All good comments Chris.
It actually isn't the ripple current so much as the crossover frequency that can bleed into the amp. At 250KHz a good crossover would be at 25KHz and most would have it lower than that so as you can see you start getting into the top end of your audible range. Most of the power supplies I do AC/Dc would have been at about 65KHz to keep me out of the 150KHz start of the conducted emissions band.
As to the filters once you start adding "CLC" into the PSU rail you can start getting peaking from the loaded/unloaded Q of the tank. These multiple resonance points can sometimes be hard to control. Typically you would then push them down below (or up above) the audible band. If I must filter I usually will use a resistive element rather than inductive. An old friend always told be it's better to burn it off as heat than radiate it as EMI. I will sometimes use capacitance multipliers quiet things up and provide a bit of isolation on the line so that line stages don't talk to each other as much.
All that being said, I am a switching power supply engineer for the last 15 years or so and before that my background is RF. I actually prefer switchers to the linear's I put in here. The whole module is too large.
I attached a voltage mode 18W supply I started for the PP KT88's. 6.3V@3A.
. There is no standby with this one and the soft-start is a little too fast. It does have OVP and thermal shutdown.
Thanks for taking the time to comment. It is always great to here what others would do. There are any number of ways to do a design. All of them are correct as long as the end product meets spec. And by spec, that means form/fit/function/cost/schedule. Pretty much in that order.
Great conversation, thanks again.
Tony
It actually isn't the ripple current so much as the crossover frequency that can bleed into the amp. At 250KHz a good crossover would be at 25KHz and most would have it lower than that so as you can see you start getting into the top end of your audible range. Most of the power supplies I do AC/Dc would have been at about 65KHz to keep me out of the 150KHz start of the conducted emissions band.
As to the filters once you start adding "CLC" into the PSU rail you can start getting peaking from the loaded/unloaded Q of the tank. These multiple resonance points can sometimes be hard to control. Typically you would then push them down below (or up above) the audible band. If I must filter I usually will use a resistive element rather than inductive. An old friend always told be it's better to burn it off as heat than radiate it as EMI. I will sometimes use capacitance multipliers quiet things up and provide a bit of isolation on the line so that line stages don't talk to each other as much.
All that being said, I am a switching power supply engineer for the last 15 years or so and before that my background is RF. I actually prefer switchers to the linear's I put in here. The whole module is too large.
I attached a voltage mode 18W supply I started for the PP KT88's. 6.3V@3A.
. There is no standby with this one and the soft-start is a little too fast. It does have OVP and thermal shutdown.Thanks for taking the time to comment. It is always great to here what others would do. There are any number of ways to do a design. All of them are correct as long as the end product meets spec. And by spec, that means form/fit/function/cost/schedule. Pretty much in that order.
Great conversation, thanks again.
Tony
not overly complicated
I don't really think this is overly complicated. Overly complicated might be a constant power filament regulator or better (and virtually impossible) a regulator that tried to maintain a set cathode temperature. How do you measure the cathode temperature?
hehe
-- mirlo
I don't really think this is overly complicated. Overly complicated might be a constant power filament regulator or better (and virtually impossible) a regulator that tried to maintain a set cathode temperature. How do you measure the cathode temperature?
hehe
-- mirlo
We should be able to do a constant temperature heater supply. You place a themistor on the copper where the heater wire attaches. You use a multi-second integrator to average the temperature. Since the CCS design is basically a voltage controlled current source you can move the Vref line around to increase or decrease the heater temperature by slowly changing the current through the heater wire.
Tony
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.