00940
Headphoneus Supremus
- Joined
- Nov 6, 2002
- Posts
- 4,493
- Likes
- 47
Over the years, I've played with quite a few little amps (so many it often diverted me from my big projects, but that's another question
). This version is the fusion of a few designs, keeping their stronger points and simplifying everything else when possible.
Let's start with the PS. The amp is to be powered by a laptop power supply. They're 19V, regulated enough, rated for tons of current, often without ground pin (no ground loop to fear) and even the worst scavenger should be able to find one for free. Obviously, they also have two problems: they're quite dirty as they're switchers and they're single rail supplies. The first problem is taken care by a filter at the DC input, using an 1mh inductor and a 1uf ceramic or mkt cap. The second problem is taken care of by a virtual ground. Using 1K resistors and 2200uf caps, it's solid enough for our use without the complication of an active virtual ground.
The amp itself now. It's based on a ne5534. I know, I know, not the sexiest chip around. However, it's a solid performer and, when the output is taken from compensation pin5, a good sounding one. This pin is able to source quite a bit of current at a lowish impedance so we can drive a simple class A buffer from it. Here, a bd139/16 biased at around 100ma. Everything, opamp, output stage, is working deeply in class A. R4 might not be necessary, just jumper it for the lowest possible output impedance.
There's a catch with the ne5534: it's a bjt input opamp. As no input caps are used, you will have a dc offset in between 0 and 10mv (rough figures), depending on the position of the pot. Nothing to worry about. To balance input currents while preserving a high enough input impedance, I also had to pick highish values for the feedback resistors, raising the noise floor up to around -88db (calculated figure). It's still a good figure. The input resistors values, when combined with a linear 100K potentiometer (and not a log one), give an usable attenuation curve. Using a linear pot often offers better matching than with log pots, especially with cheap ones.
Below is a pcb showing how compact it could be made. It's 10cm/8cm, single sided.
Very important
I tested some laptop power supplies and earth/ground could be problematic with some:
- If they have two-prong AC cords, there shouldn't be any problems.
- If they have two three-prong AC cords, I found out three possibilities:
1/ the "DC ground" is not connected to earth. No problem.
2/ the "DC ground" is connected to earth. Such a supply is not suitable. If you connect the amp to a source in which earth and ground are linked, you'll pull the virtual ground down.
3/ the "DC ground" is not connected to earth but there is a third wire connected to earth. This wire should not be connected.
Very important
Let's start with the PS. The amp is to be powered by a laptop power supply. They're 19V, regulated enough, rated for tons of current, often without ground pin (no ground loop to fear) and even the worst scavenger should be able to find one for free. Obviously, they also have two problems: they're quite dirty as they're switchers and they're single rail supplies. The first problem is taken care by a filter at the DC input, using an 1mh inductor and a 1uf ceramic or mkt cap. The second problem is taken care of by a virtual ground. Using 1K resistors and 2200uf caps, it's solid enough for our use without the complication of an active virtual ground.
The amp itself now. It's based on a ne5534. I know, I know, not the sexiest chip around. However, it's a solid performer and, when the output is taken from compensation pin5, a good sounding one. This pin is able to source quite a bit of current at a lowish impedance so we can drive a simple class A buffer from it. Here, a bd139/16 biased at around 100ma. Everything, opamp, output stage, is working deeply in class A. R4 might not be necessary, just jumper it for the lowest possible output impedance.
There's a catch with the ne5534: it's a bjt input opamp. As no input caps are used, you will have a dc offset in between 0 and 10mv (rough figures), depending on the position of the pot. Nothing to worry about. To balance input currents while preserving a high enough input impedance, I also had to pick highish values for the feedback resistors, raising the noise floor up to around -88db (calculated figure). It's still a good figure. The input resistors values, when combined with a linear 100K potentiometer (and not a log one), give an usable attenuation curve. Using a linear pot often offers better matching than with log pots, especially with cheap ones.
Below is a pcb showing how compact it could be made. It's 10cm/8cm, single sided.
I tested some laptop power supplies and earth/ground could be problematic with some:
- If they have two-prong AC cords, there shouldn't be any problems.
- If they have two three-prong AC cords, I found out three possibilities:
1/ the "DC ground" is not connected to earth. No problem.
2/ the "DC ground" is connected to earth. Such a supply is not suitable. If you connect the amp to a source in which earth and ground are linked, you'll pull the virtual ground down.
3/ the "DC ground" is not connected to earth but there is a third wire connected to earth. This wire should not be connected.
