Since i have shiit eq and i like the sound, i was curious about the circuit Thorsten posted in post #18. I built barebones circuit, without eq so far, just input pre and follower second stage. I used j113 jfet with bc560, those i have plenty.
At ~1V rms it has 0.007% distortion, so i consider its working.
I am using +/_ 42 volts as power supply, just because that's what i have. I do see some 120Hz background buzz, its like -70dB, not sure why, its swithching supply.
At ~1V rms it has 0.007% distortion, so i consider its working.
I am using +/_ 42 volts as power supply, just because that's what i have. I do see some 120Hz background buzz, its like -70dB, not sure why, its swithching supply.
Last edited by a moderator:
adason,
I did something similar. Inspired by the Schiit EQ, I prototyped a similar circuit using the filter setpoints defined in the Cello EQ. I simulated the circuit first in Circuit Maker software and then designed and fabbed circuit boards. After getting the PCBs I fully populated one with parts but never once powered it up to test or listen to it. Instead my time was consumed with my wife packing and moving our household from California to Europe. Shift forward to today, my wife and I are living in Europe, but guess what, we are about to again. This time to we are moving to where we expect to make our home. And I still haven't powered up that prototype PCB.
In any case this is what I crafted. I wonder how well it works ...
I did something similar. Inspired by the Schiit EQ, I prototyped a similar circuit using the filter setpoints defined in the Cello EQ. I simulated the circuit first in Circuit Maker software and then designed and fabbed circuit boards. After getting the PCBs I fully populated one with parts but never once powered it up to test or listen to it. Instead my time was consumed with my wife packing and moving our household from California to Europe. Shift forward to today, my wife and I are living in Europe, but guess what, we are about to again. This time to we are moving to where we expect to make our home. And I still haven't powered up that prototype PCB.
In any case this is what I crafted. I wonder how well it works ...
"What is Active volume control?"
That's a 'Baxandall active volume control' as interpreted by Doug Self. active volume control tutorial
That's a 'Baxandall active volume control' as interpreted by Doug Self. active volume control tutorial
I use Sziklai type structure in many applications and commercial products. They are very similar. So there is no reason for it not to work.
The noise being 120Hz makes it clear it is from the PSU.
Simple circuits like the one I showed usually have rather low supply rejection (only a few dB, so most of the power supply noise appears on the output). Unlike high open loop gain, high Feedback circuits.
IIRC, Goldmund also includes regulators for the frontend. So the fact that this circuit has noise and Goldmund has not, doesn't mean either power supply voltage or noise are acceptable in my circuit.
I suggest you add a pair of resistors into the power supply lines to burn off appx. 8V and add a big cap to ground from each PSU.
The circuit should draw around 20mA, so use 390 Ohm in either supply line. If we add a 3,300uF cap after that we have around 0.12Hz turnover and 60dB noise reduction at 120dB.
It will take around a minute for this circuit to be 100% stable.
Thor
Of course it has a gain below one.
As you omitted the 3.6k resistor to ground it must have a gain of 0.5.
What do you expect the circuit to do if you designed and build it for a gain of 0.5, except to have a gain of 0.5?
The "east German" EQ will behave pretty much identical.
I am not sure you actually understand the circuit, how it works and what to expect. This should t a prerequisite to tinkering with it.
This circuit has no power supply rejection and gain is depending on the resistor values in the supply lines. That is the price paid for such simplicity.
It's not an Op-amp circuit.
Thor
Naturally, the circuit is designed for unity gain.
And it uses pretty much no feedback, so distortion is tube like, monotonic and not particularly low (low distortion is not a valid design goal, inaudible distortion is).
I suggest you first start by understanding how the circuits operate and what to expect from them.
But you are right, don't bother with this circuit anymore. Just build something that doesn't require thinking or understanding.
Thor
This in not an op-amp circuit. By reducing the supply voltage you reduce the current in the devices but you retain the load the same. This means you use a greater part of the available current swing for the same signal and distortion rises.
You really need to learn about basic electronics. These are very, very simple and basic circuits, in a lot of ways they are what you learn basic electronics with.
Incidentally, you could swap BC560 for something like BD140 and up the current in the circuit, by reducing the resistors that set the current and lower distortion this way, if you find it bothersome.
Thor
Last edited:
Do you understand how the circuit works?
The Sziklai J-Fet + PNP runs at appx 11mA, of that around 0.7mA pass through the J-Fet.
This gives around 3mA/V transconductance for the J-Fet. The effective drain load is around 400 Ohm (1k in parall with the base impedance of the BJT). So gain at the BJT base is around unity.
All actual transconductance comes from the BJT, the compound unit has about 200mA/V transconductance. This implies an effective source impedance of around 5Ohm. The Drain is very high impedance, it is near a perfect current source.
The combination of all resistors on the source to AC ground form the current setting resistance for the voltage to current conversion performed by the circuit.
That is, the 3.6k to -40V (which should be free of noise) mainly sets the DC operating condition and is part of the AC circuit. The 3.6k to ground, which you designed out of the circuit, also forms part of the AC circuit.
Together these two resistors in parallel are an AC impedance of 1.8kOhm. if we apply 1.8V peak-peak to to the gate, the current through the Sziklai compound device will be modulated by 1.8V / 1.8kOhm = 1mA peak-peak. If you do not add the second 3.6k resistor to ground, this current is halved.
Actually the appx 5Ohm source impedance adds a small Error, so we only see 0.997 mA current modulation. Further, as this Impedance is dependend on the current in the device, which we are changing, we see nonlinear distortion as consequence.
Now our 11mA DC with a superimposed 0.997mA peak-peak current modulation flow from the negative rail, through the resistors on the source and the device and then exits from the drain and flows via a 1.8kOhm resistor to positive rail.
The current flowing will drop appx. 19.8V DC across the 1.8kOhm resistor, with a Voltage modulation of 0.997mA * 1.8kOhm = 1.794V peak-peak.
So when correctly assembled, as originally designed, the circuit has a gain just a touch below unity, at -0.025dB. The output buffer is added to make sure the circuit remains unaffected by the load on the output.
There a few other lessons here.
The circuits output is referenced to the positive rail, not "ground" (anyway ground isn't), so any noise on the positive rail is added to the signal.
The circuit's input is referenced to the negative rail.
Any noise on the negative rail will be added to the input signal.
Current through the circuit is directly proportional to the power supply voltage, it only works in a meaningful way with a noise free powersupply of the correct value.
If you want to change anything, you need to change the design accordingly.
So now you hopefully understand the circuit a little better, you can understand what behaviour to expect and why your random changes and experiments had the results they had.
While I have not build this specific circuit, I have the basic building blocks in many commercial products.
I have absolute confidence that if you assemble the circuit correctly and give it a noise free power supply, it will work fine and just as designed.
Will it meet your expectations in terms of performance etc.? That depends on your expectations.
Thor
The Sziklai J-Fet + PNP runs at appx 11mA, of that around 0.7mA pass through the J-Fet.
This gives around 3mA/V transconductance for the J-Fet. The effective drain load is around 400 Ohm (1k in parall with the base impedance of the BJT). So gain at the BJT base is around unity.
All actual transconductance comes from the BJT, the compound unit has about 200mA/V transconductance. This implies an effective source impedance of around 5Ohm. The Drain is very high impedance, it is near a perfect current source.
The combination of all resistors on the source to AC ground form the current setting resistance for the voltage to current conversion performed by the circuit.
That is, the 3.6k to -40V (which should be free of noise) mainly sets the DC operating condition and is part of the AC circuit. The 3.6k to ground, which you designed out of the circuit, also forms part of the AC circuit.
Together these two resistors in parallel are an AC impedance of 1.8kOhm. if we apply 1.8V peak-peak to to the gate, the current through the Sziklai compound device will be modulated by 1.8V / 1.8kOhm = 1mA peak-peak. If you do not add the second 3.6k resistor to ground, this current is halved.
Actually the appx 5Ohm source impedance adds a small Error, so we only see 0.997 mA current modulation. Further, as this Impedance is dependend on the current in the device, which we are changing, we see nonlinear distortion as consequence.
Now our 11mA DC with a superimposed 0.997mA peak-peak current modulation flow from the negative rail, through the resistors on the source and the device and then exits from the drain and flows via a 1.8kOhm resistor to positive rail.
The current flowing will drop appx. 19.8V DC across the 1.8kOhm resistor, with a Voltage modulation of 0.997mA * 1.8kOhm = 1.794V peak-peak.
So when correctly assembled, as originally designed, the circuit has a gain just a touch below unity, at -0.025dB. The output buffer is added to make sure the circuit remains unaffected by the load on the output.
There a few other lessons here.
The circuits output is referenced to the positive rail, not "ground" (anyway ground isn't), so any noise on the positive rail is added to the signal.
The circuit's input is referenced to the negative rail.
Any noise on the negative rail will be added to the input signal.
Current through the circuit is directly proportional to the power supply voltage, it only works in a meaningful way with a noise free powersupply of the correct value.
If you want to change anything, you need to change the design accordingly.
So now you hopefully understand the circuit a little better, you can understand what behaviour to expect and why your random changes and experiments had the results they had.
While I have not build this specific circuit, I have the basic building blocks in many commercial products.
I have absolute confidence that if you assemble the circuit correctly and give it a noise free power supply, it will work fine and just as designed.
Will it meet your expectations in terms of performance etc.? That depends on your expectations.
Thor
Hi Thorsten, Happy New Year!
Thanks for detailed explanation. I appreciate it.
I added massive, trully massive, filter CRCRCRC after the switching power supply, as much as it could handle. I kept adding 100r 15000uF till power supply refused to start. I think i got eight caps now. Anyway, that took care of poor psrr and 120Hz component is -80dB.
Power supply voltage after filter offcourse dropped, now its about +/_30Vdc. Distortion increased modestly.
Adding 100r 470uF in postive rail between stages had little or negative effect, as a matter of fact, some upper harmonics showed up.
Thanks for detailed explanation. I appreciate it.
I added massive, trully massive, filter CRCRCRC after the switching power supply, as much as it could handle. I kept adding 100r 15000uF till power supply refused to start. I think i got eight caps now. Anyway, that took care of poor psrr and 120Hz component is -80dB.
Power supply voltage after filter offcourse dropped, now its about +/_30Vdc. Distortion increased modestly.
Adding 100r 470uF in postive rail between stages had little or negative effect, as a matter of fact, some upper harmonics showed up.