In the Original Circuit, the Plus and Minus Supply Rails were 64 volts, so I wound my Power Transformer to supply the same DC voltage. when first carefully powering up the
circuit from My Variac, I did notice that the zener current did seem a bit low, so maybe I will do as you suggest and lower the resistors to 4K7. at the moment, I don't have 5K1. There is 52 Volts dropped across each resistor, so using 4K7, the resultant current is 11 mA, with dissipation of .57 watt. They are 1 Watt Zeners, so do You think this should be okay? -- As for the Output Device Gate Protection, I see on the Datasheets that these latfets already come with back to back Zener diodes connected between the Gate and Source Terminals, so I guess this is one thing less to worry about.
Thanks for All!
circuit from My Variac, I did notice that the zener current did seem a bit low, so maybe I will do as you suggest and lower the resistors to 4K7. at the moment, I don't have 5K1. There is 52 Volts dropped across each resistor, so using 4K7, the resultant current is 11 mA, with dissipation of .57 watt. They are 1 Watt Zeners, so do You think this should be okay? -- As for the Output Device Gate Protection, I see on the Datasheets that these latfets already come with back to back Zener diodes connected between the Gate and Source Terminals, so I guess this is one thing less to worry about.
Thanks for All!
Some small mods to your circuit
I have finally found the time to do some design work and to come up with some small mods to your circuit that should clear up some issues you are having.
Here is what the sim DOES NOT address:
Bypass caps for the opamp. No need to put them in a sim. Power supplies and wires are perfect in LTSpice! They do need to be on the PCB though.
I also did not add the Zobel networks between the output and the 8 ohm load representing the speaker. Since we don't have a speaker model anyway, they likely won't make much difference in a simulation, but they need to be there in the finished product.
Here is what I did to modify your circuit:
I eliminated the offset adjust pot. In the original configuration, this pot was being used to correct for offsets in the output stage. The offsets in the opamp are relatively trivial. Using offset pots in this way will tend to bias the opamp such that it may not have its full linear excursion available. I added C3, which allows the feedback network to have unity gain at DC, so any offset that is seen now is purely from the offset of the opamp.
I chose a different opamp that has a somewhat lower offset voltage, and a higher unity-gain bandwidth. It also can supply the current needed to drive the emitters of Q1 and Q2. It is available in a SOIC package, but you can get inexpensive adapters to make it fit a dip socket ok.
I provided a means to adjust the quiescent currents of the output transistors by changing a single resistor, R20. This circuit works quite well and lets the bases of the 2 transistors float so that the output voltage can stabilize at 0 volts with the feedback. This is accomplished with transistors Q3 and Q4, although the choice is not very critical. This circuit is one that Bob Cordell refers to as a "bias spreader". I also increased the current in the network feeding the bases of Q1 and Q2, which makes the bias point a little more stable. In practice, I would likely replace R20 with a pot. Also, the circuit is simple enough that it could be easily kludged onto your existing board by eliminating the diodes, etc. in the base circuits.
I also worked some with the compensation capacitor, C2, checking phase margin, bandwidth, and distortion. The value I would up with seems to be a good compromise; I get around 60 deg of phase margin, and there is a slight overshoot on squarewaves. For a musical instrument amplifier, bandwidth much beyond about 10K or so is not very useful. (Think of what sort of speakers they usually feed! A 12" speaker is not going to get much beyond 3 or 4 kHz at best.) Compensation is dependent on the input resistance of the opamp circuit, R13, R14, and C1 so I would not change these values very much at all.
Note: the models are from this site.
The circuit seems to perform well in simulation, but remember that "nature takes sides with the hidden flaw" and my corollaries:
1.) Oscillators are easy to build unless you want to.
2.) A real board layout with real components trumps all simulations.
Good luck. Let us know how it goes.
I have finally found the time to do some design work and to come up with some small mods to your circuit that should clear up some issues you are having.
Here is what the sim DOES NOT address:
Bypass caps for the opamp. No need to put them in a sim. Power supplies and wires are perfect in LTSpice! They do need to be on the PCB though.
I also did not add the Zobel networks between the output and the 8 ohm load representing the speaker. Since we don't have a speaker model anyway, they likely won't make much difference in a simulation, but they need to be there in the finished product.
Here is what I did to modify your circuit:
I eliminated the offset adjust pot. In the original configuration, this pot was being used to correct for offsets in the output stage. The offsets in the opamp are relatively trivial. Using offset pots in this way will tend to bias the opamp such that it may not have its full linear excursion available. I added C3, which allows the feedback network to have unity gain at DC, so any offset that is seen now is purely from the offset of the opamp.
I chose a different opamp that has a somewhat lower offset voltage, and a higher unity-gain bandwidth. It also can supply the current needed to drive the emitters of Q1 and Q2. It is available in a SOIC package, but you can get inexpensive adapters to make it fit a dip socket ok.
I provided a means to adjust the quiescent currents of the output transistors by changing a single resistor, R20. This circuit works quite well and lets the bases of the 2 transistors float so that the output voltage can stabilize at 0 volts with the feedback. This is accomplished with transistors Q3 and Q4, although the choice is not very critical. This circuit is one that Bob Cordell refers to as a "bias spreader". I also increased the current in the network feeding the bases of Q1 and Q2, which makes the bias point a little more stable. In practice, I would likely replace R20 with a pot. Also, the circuit is simple enough that it could be easily kludged onto your existing board by eliminating the diodes, etc. in the base circuits.
I also worked some with the compensation capacitor, C2, checking phase margin, bandwidth, and distortion. The value I would up with seems to be a good compromise; I get around 60 deg of phase margin, and there is a slight overshoot on squarewaves. For a musical instrument amplifier, bandwidth much beyond about 10K or so is not very useful. (Think of what sort of speakers they usually feed! A 12" speaker is not going to get much beyond 3 or 4 kHz at best.) Compensation is dependent on the input resistance of the opamp circuit, R13, R14, and C1 so I would not change these values very much at all.
Note: the models are from this site.
The circuit seems to perform well in simulation, but remember that "nature takes sides with the hidden flaw" and my corollaries:
1.) Oscillators are easy to build unless you want to.
2.) A real board layout with real components trumps all simulations.
Good luck. Let us know how it goes.
Hello
Thank you so much for going to the trouble to do what You have done to try and Help.
Unfortunately, I can't see what you have done, as I am unable to open the files at the moment, as it says that I don't have the App. to do that. What App do I need to open them? I run Windows 10 on this machine.
Thanks, and Kindest Regards; Telnet100
Thank you so much for going to the trouble to do what You have done to try and Help.
Unfortunately, I can't see what you have done, as I am unable to open the files at the moment, as it says that I don't have the App. to do that. What App do I need to open them? I run Windows 10 on this machine.
Thanks, and Kindest Regards; Telnet100
How run .asc files
LTSpice is available free at Analog Devices website. It is a totally cool simulator with just about any feature you could want, although the GUI is a bit quirky.
The .txt file has models for the MOSFETs. Just put both files in the same folder, start up LTSpice, and open the .asc file and you should be in business.
LTSpice is available free at Analog Devices website. It is a totally cool simulator with just about any feature you could want, although the GUI is a bit quirky.
The .txt file has models for the MOSFETs. Just put both files in the same folder, start up LTSpice, and open the .asc file and you should be in business.
I suspect several resistors (and maybe caps) are the wrong value, especially the gate resistors which should never be more than 1k. My guess is they are out by a factor of 10, ie 2k2 ->220 and 3k3 ->330. But any op-amp circuit will be hard to stabilize. I would test it out in simulation with lots of phase margin, and give any high-Z nodes lots of clearance in layout and keep them short.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.