Capacitor size formula for ripple current.

[sobell1980]

Hi all

Could anyone help with a formula into micro farads for ripple current capacitor size? O.5 Volts ripple every 20 milliseconds. Any idea what size capacitor i would need. Many thanks as always
Dave.

[jimmc101]

The formula you want is C.V = I.T
Where C is in Farads, V in Volts, I in Amps and T in seconds.
Rearrange to give C = I.T/V and substitute your figures C = I. 20mS/0.5V = I.40mF.
i.e. for 1A, C = 1*40mF = 40mF (or 40,000uF if you prefer)
for 100mA C = 0.1.40mF = 4mF (= 4000uF)
etc.

If you are talking about a half wave rectifier the above is an approximation as the rectifier will be conducting for part of the cycle.

Jim

[paulsherwin]

If you're talking about physical capacitor size, you can't tell very much from that. It all depends on the technology in use when the cap was made.

[G8HQP Dave]

Yes, is the OP asking about cap value for a particular ripple voltage, or cap size to survive a particular ripple current?

For the former, assume 10ms of discharging current for full-wave and 20ms for half-wave. This is not entirely accurate but close enough, given typical capacitor tolerances. Then use C V = I t.

N.B. 'S' is siemens - conductance (otherwise known as mho - inverse ohm). 's' is seconds - time.

[_Clint_]

Sorry to bring an old post up, but just for clarification:

If I am seeing 23V of Ripple on a 350VDC Line with 150mA of Current Draw in the circuit; I would design the circuit so the Smoothing Capacitor was 22uF Minimum ?

...... and on that Basis would the Capacitor see approximately 5% of the Current (Depending on ESR) ..... So I would look for a capacitor with at least a Ripple rating of 10mA @ 100Hz ?

[Station X]

I'm not an expert in this area, but other threads suggest that the ripple current rating of a smoothing cap should be from one to two times the DC current (sic).

https://www.vintage-radio.net/forum/...ad.php?t=37490

[_Clint_]

Nor I,
always going over is better however I am trying to work it from a design point of view. It would be hard to find a capacitor nowadays at that voltage rating to have less than 10x the mA ripple rating above.

Ripple current rating is there to limit heating, if your load current is less than the rated ripple current you can't go wrong (the ripple can not be more than the load). It (ripple current) is less than the load current because about half of the load will be taken by a direct path not involving the capacitor. Rule of thumb, ripple rating equal to half the load current is OK as a minimum.

[Alan Stepney]

If it is a valve circuit do remember that valves have a maximum capacitance that they will work into (the reservoir capacitor.)

That figure, which varies according to the valve type, can be found on the valve data.

[kalee20]

Originally Posted by merlinmaxwell

Ripple current rating is there to limit heating, if your load current is less than the rated ripple current you can't go wrong (the ripple can not be more than the load)

I'm afraid it can!

Consider a peak rectifying circuit, where the rectifier supplies short high-current charging pulses, 1A for 5% of the time (95% of the time the capacitor is discharging). The load is 50mA. So for 5% of the time, the capacitor charges at 950mA, and for 95% of the time it discharges at 50mA.

The capacitor sees a ripple current of root (50mA-squared x 0.95 plus 950mA-squared x 0.05) which I make 218mA, rather more than the load current.

It is a bit of an extreme case, and equal to the load current is generally OK as charging pulses are broader and smaller in amplitude. But the basic premise here can't be justified.

[Radio Wrangler]

The capacitor sees the exaggerated ripple as Kalee has described, and the rectifier sees it too. This is why many valve rectifiers had max values for reservoir capacitors in their data sheet. Increase the C and the droop betwen charging cycles is less, so the charging time is less, so the charging current must be more for the same output. You get less voltage ripple, but more ripple current. The high surges scale up losses too.

Thee are books of charts for rectifier design, and then there's LT Spice which is free and worth learning.

David

I sit corrected!

[_Clint_]

In a full wave rectifier (GZ32) at 50Hz Supply you only have 10mS for a complete cycle, I measure 2.5mS of that cycle is charging the capacitor; thats 25% Charge time.

So at a hypothetical 150mA pull, am I right in calculating a Ripple Current of 37.5mA ? or am I missing something ?

[Radio Wrangler]

No, if your amplifier is taking 150mA of HT current, and I assume it's an amplifier from that current, the rectifier will be averaging 600ma over the 2.5ms charge time and will be averaging 0mA over the 7.5ms off time.

The charge time won't be perfectly constant current, so the peak current will try to be larger still. Semiconductor rectifiers will carry this, but valve rectifiers will exhibit current limiting in saturation, when the current is limited by the peak emission from the cathode.

Rectifier valves have quite a hard life. Their losses are set by the surge current, not by the DC current.

This is where choke input filters score. The action of the choke tries to create a constant current charging effect, the choke stores and releases energy and the charging time spreads, reducing the peak current. With a large enough choke, the charge time extends to the full cycle the peak current reduces and the current ripple reduces. The counter-intuitive bit is that the lower the HT current can drop to, the more Henries of choke you need to stay in continuous conduction. The more the Max HT current can go to sets the saturation current that choke must go to.

So choke input is great for class-A stuff where the HT current doesn't change much, but the phsical size of the choke shoots up for things where the HT current varies somewhat. If the choke is too small and continuous conduction ends at low load, the HT voltage rises to the peak of the AC wave. With continuous conduction, the rectified HT is at the mean of the rectified wave.

So choke input firlers have some problems, and cost money, size and weight, but they really ease the rectifier's job, and make life easier for capacitors, too.

David

[_Clint_]

Hi David, thanks for the reply.


I still have a issue, we are looking at one part of a cycle where I am quoting Amperage Current in RMS and all Capacitors are Rated in RMS Ripple Current.

In investigation it would seem the Capacitors are rated to keep their temperatures steady, Most Modern Capacitors are Tested at 100Khz and 120Hz with a tangent given to work out the multiplier. So that said even if we cannot accurately design the values we can simply trial and error monitoring the temperature of the Capacitor as long as we don't exceed the known parameters.

The Capacitor I am looking at for this is stated at 33uF but the maximum value in the GZ32 Datasheet shows 22uF; however I ran an Panasonic cap here with a Ripple rating @ 100Hz of 275mA it measured 30uF and had a 105C rating; in circuit it reached 31.5C with a room temp of 23C after an hour of running. This Reduced the ripple by 50% (64V - 32V) over a 16uF.

The valve however has a Maximum Voltage at which a maximum current can be drawn, easy not to exceed.............. BUT also and I think more importantly a Maximum input impedance verses output Ripple Current.

The question I had was how would moving these values out of the GZ32's comfort zone affect the valve; I am guessing it loss in efficiency moving the Power output down due to the creation of more heat?

Interestingly, although I would want more time on this to be sure, it looks like the Rectifier is running 1.5C lower at its hottest spot with the bigger 32uF capacitor,

So I started to think of the Tube as having a 3D efficiency map, which would be created with: input impedance, Output Ripple Voltage and Capacitor size, maybe I have moved it further into its efficiency map?

Sorry if I am rambling but I hate it when I cant understand why, which is the question I think should be asked more often

[G8HQP Dave]

You can exceed (by a small amount) one aspect of valve rectifier limits provided that the others are well within limits. Limits include voltage, current draw, reservoir capacitor value and series resistance.

[trobbins]

Your best friend is Duncan Tools PSUD2 program - very simple to configure for nearly all power supplies and filters you are likely to work with. It will simulate the current waveform a lot more accurately than any postage stamp calculation, and identifies the peak diode current level and the capacitor rms current level, to help you compare with datasheet ratings.

You will need to measure power transformer winding resistances, as they are key aspects of valve diode design.

[Herald1360]

The cunning trick for the large choke at low currents problem for pushpull class B amps (or SSB TXs) is the swinging choke. This beast has its core gap arranged so that at lower currents the core is not saturated and the inductance is high; as current increases the core starts to saturate and inductance falls, though not so far as to drop below the critical inductance for the higher current. It's still a bit more costly than a simple choke for the max current but can be quite a bit cheaper than a straight choke of adequate inductance for the low currents that can also handle the highest currents.

[G6Tanuki]

"Swinging chokes" - I last came across them in association with mercury-vapour rectifiers. They always struck me as being a horribly bulky way to cobble together a power supply out of 'temperamental' rectifiers and situations where you couldn't get decently compact/reliable HV smoothing-capacitors.

For the last 3 decades the Kilowatt class-AB1 RF linear amps I've been designing have all used silicon diodes and big banks of series-parallel electrolytics with bleeder-resistor networks across them. No explicit current-limiting at all - I rely on the secondary resistance and leakage inductance of the transformer to provide that (and a bridge of eight 35-amp 1.5Kv diodes). I do use simple 'soft-start' circuitry to prevent breakers tripping at power-up but that's about all.

1200uF of electrolytics charged to 2.2KV need some respect.

[Radio Wrangler]

Yes, the capacitor people know what their ESR is and set an RMS current spec so that temperature rise and rate of loss of water vapour give the capacitor its minimum specified life expectancy. How that RMS value pans out in terms of current waveform is left to the user. It certainly isn't sinusoidal.

As trobbins, says, the Duncans tools software is good and does things a simple calculation can't.

LTspice is free and very very versatile. You can model whatever you want. Do transformers with modelled winding resistances, models of rectifiers, other valves etc and look at accurately simulated waveforms. There are tools to calculate RMS values etc. There's a leaning curve, but the reward at the end makes it worth it.

David