RCruz said:
Hi ikoflexer
So the initial ExpressPCB is also wrong ?
(I am starting to like this software)
Ricardo
Hey Ricardo,
Yes, all the pcb related to my initial post about the regulator should have D and G reversed on the mosfets. I'm awfully sorry about that.
I agree, expresspcb is a nice piece of software, and it's free. The nice feature where you can import the netlist of a schematic from ltspice (or other software) is particularly nice, BUT that can also be a source of failure. The pin numbering in ltspice is often different than the pin numbering in expresspcb. That's how D and G were reversed on the first pcb.
I'm inclined to just post links to my website from now on, that way I can keep in one place the latest version of everything, with fixes and all that. I'd feel awful if someone got and implemented the broken version.
Pfft, man, what a frustrating thing. ExpressPCB uses on most components square pads for positive pins, and round pads for negative pins... except for LEDs. I just realized that now, and made the minor change once again. I'm over and done with the regulator pcb for now, I tested the LED version as well and it works as per this attachment. Tested with both bf245a and sk170 jfets; my scope does not show a difference between the two. Perhaps there is sound wise, but I don't have phono circuit done yet. Nice to see the filtered input trace swing up when some power device in the house turns on and off in the house, but the regulated does not move. Neato!
Contents of sreg10.zip
sreg10bottom.pdf
sreg10silkscreen.pdf
salasshunt10.pcb
Contents of sreg10.zip
sreg10bottom.pdf
sreg10silkscreen.pdf
salasshunt10.pcb
ikoflexer said:
We must clarify the alternatives for the CCS Voltage reference and their configuration here:
In Ikoflexer's board you can either use 3X 1.7V LEDS or a resistor (R10) voltage drop governed by the JFET on its tail.
a) If using the LEDS, at least a 6mA IDSS JFET (J22) must be used so to keep them in their low noise region. With 6R8 R1 the CCS will run at about 200mA. A recommended value.
b) If using a 1600R R10, you better use a 3mA IDSS JFET (J1). That will give again about 200mA with a 6R8 R1, or 140mA with a 10R R1 as per original Ikoflexer's value. More IDSS will magnify the resistor voltage drop, and its positive difference from VGS (about 3.4V). That difference, divided by R1 sets the running current. A 6 mA IDSS will skyrocket the CCS to 0.9A for 6R8 R1 or 0.62A for 10R. So be careful! You can downsize R10 if you are left with some 2SK170BL between 6.5-11mA IDSS after matching for the phono, and you want to use them in the shunt, which is the most probable scenario.
In this case, say you have a 7mA 2SK170, use 680R for R10, and your CCS will run at 200mA with 6R8 R1.
ikoflexer said:
Don't expect to see such noise subtleties on o scope. It takes an FFT with well sorted out loop gnd and -120 to -130dB noise floor so to see what version rises higher as the frequency goes up. The LEDS need 5-10mA to be fair with their noise capabilities. I use my strongest IDSS matching left over 2SK170BL always there.
As for what you saw for low line noise and transient susceptibility, its something I forgot to mention I think, but yes, it works much better for isolation than the (series & feedback) LM317 that I earlier had.
es44 said:I'm finally finished with my first attempt
Did the 36dB version, but need a little more gain, so next one will be the 46dB. It seems i can just lift the voltage 4 volts, and change a few resistors. Should be easy. ( well, at least for some, i always find a way to mess up, hehe )
Thanks for making this possible Salas.
An externally hosted image should be here but it was not working when we last tested it.
You can even bypass R1 (the 100R resistor under first stage JFET) with a 1000uF Panasonic FC or Elna Silmic or Black Gate NX 6.3V and it will become 41dB gain without further changes.
if you want to use two shunts then you'll get the best out of them by having dual secondaries. This can be separate transformers one to each channel or a dual secondary with an isolated winding to each channel.
If you are building a dual polarity version then you need dual polarity supplies. This will require four isolated secondary windings.
It is normally recommended that Vac ~= regulated Vdc.
However these shunt regulators are approaching low voltage drop out (LDO) performance, so we can manage with a bit less on the AC side.
For 33Vdc use either 28Vac or 30Vac and with dual rectifiers use 28 or 30-0, 28 or 30-0
If you are building a dual polarity version then you need dual polarity supplies. This will require four isolated secondary windings.
It is normally recommended that Vac ~= regulated Vdc.
However these shunt regulators are approaching low voltage drop out (LDO) performance, so we can manage with a bit less on the AC side.
For 33Vdc use either 28Vac or 30Vac and with dual rectifiers use 28 or 30-0, 28 or 30-0
