If you are using less than 25v, better go with the FM series rather than FC, they have a lower ESR and cost half the price, even at 100 units.
2.700uF 25v
Digi-Key - P12385-ND (Manufacturer - EEU-FM1E272)
Looking at the simulations of my proposed supply, it looks like the first C and first R should be rated alot higher than the subsequent C's and R.
You get some huge peak currents in the first C and R. More than 5 x 2200 uF high ripple current caps can take.
7,2 A output current, gives peaks of around 18,5 A in the first C and R.
You get some huge peak currents in the first C and R. More than 5 x 2200 uF high ripple current caps can take.
7,2 A output current, gives peaks of around 18,5 A in the first C and R.
Last edited:
Makes sense.
But I have a question about total ripple current of a CRCRC supply, can you add the ripple current of all the caps together or are you in the end limited to the ripple current of the last C?
PsudII allows you to examine the ripple current of each group of C.
Each group of C can be a single cap or an assembly of parallel caps.
You will see that the ripple current in the first C is quite different from the ripple current in the next or last C. Do not add the ripples, They are predictions of what happens at different parts of the circuit.
As far as the amplifier is concerned, the predicted ripple voltage on the last C is what the amp must tolerate/attenuate before it reaches the speaker.
Each group of C can be a single cap or an assembly of parallel caps.
You will see that the ripple current in the first C is quite different from the ripple current in the next or last C. Do not add the ripples, They are predictions of what happens at different parts of the circuit.
As far as the amplifier is concerned, the predicted ripple voltage on the last C is what the amp must tolerate/attenuate before it reaches the speaker.
That I know, but an amplifier needs a certain amount of current for any given output power. Total ripple current of the caps should be higher than the required output current as not to overload the caps.
That was basically what I was asking. Does the ripple current off all the caps factor in the total or is it only the last C's ripple current that matters.
Or maybe I'm just completely wrong here.
PsudII is simulation software.
It models and predicts what the operating conditions are for your particular circuit.
It tells you that the ripple current through the first C is Xamps, when Yampsdc are drawn from the output.
You then select capacitors that together (or alone) provide sufficient ripple capacity to prevent overheating in the chassis that you provide.
The ripple in the last C is not usually an issue in CLC PSU.
The last C provides transient current to meet demand by the amplifier. That is not a continuous demand. It does not cause overheating in the last C, if the last C has been chosen appropriately to suit the amplifier and speakers.
no and yes.
No, the first cap provides almost none of the speakers demand for transient current. The first cap smooths out the rectified AC. That is it's primary purpose. The demands placed on it for this duty are first and foremost High Ripple Capacity. Read what Psud tells you about the voltage on the first C. Read what Psud tells you about the current through the first C.
Yes, you can use a lower ripple capacity capacitor for the last C. Almost all the transient current demand from the speaker via the amplifier is met by this last C. Because this last C is so intimately involved in the speaker transient current demand, it has a fairly big influence on the sound coming out of the speaker.
It is this sound quality issue that determines the final choice for the last C. Ripple can be ignored for this last C.
Read what Psud reports on voltage on the last C. Read what Psud reports for current through the last C. Note these two reports and the previous two, are for the condition when the model is delivering a continuous DC current.
So the first and possible second C if you have that, does the "heavy" lifting, while the last C just delivers the transient current needed.
Looking at the last C it doesnt have alot of current going through it, compared to the first and second C.
Increasing the first C makes the last C have even less current through it.
Current capability of the C's should be in order of highest ripple current, 1st C>2nd C>3rd C
Looking at the last C it doesnt have alot of current going through it, compared to the first and second C.
Increasing the first C makes the last C have even less current through it.
Current capability of the C's should be in order of highest ripple current, 1st C>2nd C>3rd C
Hi,
the starting peak ripple is very short duration, so can be ignored for ESR induced heating.
However, normal capacitors must be limited in rate of charge, i.e. maximum peak amperes of charge current.
Special high charge/discharge rate capacitors are made for that type of duty. They are generally for industrial use and I have no idea how they "sound".
A spot welder would use these high discharge caps.
the starting peak ripple is very short duration, so can be ignored for ESR induced heating.
However, normal capacitors must be limited in rate of charge, i.e. maximum peak amperes of charge current.
Special high charge/discharge rate capacitors are made for that type of duty. They are generally for industrial use and I have no idea how they "sound".
A spot welder would use these high discharge caps.
yes, the first C is doing the heavy lifting of converting AC to DC. Once first C has achieved that reliably it's selection and job is done.
Now move to the last C in the model.
A ClassA amplifier draws a continuous quiescent current. The CCS in the model approximates that quiescent current draw.
Now start feeding current to the speaker. the current draw varies. That is not modeled in PsudII. If we were to change to CCS to a squarewave demand varying from 0amperes to 5amperes, the model predictions would change. First C would hardly change, it smooths the rectified AC, the last C meets the transient demand of the speaker. It supplies almost all the transient current sent to the speaker.
Can we set up a variable current demand in PsudII?
If we can you would be able to read I through the last capacitor and you would see that it nearly matches the speaker demand. Progressively less of the speaker transient demand filters back through the CRCLCC of the PSU.
That is why I suggested CC for the last C, to meet the demand of the speaker, this is crucial to good sound quality.
It can't be for 500ms !
The length of a whole half wave of mains power is either 10ms or 8.3ms.
The longest pulse while trying to charge the first C from the transformer is very roughly half of those time intervals, say 4 to 5ms.
A soft charge circuit is one way to control the peak charging current. High Power NTC are specifically manufactured for precisely this duty and they are placed in the secondary circuit, not the primary circuit, where the soft start is located.
Which leads to the question - what would be the ideal choice for cap in that position? Also, would there be any benefit to having that pair of caps (the + and - rail) connected at the amp pcb's rail inputs? I.E., having it mounted physically as close as possible?