I could do that, Iko, and I'm guessing that the shunt capacitance (not unlike C2's position) might aid in stability. OLG would skyrocket, which IMO isn't necessary. I'm not sure whether a 2SK170 is quieter than a resistor, I've heard they're pretty quiet. But, with only .68V across it, would it perform much better than a resistor?
On a random thought, R8 might be replaced by a Jfet CCS, which would help protect during input overload. But I'm not sure of the implications for offset correction.
- keantoken
On a random thought, R8 might be replaced by a Jfet CCS, which would help protect during input overload. But I'm not sure of the implications for offset correction.
- keantoken
Hey Iko, were you able to get what you wanted from modifying the circuit?
I'm playing in DipTrace to find a good layout for the veroboard. It is challenging to get it all on one side but the traces are converging... A few jumpers are permissible.
I have some questions...
1: Should I separate signal and power grounds, does it matter at such low levels?
2: Should I connect the case to ground, and if so which ground? Any interference caught by the case would have to go through the source ground.
- keantoken
I'm playing in DipTrace to find a good layout for the veroboard. It is challenging to get it all on one side but the traces are converging... A few jumpers are permissible.
I have some questions...
1: Should I separate signal and power grounds, does it matter at such low levels?
2: Should I connect the case to ground, and if so which ground? Any interference caught by the case would have to go through the source ground.
- keantoken
Attachments
Hey kt, I've started working a few days ago on a pcb layout for myself, which I can share with you if you'll ever want it. I have very little time and too many things on my plate at the moment, and the regulator is the priority, meaning it will probably take me quite a while to finish it. I like your circuit as I said, and definitely want to give it a try for real.
About grounding I'd do the usual practice, keep the power ground separate from the signal ground and then connect the two in a star point. Did you decide on a power supply for this? Batteries?
About grounding I'd do the usual practice, keep the power ground separate from the signal ground and then connect the two in a star point. Did you decide on a power supply for this? Batteries?
I'm using a transformer. With the high bias currents It would make short work of batteries. Of course the bias can be turned down depending on how sensitive your headphones are. I'm using a total of 24mF of caps.
Yes, I would like to see your PCB layout too if you don't mind. I've arranged mine so that the input transistors can be zip tied or heatshrinked together for best thermal management.
- keantoken
Yes, I would like to see your PCB layout too if you don't mind. I've arranged mine so that the input transistors can be zip tied or heatshrinked together for best thermal management.
- keantoken
I probably will, since I hate suspense.
I also have to remark on the choice of grounding. I've created a virtual ground by using two resistors connected across the rails. This protects against DC offset, and it also limits the amount of DC ground current, which helps protect the circuit if something goes wrong.
Of course, I'm hoping things don't go wrong.
If I could get a PCB thrown together with all the available mods for low distortion vs. H2, and get it done by some company, this would go by much quicker. Maybe there could even be a group buy?
But I think that will wait until I get this working.
Also, there is some standard that puts a 120 ohm resistor in series with any headphone amplifier's output. This protects against a variety of things, but I also wonder about the sonic implications. Of course, this is just an "option" that would go on the final board.
- keantoken
I also have to remark on the choice of grounding. I've created a virtual ground by using two resistors connected across the rails. This protects against DC offset, and it also limits the amount of DC ground current, which helps protect the circuit if something goes wrong.
Of course, I'm hoping things don't go wrong.
If I could get a PCB thrown together with all the available mods for low distortion vs. H2, and get it done by some company, this would go by much quicker. Maybe there could even be a group buy?
But I think that will wait until I get this working.
Also, there is some standard that puts a 120 ohm resistor in series with any headphone amplifier's output. This protects against a variety of things, but I also wonder about the sonic implications. Of course, this is just an "option" that would go on the final board.
- keantoken
On the series resistor, there is the thought of using a rheostat, which could actually act sort of like a tone control. Label it "damping".
Well, I plugged it in and turned it on. As expected, it doesn't work. Hum is monstrous, and there is an odd crackling noise deep within the circuit. Oddly enough however, I don't smell anything new and as far as I can tell nothing is getting hot. So I guess this leaves capacitors... 12V rails are at 2V for some reason. Looks like my 10mF caps are gone, though I sure hope not. Outputs didn't get hot, do it doesn't look like they are the cause. 15V rails are fine at 16V.
Aha! I connected the output stage with the wrong polarity... Oops...
- keantoken
Well, I plugged it in and turned it on. As expected, it doesn't work. Hum is monstrous, and there is an odd crackling noise deep within the circuit. Oddly enough however, I don't smell anything new and as far as I can tell nothing is getting hot. So I guess this leaves capacitors... 12V rails are at 2V for some reason. Looks like my 10mF caps are gone, though I sure hope not. Outputs didn't get hot, do it doesn't look like they are the cause. 15V rails are fine at 16V.
Aha! I connected the output stage with the wrong polarity... Oops...
- keantoken
Well, I don't think I blew the MOSFETs, I've looked at the datasheet and there's no way I can summon enough current or voltage to cook them.
But it looks like the lower Allison transistor has been cooked. Vcb reads 0 ohms. I've cut the collector connection and I'm going to use the transistor's BE junction as a diode, which according to simulation has little affect on performance if any.
- keantoken
But it looks like the lower Allison transistor has been cooked. Vcb reads 0 ohms. I've cut the collector connection and I'm going to use the transistor's BE junction as a diode, which according to simulation has little affect on performance if any.
- keantoken
Alright, it looks like the lower MOSFET is drawing too much current for some reason, it's getting hot and the cap on it's rail reads a few mV, while the other read 5V. Something is also buzzing (it's not the trafo and I don't think it's a cap - however I've heard MOSFETs buzz before). This sounds like the amp circuitry is messed up somehow.
- keantoken
- keantoken
For those who worry about matching, here is the spectrum with a mismatch of 40mV for the input and feedback transistors, to give a sense of scale. 3rd harmonics are higher, but nothing awful. Some may prefer more 3rd harmonics, and it would be hard to make 3rd harmonics equal or stronger than 2nd harmonics with this specific design, since it naturally cancels odd harmonics.
- keantoken
- keantoken
Attachments
I'm not using a load at the moment. I know it's just the one MOSFET because only one of them gets hot.
Since the bias resistors are fixed, I can't change the bias current.
This is very puzzling. I have no load connected, yet the lower MOSFET is somehow dissipating a lot of heat, while the top MOSFET is cool. Without a load, all the the current of the lower MOSFET has to go through the top MOSFET. With all the heat it's putting out, the driver transistor should be fried. But no. It's not even hot.
- keantoken
Since the bias resistors are fixed, I can't change the bias current.
This is very puzzling. I have no load connected, yet the lower MOSFET is somehow dissipating a lot of heat, while the top MOSFET is cool. Without a load, all the the current of the lower MOSFET has to go through the top MOSFET. With all the heat it's putting out, the driver transistor should be fried. But no. It's not even hot.
- keantoken
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.