Sonic Wonder
Member of the Trade: Goldpoint Level Controls
- Joined
- Dec 6, 2010
- Posts
- 44
- Likes
- 12
.
Virtual Ground (regulated!) - and Rail Splitter Circuits!
- Thread starter Sonic Wonder
- Start date
KimLaroux
1000+ Head-Fier
- Joined
- Aug 2, 2011
- Posts
- 1,090
- Likes
- 70
Because those 4 diodes are supposed to have 4 volts across them, not 2.69.
So you've been using a simulator for those? I seem to remember you telling me you did not.. but whatever. The thing is, I don't trust a simulator for this application. We're using components in ways they were not meant to be used.
So you've been using a simulator for those? I seem to remember you telling me you did not.. but whatever. The thing is, I don't trust a simulator for this application. We're using components in ways they were not meant to be used.
DingoSmuggler
100+ Head-Fier
- Joined
- Jul 11, 2011
- Posts
- 283
- Likes
- 28
Quote:
Check the data sheet at the same current he is using here, just under 4mA.
The adjustment pins should only use up to 0.1mA each, and you already have 4mA - What advantage does more current give you?
I though low power was one of the design goals, but i do realise a lot of ideas are being thrown around.
I think testing it will give you some idea, but testing will probably just deliver the best result when at a given current/temp/etc the diode voltages match up best with the internal v-ref of the regs. But then in variable load conditions, and variable ambient temps, how good is this at maintaining good performance, and efficiency?
I've found this thread very interesting, but haven't had the time yet to sot down and go through any of the circuits in any level of depth. (not that i go all that deep lol)
Check the data sheet at the same current he is using here, just under 4mA.
The adjustment pins should only use up to 0.1mA each, and you already have 4mA - What advantage does more current give you?
I though low power was one of the design goals, but i do realise a lot of ideas are being thrown around.
I think testing it will give you some idea, but testing will probably just deliver the best result when at a given current/temp/etc the diode voltages match up best with the internal v-ref of the regs. But then in variable load conditions, and variable ambient temps, how good is this at maintaining good performance, and efficiency?
I've found this thread very interesting, but haven't had the time yet to sot down and go through any of the circuits in any level of depth. (not that i go all that deep lol)
Sonic Wonder
Member of the Trade: Goldpoint Level Controls
- Joined
- Dec 6, 2010
- Posts
- 44
- Likes
- 12
.
Sonic Wonder
Member of the Trade: Goldpoint Level Controls
- Joined
- Dec 6, 2010
- Posts
- 44
- Likes
- 12
.
KimLaroux
1000+ Head-Fier
- Joined
- Aug 2, 2011
- Posts
- 1,090
- Likes
- 70
Just looked at the datasheet. Turns out even diode's forward voltage is dependent on current. So it's still not a solution.
I think you're just chasing fairies here. When you consider everything, there's no way it's gonna work in real world. You'll need an active control system. At which point there's no reason to use regulators anymore.
Wikibaki's design is just so much better, and not much more complicated. What's wrong with it?
I think you're just chasing fairies here. When you consider everything, there's no way it's gonna work in real world. You'll need an active control system. At which point there's no reason to use regulators anymore.
Wikibaki's design is just so much better, and not much more complicated. What's wrong with it?
Sonic Wonder
Member of the Trade: Goldpoint Level Controls
- Joined
- Dec 6, 2010
- Posts
- 44
- Likes
- 12
Well the current is not going to change enough to really make much of a difference in the forward voltage drop, likely. KT88's point about temperature affecting current is true too. Um, the LM317/LM337 rail splitter virtual ground looks quite usable for what I'm doing with it. It's not super precise, sort of a brute force approach, but also interesting, to say the least. However, in my case, the proof is in the actual testing.
The virtual ground introduced at the beginning of this thread is a great fallback. As I claimed, it really sounds better with my headphone amp circuit than anything else I tried (other than running it directly from batteries, actually). The big deal is that main devices are husky voltage regulators. The ground point is held in place, but solidly! "I suspect that the reason for it's unusually good sound quality is this virtual ground is not made by "driving the ground point", but rather by "aggressively holding it at one potential" via the two complimentary voltage regulators."
Try it!
The virtual ground introduced at the beginning of this thread is a great fallback. As I claimed, it really sounds better with my headphone amp circuit than anything else I tried (other than running it directly from batteries, actually). The big deal is that main devices are husky voltage regulators. The ground point is held in place, but solidly! "I suspect that the reason for it's unusually good sound quality is this virtual ground is not made by "driving the ground point", but rather by "aggressively holding it at one potential" via the two complimentary voltage regulators."
Try it!
KT88
500+ Head-Fier
- Joined
- Nov 7, 2004
- Posts
- 948
- Likes
- 26
goldpoint, regarding what you wrote before editing the post, I can't build it because I don't have a zener with that nominal voltage, and placing an international order from mouser/digikey for a zener diode will cost too much 
However, there's really not much to it. Its a simple circuit, components variations are taken care of by the trimmer, temperature variations are of less importance due to the low tempco, and all that's left it variation in the current (due to variations in battery voltage). This is indeed the only real problem at the moment and could cause the voltage across the zener (or the 4 diodes in your circuit) to change by 10's of mV's. The simplest solution to this is to use a reference IC instead of the zener in the same circuit. It uses orders of magnitude less current for proper voltage generation, and being an active device with self biasing its much less sensitive to variations in the supply voltage.
If you are completely against using a reference IC we can try and think of a way to regulate the current in the zener, but using a JFET (CRD) isn't a good option because it has a minimum voltage drop across it to maintain regulation which will add to the zener voltage, and it will not give a good regulation in terms of the outputs being close to the supply/2. That's because it regulates the current and that current will still go through a resistor to the second supply line and it'll only be a constant offset above (or below, depends on how you connect it) that supply. I guess we could think of some more sophisticated way to regulate the current through the zener, but it'll probably hurt the simplicity of the circuit quite significantly so the question is if we even want to do this.
However, there's really not much to it. Its a simple circuit, components variations are taken care of by the trimmer, temperature variations are of less importance due to the low tempco, and all that's left it variation in the current (due to variations in battery voltage). This is indeed the only real problem at the moment and could cause the voltage across the zener (or the 4 diodes in your circuit) to change by 10's of mV's. The simplest solution to this is to use a reference IC instead of the zener in the same circuit. It uses orders of magnitude less current for proper voltage generation, and being an active device with self biasing its much less sensitive to variations in the supply voltage.
If you are completely against using a reference IC we can try and think of a way to regulate the current in the zener, but using a JFET (CRD) isn't a good option because it has a minimum voltage drop across it to maintain regulation which will add to the zener voltage, and it will not give a good regulation in terms of the outputs being close to the supply/2. That's because it regulates the current and that current will still go through a resistor to the second supply line and it'll only be a constant offset above (or below, depends on how you connect it) that supply. I guess we could think of some more sophisticated way to regulate the current through the zener, but it'll probably hurt the simplicity of the circuit quite significantly so the question is if we even want to do this.
vixr
500+ Head-Fier
- Joined
- Oct 14, 2005
- Posts
- 988
- Likes
- 24
This thread is amazing...thank you all for sharing this.
DingoSmuggler
100+ Head-Fier
- Joined
- Jul 11, 2011
- Posts
- 283
- Likes
- 28
Had a bit of a think about this circuit tonight and have come up with a different approach, still using LM317 & LM337, but using CSSes on the outputs to bias it.
Advantages? Tolerant of variable supply voltages due to CCS design.
Should be pretty good thermal wise, as the performance doesn't depend on the temp-co of the transistors relative to the v-reg, only each to its own complement which should stay pretty close. The CCS current will change with temp, but i don't think that will matter too much for the operation of the circuit.
Resistor values are just pulled from a hat, certainly not optimised, but should give a good idea of the concept.
It seems pretty nifty, and works in a sim but not totally convinced it will go as well in the real world. Maybe with a few tweaks and add in some trimming capability.
https://www.circuitlab.com/circuit/tr4a5y/virtual-ground/
Advantages? Tolerant of variable supply voltages due to CCS design.
Should be pretty good thermal wise, as the performance doesn't depend on the temp-co of the transistors relative to the v-reg, only each to its own complement which should stay pretty close. The CCS current will change with temp, but i don't think that will matter too much for the operation of the circuit.
Resistor values are just pulled from a hat, certainly not optimised, but should give a good idea of the concept.
It seems pretty nifty, and works in a sim but not totally convinced it will go as well in the real world. Maybe with a few tweaks and add in some trimming capability.
https://www.circuitlab.com/circuit/tr4a5y/virtual-ground/
Sonic Wonder
Member of the Trade: Goldpoint Level Controls
- Joined
- Dec 6, 2010
- Posts
- 44
- Likes
- 12
Hello KT88,
Good idea - the voltage reference. Um, I am not "against" that at all. Draw your preferred schematic for that and post it, please? However, it appears that this circuit, with or without a trimmer pot, works well for my application. The voltage regulators hold the ground point at one potential better, it seems, better than other circuits which "drive" the ground point with discreet transistors with or without opamps - my guess as to why the original posted circuit sounds so good too...
Good idea - the voltage reference. Um, I am not "against" that at all. Draw your preferred schematic for that and post it, please? However, it appears that this circuit, with or without a trimmer pot, works well for my application. The voltage regulators hold the ground point at one potential better, it seems, better than other circuits which "drive" the ground point with discreet transistors with or without opamps - my guess as to why the original posted circuit sounds so good too...
Sonic Wonder
Member of the Trade: Goldpoint Level Controls
- Joined
- Dec 6, 2010
- Posts
- 44
- Likes
- 12
We certainly have been having fun here!
KimLaroux
1000+ Head-Fier
- Joined
- Aug 2, 2011
- Posts
- 1,090
- Likes
- 70
Quote:
Interesting idea. Care to explain the theory behind it? I don't see what's keeping the output voltages at half the supply voltage.
Yeah I'm a newd, I know. Trying to learn.
Had a bit of a think about this circuit tonight and have come up with a different approach, still using LM317 & LM337, but using CSSes on the outputs to bias it.
Advantages? Tolerant of variable supply voltages due to CCS design.
Should be pretty good thermal wise, as the performance doesn't depend on the temp-co of the transistors relative to the v-reg, only each to its own complement which should stay pretty close. The CCS current will change with temp, but i don't think that will matter too much for the operation of the circuit.
Resistor values are just pulled from a hat, certainly not optimised, but should give a good idea of the concept.
It seems pretty nifty, and works in a sim but not totally convinced it will go as well in the real world. Maybe with a few tweaks and add in some trimming capability.
https://www.circuitlab.com/circuit/tr4a5y/virtual-ground/
Interesting idea. Care to explain the theory behind it? I don't see what's keeping the output voltages at half the supply voltage.
Yeah I'm a newd, I know. Trying to learn.
wakibaki
1000+ Head-Fier
- Joined
- May 26, 2011
- Posts
- 1,088
- Likes
- 68
Quote:
Quote:
I think the idea is that if you connect the output to ground, then it will be at 0V.
w
Had a bit of a think about this circuit tonight and have come up with a different approach, still using LM317 & LM337, but using CSSes on the outputs to bias it.
Advantages? Tolerant of variable supply voltages due to CCS design.
Should be pretty good thermal wise, as the performance doesn't depend on the temp-co of the transistors relative to the v-reg, only each to its own complement which should stay pretty close. The CCS current will change with temp, but i don't think that will matter too much for the operation of the circuit.
Resistor values are just pulled from a hat, certainly not optimised, but should give a good idea of the concept.
It seems pretty nifty, and works in a sim but not totally convinced it will go as well in the real world. Maybe with a few tweaks and add in some trimming capability.
https://www.circuitlab.com/circuit/tr4a5y/virtual-ground/
Quote:
I think the idea is that if you connect the output to ground, then it will be at 0V.
w
DingoSmuggler
100+ Head-Fier
- Joined
- Jul 11, 2011
- Posts
- 283
- Likes
- 28
Quote:
I tried all sort of things last night, and ended up with this circuit, but it's not ready yet, but the basic idea is there. It was past 1am in the morning when i posted, I just wanted to get it up there
Each Vreg has a 1.25V reference between ADJ & OUT. R1/Q1, and R2/Q2 use this Vref to set up CCS. The voltage across the resistor R1 will be the Vref - BE diode drop, so about 0.6V. Solving ohms law will give the CCS current. So with the values provided its roughly 1mA, but might work better with a higher current, not sure yet.
Now the idea is, both current sources are joined through the 1-ohm resistors (probably could go smaller there), so with 1mA flowing through 2-ohms, the 2 outputs are drawn together so that they are 2mV apart. BUT...
.gif)
The LM317 will be able to source the current for the CCS attached to it, and likewise the LM337 will be able to sink the current for its CCS. I'm hoping this doesn't upset the whole design. But really we should just need some current to flow through the resistors, not all the current. Maybe we just need to reduce the 1 ohm resistors to 0.1ohm so that the impedance is similar to the output impedances of the Vregs.
Indeed!
I realised there was a bit of an issue here, but haven't decided how best to tackle it, but I think there should be a simple solution.
I think it works in the sim though, as each transistor is turning on enough to push the output into the middle. I don't think real world transistors would be as kind.
Certainly more work needs to be done.
As am I, a newb, trying to learn.
I find the best way to learn is by doing, and this thread just happened to capture my attention as such.
Now whether this circuit has potential to succeed, or is going to fail, I'm not sure yet, but certainly I will learn a few things in the process.
I tried all sort of things last night, and ended up with this circuit, but it's not ready yet, but the basic idea is there. It was past 1am in the morning when i posted, I just wanted to get it up there
Each Vreg has a 1.25V reference between ADJ & OUT. R1/Q1, and R2/Q2 use this Vref to set up CCS. The voltage across the resistor R1 will be the Vref - BE diode drop, so about 0.6V. Solving ohms law will give the CCS current. So with the values provided its roughly 1mA, but might work better with a higher current, not sure yet.
Now the idea is, both current sources are joined through the 1-ohm resistors (probably could go smaller there), so with 1mA flowing through 2-ohms, the 2 outputs are drawn together so that they are 2mV apart. BUT...
Indeed!
I realised there was a bit of an issue here, but haven't decided how best to tackle it, but I think there should be a simple solution.
I think it works in the sim though, as each transistor is turning on enough to push the output into the middle. I don't think real world transistors would be as kind.
Certainly more work needs to be done.
As am I, a newb, trying to learn.
I find the best way to learn is by doing, and this thread just happened to capture my attention as such.
Now whether this circuit has potential to succeed, or is going to fail, I'm not sure yet, but certainly I will learn a few things in the process.
Users who are viewing this thread
