Hi,
I am trying to put together a valve amplifer. I understand filters but I am kind of stuck about the values of capacitors I will be using within the filter.
The amplifer is aimed at being high end and, since the amplification is by valves, a slow start up time is not a real concern.
Obviously, as I increase the capacitors' storage values the charge time becomes longer and the filtration better. But I am having trouble with the reasoning for the values and their effects. By that I mean, for instance, why should I use a larger value for the first capacitor in the filter and smaller one in the second, or the reverse? Why shouldn't I balance the capacitances and so on.
My only reasoning so far is that if I have the largest value infront of the inductor then it will draw the largest initial current to charge itself, so subjecting the inductor to a lower surge value, and so the induction value of the choke itself could be increased. But then I ask, well why not just add the value of the second capacitor after the inductor to the first and have one huge capacitor before the inductor and none after it? I guess another way of phrasing this is, what is the effect of splitting the capacitance around the filter? This is the question that confuses me the most. The induction choke will not introduce ripple itself, it will only act to filter it. Since the filters are defined as LC, that being a choke and then a filter, I would guess there is some kind of reason for there being a capacitance after the choke also.
I am looking to obtain an ultra linear supply which may take as much as one or two minutes to even reach it's correct voltage if need be. The goal is to have the lowest possible noise by sacrificing start up time. I have downloaded Duncan Monro's PS tool but with such large values it is taking forever to obtain the graphs and for some reason it doesn't seem to want to plot them any longer than 40 second or so with the values I'm using.
Also, the question of multiple Pi filters arises. Is it better to have lots of smaller filters or one larger one of the same value?
I have been playing around in Duncan's PS program for a few hours now and while I can see the alterations in things such as charge time, with charge times so long I am having trouble analysing the noise of the output, since the program finishes the graph before the supply ever enters it's linear stage (I know you can increase the diagnostic time).
It would be really cool if Duncan could include some test mode for measuring the noise alone.
I would guess the effect of splitting up the elements is to reduce the density of energies right? Heat for one. Although I can't imagine that being a major problem. Weight would be another, so I need not have one giant 20kg choke on one side of the chasis. But splitting up the elements also increases layout complexity, the number of joints that I will need to solder, greater unsurity and uses up more space. On the other hand, having the elements all as one part means that if the part brakes the cost of replacement is higher and the chances of related failures also increases.
I am guessing, since the entire universe seems to work against things being easy
that splitting the elements up is the correct way to go?
I realise this has been asked before, I think about two years ago, however I feel that I should check this out in a bit more detail before ordering the parts.
I will really appreciate your help with this since no where seems to go into detail for why they use split elements (components, values and abilities I mean).
All the best!
John
I am trying to put together a valve amplifer. I understand filters but I am kind of stuck about the values of capacitors I will be using within the filter.
The amplifer is aimed at being high end and, since the amplification is by valves, a slow start up time is not a real concern.
Obviously, as I increase the capacitors' storage values the charge time becomes longer and the filtration better. But I am having trouble with the reasoning for the values and their effects. By that I mean, for instance, why should I use a larger value for the first capacitor in the filter and smaller one in the second, or the reverse? Why shouldn't I balance the capacitances and so on.
My only reasoning so far is that if I have the largest value infront of the inductor then it will draw the largest initial current to charge itself, so subjecting the inductor to a lower surge value, and so the induction value of the choke itself could be increased. But then I ask, well why not just add the value of the second capacitor after the inductor to the first and have one huge capacitor before the inductor and none after it? I guess another way of phrasing this is, what is the effect of splitting the capacitance around the filter? This is the question that confuses me the most. The induction choke will not introduce ripple itself, it will only act to filter it. Since the filters are defined as LC, that being a choke and then a filter, I would guess there is some kind of reason for there being a capacitance after the choke also.
I am looking to obtain an ultra linear supply which may take as much as one or two minutes to even reach it's correct voltage if need be. The goal is to have the lowest possible noise by sacrificing start up time. I have downloaded Duncan Monro's PS tool but with such large values it is taking forever to obtain the graphs and for some reason it doesn't seem to want to plot them any longer than 40 second or so with the values I'm using.
Also, the question of multiple Pi filters arises. Is it better to have lots of smaller filters or one larger one of the same value?
I have been playing around in Duncan's PS program for a few hours now and while I can see the alterations in things such as charge time, with charge times so long I am having trouble analysing the noise of the output, since the program finishes the graph before the supply ever enters it's linear stage (I know you can increase the diagnostic time).
It would be really cool if Duncan could include some test mode for measuring the noise alone.
I would guess the effect of splitting up the elements is to reduce the density of energies right? Heat for one. Although I can't imagine that being a major problem. Weight would be another, so I need not have one giant 20kg choke on one side of the chasis. But splitting up the elements also increases layout complexity, the number of joints that I will need to solder, greater unsurity and uses up more space. On the other hand, having the elements all as one part means that if the part brakes the cost of replacement is higher and the chances of related failures also increases.
I am guessing, since the entire universe seems to work against things being easy
that splitting the elements up is the correct way to go?I realise this has been asked before, I think about two years ago, however I feel that I should check this out in a bit more detail before ordering the parts.
I will really appreciate your help with this since no where seems to go into detail for why they use split elements (components, values and abilities I mean).
All the best!
John
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.