I have an irritating problem that I just do not understand. I am using the Schottky diodes as protection on the output of my power amplifier. I also have a TVS, signal diodes, and a zener across the supply to protect the output of the power amplifier which will be connected to the AC mains. Right now, it is not connected to the mains and there is no coupling transformer attached at the moment either.
The problem is that I notice a 100 MHz oscillation on the negative cycle of an input sinewave. The amplifier is +12V single supply, biased to Vdd/2 and the output is 64kHz at 10Vpp which is AC coupled thru a 330uF capacitor to a 4Ω load. The oscillation is correlated to current draw. If the output current is low, <100mA, then the oscillation does not occur. If I raise it up beyond 100mA then the oscillation appears. This can happen if only one Schottky is populated.
I tested the amplifier with 2 500pF capacitors in place of the Schottkys to see if it was the capacitance causing a phase margin issue and the amplifier is stable. I used a varactor diode that was lying around to see by chance if a ΔF change through the output swing could cause it, but the amplifier remained stable.
I moved the Schottkys to different positions on the board thinking that it could be a point of resonance, but that did not reduce the oscillation.
The only mitigation I have found so far is that I can lift one side of the Schottky and then solder a 5" wire to the other side and then finish the connection to the pad on the board. BAM, the oscillation goes away.
The resistance of a 5" piece of wire has to be <1mΩ for practical purposes, but maybe it is the inductance of the wire? I am at a total loss.
I am hoping someone has seen this before and says, "Oh yeah, it's because...."


Thanks.

 

Hello,

Do you have decoupling capacitors accross the powersupply and on the bias voltage?

Greetings,
Bertus

 

Hello,

Do you have decoupling capacitors accross the powersupply and on the bias voltage?

Greetings,
Bertus

Yes, the bypass capacitors (not shown) are 220uF||10uF||1uF on the +12Vdd pin. The bias voltage is generated from a simple divider from the +12Vdd line as well with a 1uF||.1uF cap at the bias node.

 

Add a low value resistor directly in series with the opamp output pin (between that and the feedback resistor).

This is a standard mod when driving a capacitive load, many amps become unstable if directly connected to some loads.

 

Hello,

The diodes do not meet the specs from the datasheet:

It says 3A schttky diodes.
The BA320 is a fast diode:
BA320


General Purpose Fast Rectifier
Various
V(RRM)(V) Rep.Pk.Rev. Voltage=100
V(FM) Max.(V) Forward Voltage=1.1
@I(FM) (A) (Test Condition)=100m
I(O) Max.(A) Output Current=250m
t(rr) Max.(s) Rev.Rec. Time=50n
@I(F) (A) (Test Condition)=10m
@I(R) (A) (Test Condition)=1.0m
Semiconductor Material=Silicon
Status=Discontinued
Package=DO-35
Military=N

The BAV99 is a 0.2 A fast recovery diode.

These both are not schottky diodes.

Greetings,
Bertus

 

Well...

1) Typo in the schematic on my part. It is B320A which is a Schottky from Diodes Inc.

2) In place of the Schotty diodes, I used 500pF capacitors which is far greater than the capacitance of the Schottky diodes themselves and there was no instability.

Typical values of isolation resistors I have seen are 50Ω to 100Ω, but a piece of 5" wire used to complete the connection of the Schotty to the board pad is ideally 0Ω.

I would argue that it is not a capacitive load drive issue.

 

Hello,

The B320A has an avarage current of 3A.
The datasheet recomends a 3A continuous rated diode.

The other diodes are not needed.

Besides they use 47uF and a 0,1 uF capacitors for the decoupling.

Greetings,
Bertus

 

Hello,

The B320A has an avarage current of 3A.
The datasheet recomends a 3A continuous rated diode.

The other diodes are not needed.

Besides they use 47uF and a 0,1 uF capacitors for the decoupling.

Greetings,
Bertus

I do not need a continuous rated diode for the application; regardless, how would a Schottky cause an oscillation at 100 MHz? Why would a theoretical 0Ω connect eliminate it?

 

Hello,

What is the shape of the 64 kHz input signal?
If it is a squarewave the diodes together with the wire can "filter" out the 100 MHz of it, as the diode is a not linear device.

Greetings,
Bertus

 

Hello,

What is the shape of the 64 kHz input signal?
If it is a squarewave the diodes together with the wire can "filter" out the 100 MHz of it, as the diode is a not linear device.

Greetings,
Bertus

The output is a sine wave with a 100 MHz oscillation superimposed on the negative cycle only.