A Jfet sourced Norton reference to test has been incidentally discussesed a few pages before. Thanks for your input and schematic. Going 15-30% of a Jfet's Idss with a resistor between its G,S so to find its best point can be done with the trimmer and then change the main Vref resistor instead in a preditermined application. Tends to bigger Rref values and that is more Johnson and impedance to easier pick up hum on the other hand. Fully adjusting a steady properly sized main Vdrop is tempting use in our reg for the trimmer non the less. To lift Vbe with leds or diodes so to give space for 2xVt I would utilize not for a J2xx type but for a much larger Vt one because that parameter makes them better for noise in a Norton. It remains to be seen if the Jfet idea will hold better in the typical reg's working conditions than the pnp Vbe referenced trimmer feeding the underneath bjt Ref source used till now in practice. I hope so, because its less parts, cleaner, no Ib currents to settle.
I missed it for sure
Dear Salas.
I apologize for not reading the whole thread thoroughly.
Nevertheless the idea of using current reference, instead of the voltage one kept bugging me - so I brought my own version to others' attention.
Jfet depicted at the drawing was used just for clarity
It does work, at least with LM334 and the components you see: at the moment from regulated power supply.
If you'd like I could dig out old circuit, utilizing voltage reference in shunt regulator (if I only can bring to life old hard drive).
Some secondary means of the load protection would be reasonable to introduce: like Zener triggered thyristor, blowing the fuse in case of regulator failure. Like overcurrent protection in series regulator...
In respect to hum and noise - your curcuit is feeding ripples directly to the shunt BJT's base-emitter junction through quite large capacitor.
Alex
Dear Salas.
I apologize for not reading the whole thread thoroughly.
Nevertheless the idea of using current reference, instead of the voltage one kept bugging me - so I brought my own version to others' attention.
Jfet depicted at the drawing was used just for clarity
It does work, at least with LM334 and the components you see: at the moment from regulated power supply.
If you'd like I could dig out old circuit, utilizing voltage reference in shunt regulator (if I only can bring to life old hard drive).
Some secondary means of the load protection would be reasonable to introduce: like Zener triggered thyristor, blowing the fuse in case of regulator failure. Like overcurrent protection in series regulator...
In respect to hum and noise - your curcuit is feeding ripples directly to the shunt BJT's base-emitter junction through quite large capacitor.
Alex
Alex, I've done this too, at times. It's hard to catch all the details discussed in a thread of tens of pages. Indeed we've considered about every aspect of vref and other regulator issue here and in two other regulator threads. But don't get discouraged, I think every input based on good intentions is valuable. I personally bugged poor salas
with my aberrations ... quite a lot. Alex, I've had this worry too. Recently one member's regulator broke and Vin went directly to Vout. Luckily for him nothing was damaged, but this made more clear for me that protection is good to have. Of course, this would add a complication to the circuit, but it's nice to at least have the option.
Its ripple rejection is better than its output impedance spec actually. But that would be compromized by long cables typical in tube builds anyway. Force/sense could be beneficial. Without 1uF the ref's impedance rises at lows and then the hum can be picked up easier. If you look at some Gary Pimm's measurements he was getting better results the higher the bypass cap over his Nortons. We could not see AC on this one's output. We prefilter always non the less, as it was intended to be done. Does an LM334 CCS have the higher frequency performance of the Idss Jfet's?
Anyway, if you have an alternative CCT with LM CCS and comparable component count and same architecture more or less, I am sure that people would like to build it too. I encourage you to investigate and contribute.
Regards
Looks nice indeed!
Rather economic between two boards, there's the regs connections, but tubes are big parts, put out heat, need space, add the pcb foil length to anodes. The average good tube reg can show 15R-25R ouput impedance AT its output pins. Where would you estimate this construction? I wanna gauge why they still perceive it as robust in comparison with tube regs. When its very basic.
Very good! And you used the new Jfet idea or I see two little transistors there somewhere?
Thank! I try to change Jfet fot BC560, but the supply is not good, noise many than use 2 BC560.
I dont know why?
And did you cure some hum?
There are a litte hum, litte noise, The sound is good!
Yes. when I change Bc560 by 2SK170 to my board, I show that,!
noise, and the sound is not better !
Attachments
Last edited: