I used op27 everywhere. This chip has a dc offset nulling facility. It seems to make a very nice improvement to null each chip independently, staring with the input.
If you choose to build this, I recommend trying a surface mount chip. The SMD has the advantage that the holes for the sockets can be used for power supply bypassing. Also, getting to the nulling ports and output is far simpler.
If you choose to build this, I recommend trying a surface mount chip. The SMD has the advantage that the holes for the sockets can be used for power supply bypassing. Also, getting to the nulling ports and output is far simpler.
Only C4, C7 and C8 are left on the power supply board. Moving the PS takeoff to the force/sense point of the rails (R5) *may* have made that possible.
A 3 mA CCS has been added to each op amp output from the negative rail, and a 100 uF cap has been added from the output of the pre-regulator (LM 317/337) to the ground of the shunt transistor.
It's VERY quiet.
A 3 mA CCS has been added to each op amp output from the negative rail, and a 100 uF cap has been added from the output of the pre-regulator (LM 317/337) to the ground of the shunt transistor.
It's VERY quiet.
Last edited:
Sorry I was thinking about the shunt power supply. It is 27 ohm.
At this point, I have removed caps c5/c6, c9/c10, c11,c12. Kept c1/c2, c3/c4 and c7/c8.
I have a suspicion that the power transformer 229D24 ought to be 229D30, since 100mA is not being delivered by the LM317 -- only 87.5 mA. (1.05 volts over R1 )
LTspice simulations do not give 15 volts with input voltages of less than 16.8 volts to the LM317. With voltages around 20, it is possible that the 27 ohm resistor might be removed completely. Removing R3 in real life results in a rail voltage of just 2 volts. (real life input to the CCS is 16.2 volts)
This should not matter, as in my rig, in real life, 66 mA is being shunted to ground at idle. That is plenty.
Bear in mind though, I use a dual mono implementation with the opamps biased into class A. A higher voltage transformer may improve a stereo version, if the current sources can deliver 100 ma ( or more ).
At this point, I have removed caps c5/c6, c9/c10, c11,c12. Kept c1/c2, c3/c4 and c7/c8.
I have a suspicion that the power transformer 229D24 ought to be 229D30, since 100mA is not being delivered by the LM317 -- only 87.5 mA. (1.05 volts over R1 )
LTspice simulations do not give 15 volts with input voltages of less than 16.8 volts to the LM317. With voltages around 20, it is possible that the 27 ohm resistor might be removed completely. Removing R3 in real life results in a rail voltage of just 2 volts. (real life input to the CCS is 16.2 volts)
This should not matter, as in my rig, in real life, 66 mA is being shunted to ground at idle. That is plenty.
Bear in mind though, I use a dual mono implementation with the opamps biased into class A. A higher voltage transformer may improve a stereo version, if the current sources can deliver 100 ma ( or more ).
Last edited:
Finally got LTspice to work properly.... C3 is a problem for LTspice. It has been removed, and a 10 uF cap has been placed across the adjustment pin to ground. This may be changed to something much smaller, but it works for now.
1.23 volts is now measured across the CCS setting resistor -- seems like the regulator is working.
1.23 volts is now measured across the CCS setting resistor -- seems like the regulator is working.
It also works, and sounds better, (IMO) with the 10 uf cap removed. C4 is necessary according to Walt Jung.
http://waltjung.org/PDFs/Sources_101_P2.pdf
I also replaced R3 (27 ohm) with a 10 ohm resistor to better match the LT spice model (difficulty converging with 27 ohms).
http://waltjung.org/PDFs/Sources_101_P2.pdf
I also replaced R3 (27 ohm) with a 10 ohm resistor to better match the LT spice model (difficulty converging with 27 ohms).
Last edited:
The oscilloscope says different..... To get current oscillation under control, I went to a different transformer. Now running a single 216 VA 2x18 instead of a pair of 229R24. The 229R24s ought to be enough, but this works..... The 'scope says that under static load, the oscillation is gone. Also,back to the 27 ohm PRP.
The biggest frustration is that LTspice simulates differently from real life... will test the dynamic performance on the scope.
The biggest frustration is that LTspice simulates differently from real life... will test the dynamic performance on the scope.
Last edited: