Hi all,
I was mooching around the net and came here.
http://www.geocities.com/rjm003.geo/rjmaudio/diy_twi.html
The power supply has a Mosfet for line smoothing. Is anything new or good? Does it compare to the usual regulaters, LM...? I'm always curious with acircuit topology I havn't seen before. I googled it and searched here with, Mosfet line smoother, and came back with nothing. Also, didn't Technics use a mosfet in one of there designs??
Any info makes enjoyable reading.
Thankyou
Mikee55
I was mooching around the net and came here.
http://www.geocities.com/rjm003.geo/rjmaudio/diy_twi.html
The power supply has a Mosfet for line smoothing. Is anything new or good? Does it compare to the usual regulaters, LM...? I'm always curious with acircuit topology I havn't seen before. I googled it and searched here with, Mosfet line smoother, and came back with nothing. Also, didn't Technics use a mosfet in one of there designs??
Any info makes enjoyable reading.
Thankyou
Mikee55
This is also called a capacitance multiplier; that should find you lots of hits on the net.
Also see this:
http://www.passdiy.com/pdf/zen-ver3.pdf
Also see this:
http://www.passdiy.com/pdf/zen-ver3.pdf
This won't work right if the output voltage is low. Mosfets usually need about 10 volts applied to the gate to fully turn on. At low voltages (ie 5V or 3.3volts) the turn on gate voltage is likely too high. Usually BJT transistors are used.
FWIW: It would more practical to use a Linear regulator (LM317, LM7805, LM7812, etc) since they use the same method, but they include circuitry to make it easier. Using a BJT or Mosfet makes sense if your trying to regulate very high voltages (ie +100V) which is beyond the max. input voltage of most regulators.
Also note that linear regulators are very inefficient when the input voltage and output voltages have a large difference. The regulator drops the voltage by increasing the resistance across the BJT. When you pull current through the regulator it will disspates a lot of heat, because of the high internal resistance.
The circuit in at the provided web page is missing an output inductor which would help a great deal in reducing output noise.
FWIW: It would more practical to use a Linear regulator (LM317, LM7805, LM7812, etc) since they use the same method, but they include circuitry to make it easier. Using a BJT or Mosfet makes sense if your trying to regulate very high voltages (ie +100V) which is beyond the max. input voltage of most regulators.
Also note that linear regulators are very inefficient when the input voltage and output voltages have a large difference. The regulator drops the voltage by increasing the resistance across the BJT. When you pull current through the regulator it will disspates a lot of heat, because of the high internal resistance.
The circuit in at the provided web page is missing an output inductor which would help a great deal in reducing output noise.
That wouldn't regulate, it would just provide more smoothing at the expense of a larger voltage drop. You can "clamp" the output voltage of a cap multiplier (see Pass's), but I wouldn't call this regulation. The only good reason I can think to use a number of them in series is to drop an excessively high voltage and spread the wasted power between several transistors, but you'd need correspondingly bigger heatsinks for this to make a difference.
Cap multipliers are best if you don't need precise regulation and have a lot of overhead voltage to spare. If you need regulation, use a regulator.
Cap multipliers are best if you don't need precise regulation and have a lot of overhead voltage to spare. If you need regulation, use a regulator.
And here I was thinking nobody cared about the Twilight amplifier... several months down the road still my favorite design by far.
To answer the question (sort of) the MOSFETs work so well in the Twilight because there is effectively two of them stacked, on on top of the other, so the PSRR in total, form the point of view of the output, is extremely high, roughly 80 dB. [ok thats nothing on most op amps, but pretty decent in tubeland at any rate]
As TechGuy pointed out MOSFET capacitance multipliers aka line smoothers generally favor high voltage, low current applications, as you have to drop 10 V or so between the source and drain. (Less for the 2Sk213 family I used, but true as a general statement.)
The main difficulty with the circuit shown attached is that the output ripple is only as low as the ripple at the MOSFET gate. If the filter comprising the voltage divider R1, R2 and C2 is not effective, the output will be noisy no matter how high the transconductance of the MOSFET. In practice ripple rejection of more than 40 dB per stage will be increasingly difficult, more than 60 dB impossible as that's the all can squeeze from most MOSFETs anyway. An LM7812 will do about 73 dB...
As a compact, cheap, and clever way of dropping the line ripple in high voltage circuits by a factor of 100, however, I think they are great. For one thing, they don't generate a lot of noise, for another they are pretty good about suppressing higher frequency line noise, and finally they provide decent load regulation.
For low voltage circuits the equivalent quick'n'dirty vreg is the zener to the base of the BJT trick. Apparently that also sounds pretty decent.
To answer the question (sort of) the MOSFETs work so well in the Twilight because there is effectively two of them stacked, on on top of the other, so the PSRR in total, form the point of view of the output, is extremely high, roughly 80 dB. [ok thats nothing on most op amps, but pretty decent in tubeland at any rate]
As TechGuy pointed out MOSFET capacitance multipliers aka line smoothers generally favor high voltage, low current applications, as you have to drop 10 V or so between the source and drain. (Less for the 2Sk213 family I used, but true as a general statement.)
The main difficulty with the circuit shown attached is that the output ripple is only as low as the ripple at the MOSFET gate. If the filter comprising the voltage divider R1, R2 and C2 is not effective, the output will be noisy no matter how high the transconductance of the MOSFET. In practice ripple rejection of more than 40 dB per stage will be increasingly difficult, more than 60 dB impossible as that's the all can squeeze from most MOSFETs anyway. An LM7812 will do about 73 dB...
As a compact, cheap, and clever way of dropping the line ripple in high voltage circuits by a factor of 100, however, I think they are great. For one thing, they don't generate a lot of noise, for another they are pretty good about suppressing higher frequency line noise, and finally they provide decent load regulation.
For low voltage circuits the equivalent quick'n'dirty vreg is the zener to the base of the BJT trick. Apparently that also sounds pretty decent.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.