OPA134 is a good opamp. It is very stable. It has rather low distortion.
Here I use it to control a power buffer of EXICON MOSFET.
Supply should be a 2x12VAC transformer. Giving like 2x15VDC rails.
Data:
Bias should be like 500mA. This means power to cool is like 15 Watt per channel.
THD 0.00011% at 1 Watt 1 kHz into 8 Ohm.
Upper frequency 450 kHz.
Max output a bit more than 9 Watt into 8 Ohm. Of course more into 4 Ohm.
Gain margin is very good, as is Phase margin.
Enjoy 🙂
Here I use it to control a power buffer of EXICON MOSFET.
Supply should be a 2x12VAC transformer. Giving like 2x15VDC rails.
Data:
Bias should be like 500mA. This means power to cool is like 15 Watt per channel.
THD 0.00011% at 1 Watt 1 kHz into 8 Ohm.
Upper frequency 450 kHz.
Max output a bit more than 9 Watt into 8 Ohm. Of course more into 4 Ohm.
Gain margin is very good, as is Phase margin.
Enjoy 🙂
Last edited:
The input to the buffer has an impedance of 12k.
Thanks to the bootstrap.
So, it is an easy load for the OPA134 opamp.
This lowers the distortion of the circuit a bit.
Thanks to the bootstrap.
So, it is an easy load for the OPA134 opamp.
This lowers the distortion of the circuit a bit.
looks like your having fun with bootstraps.
R7/C4 radio frequency filter.
Depending on op amp compensation and bandwidth.
Can either help or ruin phase margin, depends
R4 seems like high value, might cause DC offset.
R4/R6 typically set to same value
R7/C4 radio frequency filter.
Depending on op amp compensation and bandwidth.
Can either help or ruin phase margin, depends
R4 seems like high value, might cause DC offset.
R4/R6 typically set to same value
Pardon my ignorance, but could you explain to an ignoramus how a filter outside the fb loop could ruin phase margin?
How do you adjust bias? Perhaps some part of 1k resistors, r3/r16, could the trimmer?
How do you set zero dc on output? Does opa loop play a role?
The bias is astronomically high. So if the set resistors get you in the ball game real life who cares?
Not sure if temps were set in sim or if the thermal models are accurate in Exicon models.
Remember the models being junk , so stopped using them years ago.
I set and test temps in sim over wide range and BJT models I use are thermally accurate.
They have easily aligned in sim and real life if done right.
Agree though with the DC offset, curious what it is.
The model is simplified of course, opamp and power section would have the usual power decoupling
Not sure if temps were set in sim or if the thermal models are accurate in Exicon models.
Remember the models being junk , so stopped using them years ago.
I set and test temps in sim over wide range and BJT models I use are thermally accurate.
They have easily aligned in sim and real life if done right.
Agree though with the DC offset, curious what it is.
The model is simplified of course, opamp and power section would have the usual power decoupling
Source impedance can have an effect on stability in much the same way as load admittance, see https://www.diyaudio.com/community/...lifier-stability-and-source-impedance.420032/ The effect is small when the source impedance is low compared to the open-loop input impedance around the frequency or frequencies where the magnitude of the loop gain goes through unity.
The output impedance of the filter is dominated by its capacitor at high frequencies, ensuring that the op-amp is driven by a predictable impedance at high frequencies, no matter what may be connected to the input connector. 150 pF is a lot more than the 8 pF differential input capacitance of the OPA134, so it sees essentially an ideal source at the frequencies where it matters. I would therefore expect the filter to improve stability, if it does anything at all, assuming that the feedback loop is designed to work well with a low-impedance source.
The gain bandwidth product of opa134 is 8MHz. The circuit has 8x voltage gain. Thus, the unit loop gain bandwidth is about 1MHz. I would call it very stable.
Indeed typical audio opamps from 3 to 10 MHz are not to cumbersome to stabilize in models/real life
with impedance buffer wrapped in feedback loop.
Tend to change source and power supply impedance in models and do high frequency sweeps.
Tend to stick to TI models and Tina since real time scope and rather good function generator.
Even if your bode plot shows good margin.
Much like real life with different sources and stepped input signals at high frequency.
You will catch stability issues in real time on the scope.
with impedance buffer wrapped in feedback loop.
Tend to change source and power supply impedance in models and do high frequency sweeps.
Tend to stick to TI models and Tina since real time scope and rather good function generator.
Even if your bode plot shows good margin.
Much like real life with different sources and stepped input signals at high frequency.
You will catch stability issues in real time on the scope.
This is the Scope at 1 kHz with my new version in post #1.
I have added a capacitor 10pF across the feedback resistor.
Now there is no need for input filter.
I have added a capacitor 10pF across the feedback resistor.
Now there is no need for input filter.
Don’t do that. This trend has to be stopped.
The Zero introduced here is to cancel out the 2nd Pole. However, most people apply this blindly. Here, the 2nd Pole could be caused by unnecessarily big gate stopper resistors. Although the gate stoppers are bigger than what I like, they are still in normal range. As the 2nd Pole is above the unit loop gain bandwidth, you don’t need the cap on feedback resistor.
What will happen is that the cap moves the unit loop gain bandwidth from 1MHz to a higher point, like 8MHz. That may cause instability.
Last edited:
R18 and R19 combine with the gate stoppers to further compromise the bandwidth of the output stage. If you shunt R18 and R19 with caps (eg. 100nF), you’ll be able to increase amp loop bandwidth.
If you try @stigigemla suggestion, make the nodes joining the gates to the bias resistors (ie, R1, R18 and R2, R19) as compact as possible to preclude oscillation.