Hi,
You could use a P-Channel JFET instead of Q3, but I don´t see a advantage.
The bipolar is cheaper and closer tolerated than a JFET.
jauu
Calvin
You could use a P-Channel JFET instead of Q3, but I don´t see a advantage.
The bipolar is cheaper and closer tolerated than a JFET.
jauu
Calvin
Just thinking if the servo is in line of signal then fet will have a lower intrinsic noise . But this is a baseless idea of mine. If the servo has nothing to do with signal ....
Calvin,
In post #15
What is the role of L2,L3 and C8 and C9 in buffer?
Also what is Rs1 does in circuit?
In post #15
What is the role of L2,L3 and C8 and C9 in buffer?
Also what is Rs1 does in circuit?
L1 is the output inductor, with a damping resistor added.
L2 & L3 are in the PSU and give some interference attenuation.
C8 & C9 are in the PSU and give some interference attenuation.
Rs1 is part of an output Zobel.
L2 & L3 are in the PSU and give some interference attenuation.
C8 & C9 are in the PSU and give some interference attenuation.
Rs1 is part of an output Zobel.
Thanks Andrew. What type are the Caps used / recommended in the circuit #15?
Is any of them electrolytic ?
Is any of them electrolytic ?
Hi,
C8 and C9 are polars. Use decent low-esr types.
Recently I tried polymere types and It seems that they improved the sound outcome, but don't nail me fix on that one 😎
C7 should be a bipolar type as the OPAmp can swing a couple of Volts into the negative.
L2 and L3 are inductances for RF interference attenuation.
As I like to use SMPSs and/or dc-dc converters for the supplies, some means of post-filtering are mandatory.
If the rails are clean, You may omit with the Ls, but it won't hurt at all to leave them in.
Rs1 is not part of a Zobel filter but just an element of the 'attached' load, i.e it represents cable- and connector resistances.
CL1 represents the load capacitance lumped together of cable capacitance and input capacitance of the following device.
RL1 then represents the input resitance of the following device.
jauu
Calvin
C8 and C9 are polars. Use decent low-esr types.
Recently I tried polymere types and It seems that they improved the sound outcome, but don't nail me fix on that one 😎
C7 should be a bipolar type as the OPAmp can swing a couple of Volts into the negative.
L2 and L3 are inductances for RF interference attenuation.
As I like to use SMPSs and/or dc-dc converters for the supplies, some means of post-filtering are mandatory.
If the rails are clean, You may omit with the Ls, but it won't hurt at all to leave them in.
Rs1 is not part of a Zobel filter but just an element of the 'attached' load, i.e it represents cable- and connector resistances.
CL1 represents the load capacitance lumped together of cable capacitance and input capacitance of the following device.
RL1 then represents the input resitance of the following device.
jauu
Calvin
Thanks Calvin.How does this look ?
With the new setup I am getting an offset of 13mv between in and out.Is it accepatable?
With the new setup I am getting an offset of 13mv between in and out.Is it accepatable?
Attachments
Last edited:
without the DC servo the output offset should be set up to be <<13mVdc
with a DC servo the output offset should be less than 1mVdc.
with a DC servo the output offset should be less than 1mVdc.
Hi,
to choose a 10k Volume control poti is reasonable but check that the bandwidth limit of the Highpass formed by the 10uF cap and the poti remains sufficiently low.
Choose more common values for the emitter and the source resistances.
The closest values to 8R would be 7.5R and 8.2R, resp for 4R it are 3.9R or 4.7R.
Around 4R will lead to quite high idle currents --> check for sufficient cooling of the bipolars.
You connected both inputs of the servo OPAmp to gnd .... why??
Doesn't make sense to me.
The only time You'd do this is when You tune the 3k resistor of the bipolar servo transistor, which btw ihas to be a pnp and not a npn.
Since You designed both source resistors to be the same 4R, there's no sense in using a servo anyway.
In that case simply build the Calvin buffer with offset trimming by a Poti paralleled to the lower halfs JFET-Drain resistor and omit the servo alltogether.
I wonder though, when You seemingly haven't understood the circuit in fullness, why don't You stick with the given values first and begin fudging and tuning after a successful built of the first?
jauu
Calvin
to choose a 10k Volume control poti is reasonable but check that the bandwidth limit of the Highpass formed by the 10uF cap and the poti remains sufficiently low.
Choose more common values for the emitter and the source resistances.
The closest values to 8R would be 7.5R and 8.2R, resp for 4R it are 3.9R or 4.7R.
Around 4R will lead to quite high idle currents --> check for sufficient cooling of the bipolars.
You connected both inputs of the servo OPAmp to gnd .... why??
Doesn't make sense to me.
The only time You'd do this is when You tune the 3k resistor of the bipolar servo transistor, which btw ihas to be a pnp and not a npn.
Since You designed both source resistors to be the same 4R, there's no sense in using a servo anyway.
In that case simply build the Calvin buffer with offset trimming by a Poti paralleled to the lower halfs JFET-Drain resistor and omit the servo alltogether.
I wonder though, when You seemingly haven't understood the circuit in fullness, why don't You stick with the given values first and begin fudging and tuning after a successful built of the first?
jauu
Calvin
The lowest I get is
DC component:-2.99094e-008
That is with R20 in post #15 with 1.5K and PNP Q2SB1427.
Buffer current is 53ma and current through R20 is 10ma.
Total Harmonic Distortion: 0.000033%.
Also what are these spice directives used for
.MEAS NOISE total_rms_inpuput_noise INTEG V(inoise)
DC component:-2.99094e-008
That is with R20 in post #15 with 1.5K and PNP Q2SB1427.
Buffer current is 53ma and current through R20 is 10ma.
Total Harmonic Distortion: 0.000033%.
Also what are these spice directives used for
.MEAS NOISE total_rms_inpuput_noise INTEG V(inoise)
Last edited:
Hi,
With typical mutimeters You won´t be able to reliably measure down to 1mV or even less.
So don´t worry too much about to get the lowest figures.
Just keep an eye if teh offset remains low over time (after heating up).
If You now strg+click on the V(onoise) or V(inoise) label a display opens giving You the values of lower and upper bandwidth limit and the total rms noise figure over that bandwidth.
The above script also calaculates the total rms noise figure and writes it into the Spice Error Log (which You can access via a rightclick/view/SpiceErrorLog on the results display (*.raw)
jauu
Calvin
Everything under ~10mV of output offset is reasonably low.
With typical mutimeters You won´t be able to reliably measure down to 1mV or even less.
So don´t worry too much about to get the lowest figures.
Just keep an eye if teh offset remains low over time (after heating up).
If You run a noise sim and probe the output the diagram shows the noise density value.
If You now strg+click on the V(onoise) or V(inoise) label a display opens giving You the values of lower and upper bandwidth limit and the total rms noise figure over that bandwidth.
The above script also calaculates the total rms noise figure and writes it into the Spice Error Log (which You can access via a rightclick/view/SpiceErrorLog on the results display (*.raw)
jauu
Calvin
Any 4391 model would work.
http://web.rfoe.net:8000/ziliaoxiazai/PHILIPS/models/spicespar/data/pmbf4391.html
Jan
Last edited:
I can't find the DC input to the servo. Possibly swapped input and ground on the RC.
Jan
- Status
- Not open for further replies.