I want to predict the value of V_G on the linked schematic. Please help. I am a novice in DIY.
I am trying to build a very simple (fewer number of parts than in Mezmerize DCB1) DC coupled Pass B1 2 channel audio line level buffer with total 3 pairs of pair-matched 2SK170s. In the schematic, I simplified the 115/230VAC-to-18VDC power supply to one 18V battery. It may take a couple of minutes for the circuit to settle after turning the power on, but I can leave it constantly on.
Suppose the LOAD is a 10kOhm resistor.
If the voltage across the LOAD is 1V DC, then V_G is 0.5V DC, right?
If the voltage across the LOAD is 20Hz 1Vrms AC, then what is the value of V_G?
Last edited:
Actually there is a lot more circuit there than is needed.
You could easily take two 5K resistors in series from + to - and derive
a DC ground value where they join. Some capacitance to the rails will make
that impedance low at AC values, and since you are not using this to
amplify DC, it works like glue. If you are that worried about the DC
impedance, make the resistors a lower value. Two 1K resistors will each
dissipate about 1/10 watt and provide a 500 ohm impedance DC reference.
😎
You could easily take two 5K resistors in series from + to - and derive
a DC ground value where they join. Some capacitance to the rails will make
that impedance low at AC values, and since you are not using this to
amplify DC, it works like glue. If you are that worried about the DC
impedance, make the resistors a lower value. Two 1K resistors will each
dissipate about 1/10 watt and provide a 500 ohm impedance DC reference.
😎
Thanks for your insight.
In reality, the capacitor values are not precise. Instead of two 22000uF capacitors, I am likely to have something like 26400uF on + side and 17600uF – side. Will the two 1kOhm resisters you mentioned (without the active component - a pair of 2SK170s) make circuit settle to symmetric + and - voltage division?
Yes. Any mismatch of the Jfets (as seen by DC at their output) will alter the
balanced a bit, but it really is slight. I have circuits using 20 ohm trimpots
between the Jfet Sources to null their offset, and I can run 12 sets of buffers
using 4.7K resistors as dividers. There is almost no difference, and the PSRR
of the buffers is pretty high for this sort of thing, and 1,000 uF seems to
be enough.
So the answer is: Don't worry about it, but if you want perfection trim your
DC offset on each buffer.
😎
note: (complementary follower buffers - it's still the same)
balanced a bit, but it really is slight. I have circuits using 20 ohm trimpots
between the Jfet Sources to null their offset, and I can run 12 sets of buffers
using 4.7K resistors as dividers. There is almost no difference, and the PSRR
of the buffers is pretty high for this sort of thing, and 1,000 uF seems to
be enough.
So the answer is: Don't worry about it, but if you want perfection trim your
DC offset on each buffer.
😎
note: (complementary follower buffers - it's still the same)
What do you mean by V_G? The input to output offset of such a circuit depends on how closely the JFETs are matched. Such a circuit is used on many oscilloscopes, a dual JFET is commonly used.
The diodes are to protect the buffer amp from surge when we connect or disconnect input our output cable. Bose 901 EQ has higher value input and output series resistors, but still has those diodes.
V_G is the voltage across the resistor R_G in the schematic.
We can completely remove R_G (R_G = 0 Ohm) and "drive the ground" directly with BUF634 or a high-input-impedance high-output-current op amp. (Due to high capacitance value in the schematic (22000uF), even the high current BUF634 will run at max capability for a while after turning on the power.) In this case, however, the audio signal current though the load, the amplifier connected to B1 output, flows also through a chip with negative feedback (BUF634 or an OP amp).
I did not like it, because I wanted to fully realize the "zero feedback" concept of Pass B1. So I chose a very high 5kOhm for R_G, instead of typical 0 Ohm, and use B1 instead of a high-current OP amp. This makes driving the ground easy, i.e. low current, and decoupled from the audio B1 at audio frequencies. A down side is that the ground drive is very slow.
To keep my B1 buffer constantly on, I want power consumption to be as small as possible. The B1 itself consumes small amount of energy, but there will be energy loss in the transformer. So I am going to use very small 115/230VAC-to-28VAC transformer. Perhaps 0.5W or 1W transformer. To prevent transformer overload during turn on, I will choose very slow turn on. You see the 1 Ohm series resistor to the power supply in the original Pass B1 schematic. I will use about 1 ~ 5 kOhm series resistor at that place to limit the current during turn on.
I do not follow the conventional wisdom of "oversized transformer for sound quality," since the power consumption of B1 does not fluctuate much with the music signal input.
V_G is the voltage across the resistor R_G in the schematic.
We can completely remove R_G (R_G = 0 Ohm) and "drive the ground" directly with BUF634 or a high-input-impedance high-output-current op amp. (Due to high capacitance value in the schematic (22000uF), even the high current BUF634 will run at max capability for a while after turning on the power.) In this case, however, the audio signal current though the load, the amplifier connected to B1 output, flows also through a chip with negative feedback (BUF634 or an OP amp).
I did not like it, because I wanted to fully realize the "zero feedback" concept of Pass B1. So I chose a very high 5kOhm for R_G, instead of typical 0 Ohm, and use B1 instead of a high-current OP amp. This makes driving the ground easy, i.e. low current, and decoupled from the audio B1 at audio frequencies. A down side is that the ground drive is very slow.
To keep my B1 buffer constantly on, I want power consumption to be as small as possible. The B1 itself consumes small amount of energy, but there will be energy loss in the transformer. So I am going to use very small 115/230VAC-to-28VAC transformer. Perhaps 0.5W or 1W transformer. To prevent transformer overload during turn on, I will choose very slow turn on. You see the 1 Ohm series resistor to the power supply in the original Pass B1 schematic. I will use about 1 ~ 5 kOhm series resistor at that place to limit the current during turn on.
I do not follow the conventional wisdom of "oversized transformer for sound quality," since the power consumption of B1 does not fluctuate much with the music signal input.
Last edited:
As I understand from some posts here, the DC offset of Mezmerize DCB1 is also affected by the difference of | + supply voltage | and | - supply voltage |. So I want to divide the 18V in to precise half just as the Mezmerize + and - power supplies. Purchasing a 78L09 and a 79L09 does not seem to offer precise enough voltage division. You may get +9.3V from your 78L09 and -8.8V from your 79L09. With Mezmerize + and - power supplies, one can have +9.002V and -9.000V. I want similar precision from my ground drive power supply.
Last edited:
In my schematic, there are two voltage followers for audio and one voltage follower for ground driving.
Sorry that it is hard for me to understand that my schematic is like complementary follower buffers.
The ground drive voltage follower in my schematic is like the OP amp in unity gain configuration in Figure 7 in the following document.
http://www.ti.com/lit/an/sboa059/sboa059.pdf
The ground drive voltage follower in my schematic takes the role of BUF634 in Figure 6 in the following document.
http://www.ti.com/lit/ds/symlink/buf634.pdf
Sorry that it is hard for me to understand that my schematic is like complementary follower buffers.
The ground drive voltage follower in my schematic is like the OP amp in unity gain configuration in Figure 7 in the following document.
http://www.ti.com/lit/an/sboa059/sboa059.pdf
The ground drive voltage follower in my schematic takes the role of BUF634 in Figure 6 in the following document.
http://www.ti.com/lit/ds/symlink/buf634.pdf
Last edited:
I have to cancel this. Such a high value resistor does not work in DC supply due to voltage drop.
So resistor divider DC B1 works too. Fewer number of parts is good for a novice like me.
It has higher output impedance at 0 Hz than Mezmerize DC B1.
As for trimming, I think I have to have the same polarity DC offset on both L and R, and trim the average of L and R offsets. ("If your offset is +5mV you want to make the negative reg Vout stronger." - salasdcb1shuntreg-build-guide-v4.1)
Buy the way, can anyone tell me why the conventional chip regulator DC supply is not much discussed here for DC B1 buffer?
Since 78L09 (+9V DC regulator) and 79L09 (-9V DC regulator) chips are not expensive, we can to buy extra chips and pick a pair that gives the voltages we want, right? No, because the voltage from 78Lxx or 79Lxx drifts over time (after one year, two years, . . . . )?
Since 78L09 (+9V DC regulator) and 79L09 (-9V DC regulator) chips are not expensive, we can to buy extra chips and pick a pair that gives the voltages we want, right? No, because the voltage from 78Lxx or 79Lxx drifts over time (after one year, two years, . . . . )?
Regulation for this circuit is not crucial, and the switching supply is
regulated already. You need only filter it actively or passively for
noise.
😎
regulated already. You need only filter it actively or passively for
noise.
😎
Sorry for not understanding things quickly. I think I asked my question with incorrect words, and have lack of circuit analysis skills.
My concern is the DC values, not noise.
Could anyone estimate how much bigger or smaller the offset would be in the lower circuit? Assume we are using the same pair of JFETS (the identical pair, not other pair of the same part number) as in the above circuit.
Last edited:
I matched up a pair of MPF102 for Idss at 10V. The offset of the output with reference to the input went from 5mV in the upper (balanced) example to 14mV in the lower (unbalanced) example. This was the opposite of what I predicted based on typical JFET curves. My conclusion is that thermal drift was the dominate cause. In the unbalanced example the lower JFET is warmer because it has more voltage across it and the current is the same in both JFETs.
Thanks for posting your measurement result.
After seeing your result, I think that reducing output-DC-offset by plus-minus-power-supply-offset as described in Salas' instruction may not always work.
So I drew a new schematic, which might be called AC-coupled-input B1. Could you guys tell me (just by looking at it) whether it works? I often draw something that turns out to be nonsense.
I think input coupling capacitor degrades sound quality less than output coupling capacitor. So I put coupling capacitor only at the input.
The output DC offset is reduced by manually adjusting the input DC voltage. I hope the drift of the output DC over many years is negligible.
I chose 1uF for input coupling capacitor as in the original AC coupled Pass B1. I guess it results in sufficiently low cut-off frequency due to high input impedance of the JFETs.
After seeing your result, I think that reducing output-DC-offset by plus-minus-power-supply-offset as described in Salas' instruction may not always work.
So I drew a new schematic, which might be called AC-coupled-input B1. Could you guys tell me (just by looking at it) whether it works? I often draw something that turns out to be nonsense.
I think input coupling capacitor degrades sound quality less than output coupling capacitor. So I put coupling capacitor only at the input.
The output DC offset is reduced by manually adjusting the input DC voltage. I hope the drift of the output DC over many years is negligible.
I chose 1uF for input coupling capacitor as in the original AC coupled Pass B1. I guess it results in sufficiently low cut-off frequency due to high input impedance of the JFETs.
- Status
- Not open for further replies.