Looking for OpAmp Oscillation

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Gold

Well-known member
Joined
Jun 23, 2004
Messages
3,670
Location
Brooklyn
I haven’t been successful in recognizing low level oscillation on an oscilloscope.

Yesterday I noticed a higher noise floor on one channel of a line amp. I touched the op amp and it was hot. I then realized I forgot to put rail bypass caps in that channel.

Before I fixed it I tried to look at the oscillation on the scope but it looked like broadband noise. I couldn’t tune in on a specific frequency.  Measured with the probe on the output pin.

Am I doing something wrong? Could someone post a snapshot of op amp oscillation on a scope?
 
The frequency of the oscillation may be above what the scope can easily sync/lock to.

When you hear oscillation as wide band noise, you are actually hearing rectification from the circuitry not being fast enough to keep up.

This is the same mechanism as radio station interference in input gain stages, but in the RF case the rectification decodes the radio signal modulating the RF carrier. For a relatively simple instability oscillation it decodes randomness in the oscillation (that may also make it harder for the scope to sync to).

JR

[edit if you see an unusually thick audio trace with the scope , that often suggests low level oscillation... [/edit]
 
What's the model and bandwidth of the scope? Are you using a 1 to 1 or a 10 to 1 probe? A 10 to 1 is preferred due to its higher resistance and lower capacitance.

You could try putting the probe very close to  (like 1/4 inch) the output pin and see if that shows up on the scope.  The capacitance of the probe tip could change what the pin is doing, and a small "capacitive" connection might show it better.
 
benb said:
...A 10 to 1 is preferred due to its higher resistance and lower capacitance....The capacitance of the probe tip could change what the pin is doing, and a small "capacitive" connection might show it better.

Agreed, just to add that sometimes probe's capacitance can stop temporary the oscillation at all.
 
benb said:
What's the model and bandwidth of the scope? Are you using a 1 to 1 or a 10 to 1 probe? A 10 to 1 is preferred due to its higher resistance and lower capacitance.

You could try putting the probe very close to  (like 1/4 inch) the output pin and see if that shows up on the scope.  The capacitance of the probe tip could change what the pin is doing, and a small "capacitive" connection might show it better.

It's a Tek TAS 465 100MHz. I'm using a probe that has a a switch for 1x and 10x. I was using the 1x. I'll try the 10x next time.
 
Hi Gold

I use my $3 Spectrum Analyzer, it is  AM battery powered radio with a loopstick antenna. Tune it to a blank station and move it around the PCB. If it goes silence the oscillation has no modulation, if it get louder and has a buzz the circuit has modulation of some type.

Checking for oscillation with a scope use a 10:1 probe and a 200 Ohm resistor is series with the tip (cut the leads short) this will reduce the extra stray C.
Duke
 
Audio1Man said:
Hi Gold

I use my $3 Spectrum Analyzer, it is  AM battery powered radio with a loopstick antenna. Tune it to a blank station and move it around the PCB. If it goes silence the oscillation has no modulation, if it get louder and has a buzz the circuit has modulation of some type.

Checking for oscillation with a scope use a 10:1 probe and a 200 Ohm resistor is series with the tip (cut the leads short) this will reduce the extra stray C.
Duke

Thanks for the tips. I think a battery powered AM radio is in the budget. I'll pick one up.
 
moamps said:
Agreed, just to add that sometimes probe's capacitance can stop temporary the oscillation at all.
ding ding ding... I have heard stories of huge one-off military/government electronic system builds, shipped with a scope probe still attached, hidden inside.  ::)

JR
 
Trying to start figuring out why some LME49990 op amps are running very hot in a DAC . Looked at sine on pin 6 output and it looks perfect. No oscillation afaik....


Here's data sheet
 

Attachments

  • LME49990.pdf
    515.5 KB · Views: 0
Last edited:
Unity gain stable so should not be oscillating... these uber op amps draw 10+mA so over 1/3 Watt of heat dissipation with +/- 15v rails, with load even more heat.

JR

OUTPUT DRIVE AND STABILITY
www.ti.com
page10image52206080
APPLICATION HINTS
The LME49990 is unity gain stable from both input (both stable when gain = -1 or gain = 1). It able to drive resistive load 600Ω with output circuit with a typical 27mA. Capacitive loads up to 100pF will cause little change in the phase characteristics of the amplifiers and are therefore allowable.
Capacitive loads greater than 100pF must be isolated from the output. The most straight forward way to do this is to put a resistor in series with the output. This resistor will also prevent excess power dissipation if the output is accidentally shorted.
The effective load impedance (including feedback resistance) should be kept above 600Ω for fast settling. Load capacitance should also be minimized if good settling time is to be optimized. Large feedback resistors will make the circuit more susceptible to stray capacitance, so in high-speed applications keep the feedback resistors in the 1kΩ to 2 kΩ range whenever practical.
OUTPUT COMPENSATION
In most of the audio applications, the device will be operated in a room temperature and compensation networks are not necessary. However, the consideration of output network as shown in Figure 28 may be taken into account for some of the high performance audio applications such as high speed data conversion or when operating in a relatively low junction temperature. The compensation network will also provide a small improvement in settling time for the response time demanding applications.
Figure 28. LME4990 Output Compensation Network
SUPPLY BYPASSING
To achieve a low noise and high-speed audio performance, power supply bypassing is extremely important. Applying multiple bypass capacitors is highly recommended. From experiment results, a 10μF tantalum, 2.2μF ceramic, and a 0.47μF ceramic work well. All bypass capacitors leads should be very short. The ground leads of capacitors should also be separated to reduce the inductance to ground. To obtain the best result, a large ground plane layout technique is recommended and it was applied in the LME49990 evaluation board.
page10image52206496
 
these uber op amps draw 10+mA so over 1/3 Watt of heat dissipation with +/- 15v rails, with load even more heat
Yeah, from what I can tell so far there are 12 of those 49990 chips in total. think running at 16v... Haven't checked....40va transformer....

There are little strips of aluminum just stuck on top of 4 of the chips and I was curious why not on the others. Turns out they are on single to dual adapters so (2) lme49990 right on top of each other (I'm ssuming the bottom one is the same). There was silicone used all over the place and to hold the aluminum strips in place but, under these single to dual adapters, there's what looks like some grease? I can't tell if it's broken down silicone but it looks like it could be some kind of clearish grease...idk....
any reason why packing the bottom/board side of these adapters

Single-to-dual Op-Amp Adapter - 020302 - Cimarron Technology

with grease would be a good idea?
 
Last edited:
clever but now over 1/2W from +/-15V rail.

A crude test to measure temperature is to put a drop of spit on the IC package. If the spit boils away we know it more than 100'C. IC packages are stable to a bunch higher than that, but not happy.

JR
 
Yeah they were hot enough to boil it away after 20 seconds or so. Same with the regulator heat sinks.... Weird thing, when I was scoping the output of one of the chips, I heard a pop, then everything was dead.... May have shorted something..idk.
Anyhow , looking at the giant AC wall wart transformer, it didn't seem like it was working right from the readings I was getting.. I had a suitable transformer that once I rigged up, everything works and everything is cool enough to hold onto as far as I can tell.... couldn't hold anything for more than a half second before... I got tired of waiting to get burned so seems good so far...
Goofy transformer can cause circuit to overheat? Was a 16v transformer and I'm using a 15v one now...
 
Perhaps they were using the PCB as a heatsink for the lower device.
Maybe. It doesn't seem like there are any pads underneath but it's hard to tell. Are you saying that using grease in the air gap to get heat down to the pcb in general is a way to help dissipate it?

Did you verify the voltage was correct for both rails?
Before the mishap, yes, both rails were +/-15v ish....on a couple of chips tested....
When testing the wall wart transformer after the mishap, there I could only get a reading of 10v to the centertap on one terminal but nothing across both .The power supply is similar to this one in the pic in general but 16v tansformer, with one leg grounded... didn't expect to see something like that ....

Center tap wasn't used....Just saying that's the only reading I could get once it goofed....
Old transformer was a 40VA and the one I have now is 48VA....
 

Attachments

  • wall wart supply with load resistors.gif
    wall wart supply with load resistors.gif
    10.3 KB · Views: 1
Last edited:
Are you saying that using grease in the air gap to get heat down to the pcb in general is a way to help dissipate it?

Not really, you would need to clamp the assembly down some way to keep it pressed against the PCB. That was just my first thought, that perhaps the op-amp on the bottom was being pressed against the PCB, and there was thermal paste to help couple the surfaces. Just a big lump of paste is not effective though, it needs to be a very thin layer that just fills in surface imperfections.
 
I do not advocate operating ICs with more than 100'C package temp...

Technically they can survive it, but I wouldn't bet on long term reliability
Thanks. They are running much much cooler now. Not sure why? :confused: Only thing that has changed is the transformer.

After I get to putting everything together ,I'll run it for a long while and see how it goes. Before everything would get scorching within a minute of turning it on. Now things just seem to barely get warm after several minutes but I haven't gone past that yet.
 

Latest posts

Back
Top