I would appreciate some help to improve the power supply for the audioboard in my Arcam Delta 70.2 CD player. The audioboard has its own 12VA 15V + 15V transformer giving +/-20Vdc - which is filtered by CRC:- 1000uF-6R8-1000uF. This then feeds the regs; 7805/7905 for +/-5V, LM337 for -15V and a discrete +/-12V reg for I/V and opamps.
I was thinking changing the existing Tx to a 25VA torroidal unit and applying some additional filtering and / or a pre-reg. at 20V, to feed directly into the existing arrangement.
However, there are a few things I need advice on before proceeding. Firstly, the choice of transformer. I have a 15V + 15V 25VA torroidal - which would make an easy swap - but in Duncan Amps software it doesn’t bring much of a change to the conduction time or current peaks. I also have a 22V + 22V 25VA unit which seems to be more interesting - but I'm a bit concerned about the higher voltage causing some damage - especially if the load drops for some reason. Secondly, I’m not sure whether to use a pre-regulator or not.
Modelling some options in Duncan Amps PSUD2 and assuming a current draw of 150mA (to allow for excess) shows the following info:
Existing Power Supply:- 15V + 15V 12VA - diodes - 1000uF - 6R8 - 1000uF - regs.
Output voltage - 19.9V
Ripple voltage - 179mV
Conduction time - 3.8mS
Current Peaks - 0.5A
Option 1:- 15V + 15V 25VA - diodes - 1000uF - 6R8 - 1000uF - regs.
Output voltage - 21.5V
Ripple voltage - 196mV
Conduction time - 2.25mS
Current Peaks - 0.8A
This appears to be inferior to the original arrangement - but the addition of 3R3 - 470uF - 5R before the existing filter gives the following:
Output voltage - 19.0V
Ripple voltage - 83mV
Conduction time - 3.8mS
Current Peaks - 0.5A
Better - but not a huge improvement over the original.
Option 2:- 22V + 22V 25VA - diodes - 40R – (existing filter)1000uF - 6R8 - 1000uF - regs.
Output voltage – 18.9V
Ripple voltage - 133mV
Conduction time - 6.2mS
Current Peaks - 0.36A
This seems to be the best so far and only requires the addition of a 40R resistor. However, the no-load voltage could be as high as 36V. Will this cause any problems with the kill circuit etc?
Option 3:- 22V + 22V 25VA - diodes - 20R - 100uF - 20R – (existing filter)1000uF - 6R8 - 1000uF - regs.
Output voltage - 19.9V
Ripple voltage - 87mV
Conduction time - 5.5mS
Current Peaks - 0.4A
This also seems to be quite good and only requires the addition of 20R-100u-20R before the existing filter. However, the same no-load voltage concerns exist as above.
Option 4:- 22V + 22V 25VA - diodes - 10R - 220uF - 20R - 1000uF - (LM317/337 pre-reg set to 20V)
Output voltage - 24.67V
Ripple voltage - 251mV
Conduction time - 4.6mS
Current Peaks - 0.48A
I assume that the pre-reg would reduce the ripple to 251/10,000 = 0.0251mV. This could then be fed into the existing arrangement.
Any thoughts, comments or suggestion on this would be appreciated.
I was thinking changing the existing Tx to a 25VA torroidal unit and applying some additional filtering and / or a pre-reg. at 20V, to feed directly into the existing arrangement.
However, there are a few things I need advice on before proceeding. Firstly, the choice of transformer. I have a 15V + 15V 25VA torroidal - which would make an easy swap - but in Duncan Amps software it doesn’t bring much of a change to the conduction time or current peaks. I also have a 22V + 22V 25VA unit which seems to be more interesting - but I'm a bit concerned about the higher voltage causing some damage - especially if the load drops for some reason. Secondly, I’m not sure whether to use a pre-regulator or not.
Modelling some options in Duncan Amps PSUD2 and assuming a current draw of 150mA (to allow for excess) shows the following info:
Existing Power Supply:- 15V + 15V 12VA - diodes - 1000uF - 6R8 - 1000uF - regs.
Output voltage - 19.9V
Ripple voltage - 179mV
Conduction time - 3.8mS
Current Peaks - 0.5A
Option 1:- 15V + 15V 25VA - diodes - 1000uF - 6R8 - 1000uF - regs.
Output voltage - 21.5V
Ripple voltage - 196mV
Conduction time - 2.25mS
Current Peaks - 0.8A
This appears to be inferior to the original arrangement - but the addition of 3R3 - 470uF - 5R before the existing filter gives the following:
Output voltage - 19.0V
Ripple voltage - 83mV
Conduction time - 3.8mS
Current Peaks - 0.5A
Better - but not a huge improvement over the original.
Option 2:- 22V + 22V 25VA - diodes - 40R – (existing filter)1000uF - 6R8 - 1000uF - regs.
Output voltage – 18.9V
Ripple voltage - 133mV
Conduction time - 6.2mS
Current Peaks - 0.36A
This seems to be the best so far and only requires the addition of a 40R resistor. However, the no-load voltage could be as high as 36V. Will this cause any problems with the kill circuit etc?
Option 3:- 22V + 22V 25VA - diodes - 20R - 100uF - 20R – (existing filter)1000uF - 6R8 - 1000uF - regs.
Output voltage - 19.9V
Ripple voltage - 87mV
Conduction time - 5.5mS
Current Peaks - 0.4A
This also seems to be quite good and only requires the addition of 20R-100u-20R before the existing filter. However, the same no-load voltage concerns exist as above.
Option 4:- 22V + 22V 25VA - diodes - 10R - 220uF - 20R - 1000uF - (LM317/337 pre-reg set to 20V)
Output voltage - 24.67V
Ripple voltage - 251mV
Conduction time - 4.6mS
Current Peaks - 0.48A
I assume that the pre-reg would reduce the ripple to 251/10,000 = 0.0251mV. This could then be fed into the existing arrangement.
Any thoughts, comments or suggestion on this would be appreciated.