smaller tantalum), from each opamp power pin to ground. The bypass caps, especially the ceramics, should be placed VERY close to the opamp's power supply pins, with the shortest-possible connections to the pins. Each millimeter is significant, there. You can also connect additional bypass capacitance directly between the two power pins (Use minimum lead length.).

Also, "just in case", check the opamp manufacturer's datasheet to see if any minimum gain configuration is required, for stability.

Using IC sockets, like that, might be a significant problem.

Paralleling the feedback R with a small NPO or C0G ceramic capacitor WILL help. The larger the cap value, the less high frequencies will be a problem. Using a C there will usually enable you to stop any oscillation. But you should know or find the reason for the oscillation, and correct its cause, instead, if possible. Even without any oscillation problem, it's often a good idea to use a small C, in parallel with the feedback R, just to limit the response above the frequencies of interest. You can use the RC filter equation, below, to get some idea of what C value to add to your feedback R. It will be in the pF range.

You could also try a Zobel network on the output pin. (See what the opamp manufacturer's datasheet recommends.)

You should also definitely have an RC lowpass RF filter, just before the input pin: If you have a non-inverting configuration (i.e. using + input), just use a small-ish R in series with the + input, and then a small cap to gnd from between the R and the input. You can just pick an R value and the lowpass -3dB cutoff frequency and then calculate the C value with f = 1/(2 x Pi x R x C), or C = 1/(2 x Pi x R x f), where Pi = 3.14. For example, a 1K R and a 220pF C would give a cutoff frequency of about 723 kHz. You can use smaller R values, if you scale the C value up, proportionately. NOTE that if you are using an inverting configuration (i.e. using - input), then you should use two same-value resistors in series with the - input, with the cap to ground from between the two resistors.

When prototyping with very fast (e.g. video) opamps, you have to really worry about things like wire/lead lengths (keep them very, very short). And input and output (and power) wiring can be a big problem (think "antennas"). It's usually a good idea to build the prototype on a bare PCB's copper (used as a ground plane). Then solder BNC or coax connectors to the ground plane, as close as possible to input and output pins, so you can use shielded I/O cabling. (You should not try to use a "solderless breadboard", for example, because they have far too much stray capacitance.)
 
hi Tom,

tks for a very detailed explaination, I am trying the THS 4012-THS4042 opa mps, so put them on the sockets allowed me to switch quickly, though I know I iwll hv the risk of oscilation.
THS4022 being 350Mhz and rise time of 400V is the most difficult to settle.
Initially I solder to the top of the DIP8 socket, with a ceramic 0.01uf solder directly to the the op amp pin 4 and 8, right after a 5 ohm in between the +/- PS to the op amp pin 4 and 8. No music but hiss similar to FM noise. Adding 20 pf, then 30 pf silver mica across both the inv n non inv input resulted in music, but not stable, with same hiss still. replacing the cap wit 39pf ceramic, the smallest I hv ceramic I have, cure the stability problem, but still hiss. The DC offset were at 0.5V and 1.2 V in the 2 ch. Then while using my test meter probe at the op amp output, the hiss reduce a bit. So remembering what I read from some advise, I soldered a 130 ohm right at the soic output pins. Now the op amp is functionally normal with no noise at all! It seemed that the wire connecting the ouput pins acted like antenna, and yes, I have a single wire of about 10" connected to the output...!
I do not have Zobel at the output, but there is one at the input, in front of the inpur R. This is the original design in the pre amp.
I think the mose important thing is to solder every thing directly to the pin of the op amp.
I am surprised that the offset is now much lower at about 0.15V. Still have by usual standard, but considering I have the +ve input connected directly to the volume control, this is ok, as there is an input cap at the power amp. After I am clear about the op amp sound I will need to add capacitors at the input.


best rgds

William Lee