Here are the .asc files, even better then PDF.
Hans
.
P.s. in the .Tran simulation you will have to let all caps settle for roughly 50 to 100 msec.
.
Hans
.
P.s. in the .Tran simulation you will have to let all caps settle for roughly 50 to 100 msec.
.
Last edited:
Here you are with a cleaned-up version for the SE version, the one that Ricardo has planned to build.
It also included the 10R trimpot//20R and a 100pF cap at the input.
Is this what you asked for ?
Hans
.
Joachim,
I just noticed that the used source resistances so far where 2*3.9R instead of the 3.3R that I thought.
That means that R18, the 20R resistor, has to be 35R instead to get the same 7.8R as in the simulations.
Hans
I just noticed that the used source resistances so far where 2*3.9R instead of the 3.3R that I thought.
That means that R18, the 20R resistor, has to be 35R instead to get the same 7.8R as in the simulations.
Hans
In simulation the smallest unbalance in the 10r trimmer // 25ohm creates a large offset in the OUT+ OUT- points and the circuit does not amplify.
This trimmer is critical... do you believe there will be no issues due to thermals once the build is completed ?
I mean, once we get the trimmer value right, will it keep it's value with time/temp ?
This trimmer is critical... do you believe there will be no issues due to thermals once the build is completed ?
I mean, once we get the trimmer value right, will it keep it's value with time/temp ?
The two op-amps U3&U5 are keeping the voltages equal on the two 400R drain resistances.
But their output voltages can be anything depending on differences in the Fets.
That's why the 10R trimpot came into the game.
But with a DC gain of over 10.000, the thermal stability also has to be considered, that's why I added a servo to cope with that aspect.
Initiating procedure should now be:
1) turn the trimpot without servo connected to get circa zero volt on Vout.
2) then connect the servo and all is set.
I have tested this with two Fets on one side to get a large difference between the two legs, and it worked perfectly.
Hans
.
But their output voltages can be anything depending on differences in the Fets.
That's why the 10R trimpot came into the game.
But with a DC gain of over 10.000, the thermal stability also has to be considered, that's why I added a servo to cope with that aspect.
Initiating procedure should now be:
1) turn the trimpot without servo connected to get circa zero volt on Vout.
2) then connect the servo and all is set.
I have tested this with two Fets on one side to get a large difference between the two legs, and it worked perfectly.
Hans
.
So after all we need a servo.
It does work and in simulation it also creates a bass rollof below 10Hz... nice.
What type of opamp should we use for the servo ?
It does work and in simulation it also creates a bass rollof below 10Hz... nice.
What type of opamp should we use for the servo ?
I would like to reduce the value of C1 (470n) and scale R1 R2 R24 accordingly because it will be easier for me to trim the RIAA curve if the resistors are larger.
Will I loose the splendid noise performance this way ?
Will I loose the splendid noise performance this way ?
Why will setting the Riaa be easier with larger resistor values ?
470nF is about where going larger makes hardly a difference, but going to 100nF will change EIN fron 0.24nV/rtHz to 0.31nV/rtHz.
But I just found that the servo ruins the noise, so let me find out why.
Hans
470nF is about where going larger makes hardly a difference, but going to 100nF will change EIN fron 0.24nV/rtHz to 0.31nV/rtHz.
But I just found that the servo ruins the noise, so let me find out why.
Hans
The servo affected the Cart's input voltage through the huge 300pF Fet capacitances, that's why a few resistors had to be changed.
Two 220K input resistors R4 and R5 went to 100K, a value that can't hurt in any way to my feeling, the 1Meg servo input resistor R19 went to 10K, causing a bass roll off at 10Hz, and the servo's output resistor R20 went from 10K to 220K.
Offset range that the servo can correct with R20 is +/-3.5Volt at Vout without servo, probably more than enough after having set Vout with the 10R pot without servo to ca. 0Volt.
Noise is safely back to the 0.24nV/rtHz it had without servo.
Hans
.
Two 220K input resistors R4 and R5 went to 100K, a value that can't hurt in any way to my feeling, the 1Meg servo input resistor R19 went to 10K, causing a bass roll off at 10Hz, and the servo's output resistor R20 went from 10K to 220K.
Offset range that the servo can correct with R20 is +/-3.5Volt at Vout without servo, probably more than enough after having set Vout with the 10R pot without servo to ca. 0Volt.
Noise is safely back to the 0.24nV/rtHz it had without servo.
Hans
.
Last edited:
The output impedance of Q1 can be as high as 300K, that's what I overlooked in the previous servo version.
Since the servo's output resistance R20 is in parallel to Q1, I would advise to make R20 as large as possible after having assembled the amp and having measured how much Vout drifts.
This drifting can be easily seen by monitoring the output of U6.
When it's only moves within a few Volts, R20 can be easily increased to a 470K or even to 1Meg, but U6 should not exceed +/-10Volt.
Well, that's it for today.
Hans
Since the servo's output resistance R20 is in parallel to Q1, I would advise to make R20 as large as possible after having assembled the amp and having measured how much Vout drifts.
This drifting can be easily seen by monitoring the output of U6.
When it's only moves within a few Volts, R20 can be easily increased to a 470K or even to 1Meg, but U6 should not exceed +/-10Volt.
Well, that's it for today.
Hans
Last edited:
Joachim,
Since you mentioned a bipolar version, I had already made a bipolar version for myself based on the same design topology but with halve the current in the LTP.
This bipolar version has exactly the same noise and gain specs as the IF3602.
So just for the record, here it is.
Hans
.
Since you mentioned a bipolar version, I had already made a bipolar version for myself based on the same design topology but with halve the current in the LTP.
This bipolar version has exactly the same noise and gain specs as the IF3602.
So just for the record, here it is.
Hans
.
That is the poor mans version (;
The BJT I build is the older idea without CCS and a offset voltage over the positive inputs of the OPamps.
The BJT I build is the older idea without CCS and a offset voltage over the positive inputs of the OPamps.