I started to make the power supply for a headphone amp, as the transformer I had gave me more volts than I needed I decided to make the supply a choke input type.
I thought I would read up a bit more about this type of supply.
Opening Morgan Jones's book to this section I started to read about the broadband responce of of LC filters and the Q of the circuit to avoid resonances. A Q of .5 is aimed for.
I ask, how many people, when making this type of supply take the Q of the circuit into consideration and is it practicable with the resistance values of todays chokes, Llundahls 10H choke is 36 ohms whereas older chokes seem to have larger dcr values of say 220 ohms for a 10 H choke.
Achieving a Q of .5 with a choke of 220 ohms dcr and a cap of say 220mfd is easily done by adding extra resistance in series with the choke but what about if I was using the Llundahl choke of 36 ohms, it would seem I would need a cap of 30,000mfd to achieve a Q of .5.
Have I gone wrong somewhere in my calculations or am I barking up the wrong tree?
What about power supply impeadance, is a high impeadance psu good or bad? If I didn't use a large cap with my Llundahl choke I would need a pretty big resistance in series with the choke to achieve a Q of .5.
Using my 220 ohm choke I worked out the critical current draw and to achieve this I have a load resistance of 11k, how does this resistance fit into the Q formula? Morgan calls it a notional extra choke series resistance but i'm not sure how to understand this and how it fits in.
Lastly, it seems that the resonances set up by LC networks in the power supply have resonces and harmonics well within the audio passband so how does it, how will it effect the final sound from the amp?
I thought I would read up a bit more about this type of supply.
Opening Morgan Jones's book to this section I started to read about the broadband responce of of LC filters and the Q of the circuit to avoid resonances. A Q of .5 is aimed for.
I ask, how many people, when making this type of supply take the Q of the circuit into consideration and is it practicable with the resistance values of todays chokes, Llundahls 10H choke is 36 ohms whereas older chokes seem to have larger dcr values of say 220 ohms for a 10 H choke.
Achieving a Q of .5 with a choke of 220 ohms dcr and a cap of say 220mfd is easily done by adding extra resistance in series with the choke but what about if I was using the Llundahl choke of 36 ohms, it would seem I would need a cap of 30,000mfd to achieve a Q of .5.
Have I gone wrong somewhere in my calculations or am I barking up the wrong tree?
What about power supply impeadance, is a high impeadance psu good or bad? If I didn't use a large cap with my Llundahl choke I would need a pretty big resistance in series with the choke to achieve a Q of .5.
Using my 220 ohm choke I worked out the critical current draw and to achieve this I have a load resistance of 11k, how does this resistance fit into the Q formula? Morgan calls it a notional extra choke series resistance but i'm not sure how to understand this and how it fits in.
Lastly, it seems that the resonances set up by LC networks in the power supply have resonces and harmonics well within the audio passband so how does it, how will it effect the final sound from the amp?
kopite said:
I don't yet consider myself competant enough to comment on your good technical questions. But I can say the PSU resonances will affect the sonics of the amp, so you're right to want to get this part of the design right.
Have a look at this page which is
Jim de Kort's take on the matter. Also check out the great little PSU2 modeling program from Duncan amps.
Hi Majestic,
Thanks for your reply. Yes I'm aware of Duncan's psu designer and great it is. I use it all the time but it won't give me the answers that I need.
Thanks for your reply. Yes I'm aware of Duncan's psu designer and great it is. I use it all the time but it won't give me the answers that I need.
Your number-crunching is correct. It is possible to have a low Q or a low output resistance, but achieving the two simultaneously is difficult. You generally have to compromise by adding a little series resistance to the choke and a lot of capacitance after it.
From the point of view of the resonant circuit, the load resistance adds a little more damping and you can convert this shunt resistance into a notional series resistance which you add to the choke series resistance. In practice, it doesn't generally make much difference, so it's probably not worth the bother.
You are quite right, power supply resonances are within the audio band and can make quite a difference to the sound of the amplifier. There are three ways of dealing with them:
(1) Ignore them.
(2) Tune the LF resonance such that you hope it will interact with the loudspeakers in a way that sounds good. A lot of sweeping assumptions need to be made.
(3) Reduce the Q.
One method that has proved quite popular when using an LCLC filter is to play with the current step response in PSUD whilst juggling capacitor values and positions. You usually end up with a filter having small capacitor values at the input and larger ones towards the output.
From the point of view of the resonant circuit, the load resistance adds a little more damping and you can convert this shunt resistance into a notional series resistance which you add to the choke series resistance. In practice, it doesn't generally make much difference, so it's probably not worth the bother.
You are quite right, power supply resonances are within the audio band and can make quite a difference to the sound of the amplifier. There are three ways of dealing with them:
(1) Ignore them.
(2) Tune the LF resonance such that you hope it will interact with the loudspeakers in a way that sounds good. A lot of sweeping assumptions need to be made.
(3) Reduce the Q.
One method that has proved quite popular when using an LCLC filter is to play with the current step response in PSUD whilst juggling capacitor values and positions. You usually end up with a filter having small capacitor values at the input and larger ones towards the output.
- Status
- Not open for further replies.