For future reference..........
And lower noise. I would put a cap in the range of 470 uF to 1000 uF across the voltage reference, very close to it. For a very fast op amp you may want to add a 0.1uF cap in series with a 10 ohm resistor network across the reference and electrolytic. Short leads and small 0.1uF cap (50 V MKP should work well). A voltage reference uses feedback and will most likely be unstable inside a feedback loop with no cap or too small a value cap across it. Use a low voltage cap just a little over the voltage of the reference for low Q and low inductance. Something like a 50 V high Q ( low ESR cap) will be both inductive and likely the resonate with the inductive output impedance of the reference. Think of it as a noise filter AND a zobel network. LM336 impedance vs. frequency below:
Pretty much no charge with all the dead beats here.........
janneman said:
And lower noise. I would put a cap in the range of 470 uF to 1000 uF across the voltage reference, very close to it. For a very fast op amp you may want to add a 0.1uF cap in series with a 10 ohm resistor network across the reference and electrolytic. Short leads and small 0.1uF cap (50 V MKP should work well). A voltage reference uses feedback and will most likely be unstable inside a feedback loop with no cap or too small a value cap across it. Use a low voltage cap just a little over the voltage of the reference for low Q and low inductance. Something like a 50 V high Q ( low ESR cap) will be both inductive and likely the resonate with the inductive output impedance of the reference. Think of it as a noise filter AND a zobel network. LM336 impedance vs. frequency below:
Pretty much no charge with all the dead beats here.........Attachments
Interresting, but forgive my lack of certains basics, such as how to connect a zener instead if the FET? I'm not at the level of understanding transistor bias, FETs and the maths behind these circuits.
I have learned a lot throught these postings + reasearch on regulators etc, and I must thank you all for the input.
Here is an attempt at putting zeners instead of the FETs (attached). Is it correct?
As for the integrated 5V references, i'll have to order some, but I would like to try the zeners today.
I have learned a lot throught these postings + reasearch on regulators etc, and I must thank you all for the input.
Here is an attempt at putting zeners instead of the FETs (attached). Is it correct?
As for the integrated 5V references, i'll have to order some, but I would like to try the zeners today.
Attachments
Did I mess up.........
The zeners replace the 2K4 ohm resistors to the output of the op amps. The direction shown is correct. Leave the 2SK246 jfets where they were. Sorry if I confused anyone.
For some good regulator background:
http://tangentsoft.net/elec/opamp-linreg.html
http://home.comcast.net/~walt-jung/wsb/PDFs/Improved_PN_Regs.pdf
http://home.comcast.net/~walt-jung/wsb/PDFs/Low_Noise_Power_for_Analog_Circuits.pdf
The zeners replace the 2K4 ohm resistors to the output of the op amps. The direction shown is correct. Leave the 2SK246 jfets where they were. Sorry if I confused anyone.
For some good regulator background:
http://tangentsoft.net/elec/opamp-linreg.html
http://home.comcast.net/~walt-jung/wsb/PDFs/Improved_PN_Regs.pdf
http://home.comcast.net/~walt-jung/wsb/PDFs/Low_Noise_Power_for_Analog_Circuits.pdf
Working well!
Here are the results:
Bias resistors 668 Ohms
Zeners 5.1v (measured at about 4.7v)
Opamps outputs (using 10v zener instead of 2.4k)
+ side 4.31v
- side -4.33v
Regulators are working at zener voltage !
After further measurments it looks like the zeners I was using were way off: 4.6v instead of 5.1 and they required more current. I have some other zeners (same wattage) and these are closer to specified value and do not require as much current (tried them with 1k, 2.4k 4.5k bias resistors), almost the same output. I settled for 5.6v zeners (measured at 5.25v) with 2.4k bias resistors: Regulator output at + and - 5.25v
Thanks for help !
Here are the results:
Bias resistors 668 Ohms
Zeners 5.1v (measured at about 4.7v)
Opamps outputs (using 10v zener instead of 2.4k)
+ side 4.31v
- side -4.33v
Regulators are working at zener voltage !
After further measurments it looks like the zeners I was using were way off: 4.6v instead of 5.1 and they required more current. I have some other zeners (same wattage) and these are closer to specified value and do not require as much current (tried them with 1k, 2.4k 4.5k bias resistors), almost the same output. I settled for 5.6v zeners (measured at 5.25v) with 2.4k bias resistors: Regulator output at + and - 5.25v
Thanks for help !
So, that's about 1.4v across the 2.4k or .6mA, that's the reserve for driving the pass xsistor if the supply is being loaded. Was this measurement under load? What load?
Assuming an Hfe of 100 of the pass xsistor gives you about 60mA load current spare. Is that enough? What's the raw input voltage at the reservoir cap?
Jan Didden
Assuming an Hfe of 100 of the pass xsistor gives you about 60mA load current spare. Is that enough? What's the raw input voltage at the reservoir cap?
Jan Didden
What I got........ beside tired after all this
Actually a level shifting circuit like the CCS jfet and resistor are only necessary when driving the base of the pass transistor to voltages near are above the rail voltage for the op amp. With a 12 Volt raw supply feeding the op amp it can swing to about 10 volts, in the process changing from sinking current to sourcing it. The base of the transistor is sitting at about 5.6 Volts with the current source supplying base current. The op amp will swing to about 10 Volts (the op amp supply minus a couple of volts). so we have 10 volts minus 5.6 volts divided the 2.4 K ohm resistor or 1.8 mA. This added to the current source current which is all going to the base. The collector current is Hfe times the sum of these two current. Pulling a number out of the air for the Idss of the CCS fet of 3.2 ma for quick math and assuming hfe of a 100 for more lazy math, you have (1.8mA plus 3.2 ma) times 100 or 500 mA. If one is really going to take advantage of this level shifting on the output like "the regulator who dare not speak its name"; one should run the op amp off regulated 5 volts rails for better PS rejection. If you are going to power the op amp from the raw 12 volt rails you might as well leave out the level shifting circuit and put a 10 ohm resistor between the op amp output and pass transistor base for both simplicity and lower output impedance.
Perhaps some reverse engineering with more reversals than engineering resulted in what we see on the schematic. I would go for the zener- jfet level shifter and power the op amp from the regulated output voltages like the "Engineer whose name we dare not speak" did on the super regulator. Just a little humor WJ, please don't be offended.
Actually a level shifting circuit like the CCS jfet and resistor are only necessary when driving the base of the pass transistor to voltages near are above the rail voltage for the op amp. With a 12 Volt raw supply feeding the op amp it can swing to about 10 volts, in the process changing from sinking current to sourcing it. The base of the transistor is sitting at about 5.6 Volts with the current source supplying base current. The op amp will swing to about 10 Volts (the op amp supply minus a couple of volts). so we have 10 volts minus 5.6 volts divided the 2.4 K ohm resistor or 1.8 mA. This added to the current source current which is all going to the base. The collector current is Hfe times the sum of these two current. Pulling a number out of the air for the Idss of the CCS fet of 3.2 ma for quick math and assuming hfe of a 100 for more lazy math, you have (1.8mA plus 3.2 ma) times 100 or 500 mA. If one is really going to take advantage of this level shifting on the output like "the regulator who dare not speak its name"; one should run the op amp off regulated 5 volts rails for better PS rejection. If you are going to power the op amp from the raw 12 volt rails you might as well leave out the level shifting circuit and put a 10 ohm resistor between the op amp output and pass transistor base for both simplicity and lower output impedance.
Perhaps some reverse engineering with more reversals than engineering resulted in what we see on the schematic. I would go for the zener- jfet level shifter and power the op amp from the regulated output voltages like the "Engineer whose name we dare not speak" did on the super regulator. Just a little humor WJ, please don't be offended.
Some more measurments for sake of clarity:
Load is the working circuit, probably 50-100mA for the 5v regulators, a bit more on the 15v regulator (it gets a little hot)
1st 5v regulator
Output : +5.2 and -5.2 v
Opamp Out: -3.75 and +3.82
FET (after Zener): +5.8 and -5.82
2nd 5v regulator
Output : +5.18 and -5.25 v
Opamp Out: -4.10 and +3.70
FET (after Zener): +5.79 and -5.85
3rd regulator 15v
Output : +14.26 and -14.49v
Opamp Out: +4.91 and -5.34
FET (after Zener): +5.8 and -5.82
----
I'm uzing the zeners, as you suggested. Are you also suggesting to power the Opamps from the regulated side of the rails? Whould that still be ok in terms of output current spare?
Load is the working circuit, probably 50-100mA for the 5v regulators, a bit more on the 15v regulator (it gets a little hot)
1st 5v regulator
Output : +5.2 and -5.2 v
Opamp Out: -3.75 and +3.82
FET (after Zener): +5.8 and -5.82
2nd 5v regulator
Output : +5.18 and -5.25 v
Opamp Out: -4.10 and +3.70
FET (after Zener): +5.79 and -5.85
3rd regulator 15v
Output : +14.26 and -14.49v
Opamp Out: +4.91 and -5.34
FET (after Zener): +5.8 and -5.82
----
I'm uzing the zeners, as you suggested. Are you also suggesting to power the Opamps from the regulated side of the rails? Whould that still be ok in terms of output current spare?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.