At the risk of inspiring ZM to pull his hair out and a general chorus of sneering from those who I realize know much more about this stuff than I do I am going to ask a power supply question.
This is based upon the fact that I know I will have to get new transformers and I have a spare heatsink available to me AND most importantly since I carry the anything but recessive gene of Scots "thriftiness" which I prefer to call cheapness I have continued to play around with PSUD. I hear the groans ...
With using four sections (is the first capacitor a section?) and three resistors, two 50 mR and one 25 mR and two 30mF, a 60 mF and a last 100 mF caps i am getting truly vestigial ripple. Now I realize this is all theoretical and I do not doubt the real world will be very different (and I do not have the means to measure it).
Looks like I will have to use a 32 volts transformer, even at 400 VA it costs little more than the 18V 300VA from ANTEK.
I can use the spare heatsink to mount the resistors if I have to. I would start with mounting them to the bottom panel.
i get predictions of 9uV ripple. Now the voltage comes up slowly though I wonder if I should still use some kind of standby switch for start up since I do not want to deal with relays and their support circuitry. it takes almost a minute for the voltage to completely stabilize. Is that good or bad? Is this configuration TOO slow? Of course the last 15 seconds or so are for a small rise in voltage, 1 to 2 mV. Does this mean anything?
So all of this SEEMS good but I know, other than the obvious heat and inefficiency, there must be something more to this other than the expense of the capacitors. The chassis mount 50 w resistors are not that expensive, luckily. This COULD cost less than using even a HAMMOND choke and the predicted performance is much better.
The other approach would be to use a choke instead of one of the resistors but the thing predicts better performance from the resistors. Now I do not doubt the choke would be superior at blocking high frequency junk but wonder with an amp limited to 500 hz first order low pass, is that needed?
100uV ripple with one choke and two resistors v. 9 uV ripple with resistors, only. Two chokes and one resistor goes backwards from there.
So after all of this rambling, what am I missing? What is the downside of multiple filters?
I think BRIDGE 25 in PSUD is some kind of ideal rectifier model. It certainly performs better than any of the other choices. If this does correspond to a real world part, does anyone know what it is?
This is based upon the fact that I know I will have to get new transformers and I have a spare heatsink available to me AND most importantly since I carry the anything but recessive gene of Scots "thriftiness" which I prefer to call cheapness I have continued to play around with PSUD. I hear the groans ...
With using four sections (is the first capacitor a section?) and three resistors, two 50 mR and one 25 mR and two 30mF, a 60 mF and a last 100 mF caps i am getting truly vestigial ripple. Now I realize this is all theoretical and I do not doubt the real world will be very different (and I do not have the means to measure it).
Looks like I will have to use a 32 volts transformer, even at 400 VA it costs little more than the 18V 300VA from ANTEK.
I can use the spare heatsink to mount the resistors if I have to. I would start with mounting them to the bottom panel.
i get predictions of 9uV ripple. Now the voltage comes up slowly though I wonder if I should still use some kind of standby switch for start up since I do not want to deal with relays and their support circuitry. it takes almost a minute for the voltage to completely stabilize. Is that good or bad? Is this configuration TOO slow? Of course the last 15 seconds or so are for a small rise in voltage, 1 to 2 mV. Does this mean anything?
So all of this SEEMS good but I know, other than the obvious heat and inefficiency, there must be something more to this other than the expense of the capacitors. The chassis mount 50 w resistors are not that expensive, luckily. This COULD cost less than using even a HAMMOND choke and the predicted performance is much better.
The other approach would be to use a choke instead of one of the resistors but the thing predicts better performance from the resistors. Now I do not doubt the choke would be superior at blocking high frequency junk but wonder with an amp limited to 500 hz first order low pass, is that needed?
100uV ripple with one choke and two resistors v. 9 uV ripple with resistors, only. Two chokes and one resistor goes backwards from there.
So after all of this rambling, what am I missing? What is the downside of multiple filters?
I think BRIDGE 25 in PSUD is some kind of ideal rectifier model. It certainly performs better than any of the other choices. If this does correspond to a real world part, does anyone know what it is?
Here is one with the LUNDAHL as an input choke.
Probably have it running with too much current.
Would need to use a smaller voltage power transformer.
No question the LUNDAHL's inductance and resistance make for a very controlled turn on.
I like the idea of a fourth order filter. Maybe this is what I will stick with. Who knows?
Figure I will run one amp AS IS mono and do the modded supply on the other and listen for a difference.
Probably have it running with too much current.
Would need to use a smaller voltage power transformer.
No question the LUNDAHL's inductance and resistance make for a very controlled turn on.
I like the idea of a fourth order filter. Maybe this is what I will stick with. Who knows?
Figure I will run one amp AS IS mono and do the modded supply on the other and listen for a difference.
Attachments
Last edited:
Hi Rick
Along with static load eg 2.6A to 3A for stereo, put a 2A sine or square wave 40Hz active load (ie simulated bass note) on the supply and look at the ripple on the rails.
If it's still less than 10mV peak to peak, I will be very impressed. I predict it's going to look terrible unless the last cap/s on the supply are at least 100,000uF in value.
Along with static load eg 2.6A to 3A for stereo, put a 2A sine or square wave 40Hz active load (ie simulated bass note) on the supply and look at the ripple on the rails.
If it's still less than 10mV peak to peak, I will be very impressed. I predict it's going to look terrible unless the last cap/s on the supply are at least 100,000uF in value.
Last edited:
lilling da guildy or guilding a lilly or splittin' da hair ........ 
even if your measuring caliper is way of , as long you're using it both for measuring and fabrication/cutting , everything is fine
as far I know , one don't need hysterical load function in PSUD
when you need that , it's time for learning LTSpice
even if your measuring caliper is way of , as long you're using it both for measuring and fabrication/cutting , everything is fine
as far I know , one don't need hysterical load function in PSUD
when you need that , it's time for learning LTSpice
Rick sent me that file in e-mail
- 7z is extension not included in forum software , same as original .psu one
but plain zipped is
Rick - once you must stop planning , especially when you are just changing one overkill to other
this sch is overkill , if you ask me , so - god to go
load changed to CCS , 1A65 as per one channel Iq
edit:
check is it saved in sim :
simulate for 50mS after reporting delay of 5S
- 7z is extension not included in forum software , same as original .psu one
but plain zipped is
Rick - once you must stop planning , especially when you are just changing one overkill to other
this sch is overkill , if you ask me , so - god to go
load changed to CCS , 1A65 as per one channel Iq
edit:
check is it saved in sim :
simulate for 50mS after reporting delay of 5S
Last edited:
pico,
Last cap is 100m. Thought I would start there and go up some.
I do not have test gear but I will try and see if i can come up with some.
I call them predictions and they do need some real world testing to make sense out of them. At this point I use them relatively; comparing different things to see what happens.
Take care,
Last cap is 100m. Thought I would start there and go up some.
I do not have test gear but I will try and see if i can come up with some.
I call them predictions and they do need some real world testing to make sense out of them. At this point I use them relatively; comparing different things to see what happens.
Take care,