HI All..
How do you calculate for an AC transformer in VA for DC current draw properly is my question? Usually I'm building things on here that have been built before where the appropriate size of the transformer has been spec'd out and I don't bother to ask how they came to that value.
My primary AC power is 120V and the AC voltage on the secondary will be 9V AC. The final DC voltage will be 5V and the current draw will be max 1.5 Amps
A 25VA 9V AC transformer would put out 2.77 A , I know you want headroom so the transformer isn't stressed, humming, etc. etc.
A 50 VA 9V AC transformer would put out 5.55 A , I know you don't want it too oversized because the magnetic field creates more noise I think.
What value is best?
How do you calculate for an AC transformer in VA for DC current draw properly is my question? Usually I'm building things on here that have been built before where the appropriate size of the transformer has been spec'd out and I don't bother to ask how they came to that value.
My primary AC power is 120V and the AC voltage on the secondary will be 9V AC. The final DC voltage will be 5V and the current draw will be max 1.5 Amps
A 25VA 9V AC transformer would put out 2.77 A , I know you want headroom so the transformer isn't stressed, humming, etc. etc.
A 50 VA 9V AC transformer would put out 5.55 A , I know you don't want it too oversized because the magnetic field creates more noise I think.
What value is best?
You should start with the voltage regulator spec for the minimum input voltage to maintain regulation. Add about 15% to that and set that as the trough of your ripple voltage. That, and the maximum load current, will let you determine the value of the capacitor to use. Use the smallest capacitor that will work! (NOT the largest you can!) That will minimize peak currents in the diodes and transformer. Most voltage regulators will have about 50 to 60dB of ripple rejection, so you can actually tolerate quite a bit of ripple on the input to the regulator.
The peak value of current in the diode/transformer is important to know, as it influences the capacity required of the transformer.
The peak value of current in the diode/transformer is important to know, as it influences the capacity required of the transformer.
Here is an excellent write-up on transformer selection for rectifier power supplies. It discusses the crest factor (referred to as the form factor) and how it influences transformer loading.
https://toroid.com/technical-topics-toroidal-transformers/
https://toroid.com/technical-topics-toroidal-transformers/
Two excellent sources written by Rod Elliott:
Linear Power Supply Design - Part 2: https://sound-au.com/articles/power-supply-4.htm
Beginners' Guide to Transformers: https://sound-au.com/xfmr.htm
They have lots of practical real-world advice and information without being heavy on the theory.
I'd strongly recommend you use a simulator to help get a good feel for how the component choices impact the results - especially the sensitivity of the output to changes in component/voltages/temperature values etc. I use LTSpice as it's free, fairly easy to use (but you have to get used to the quirks as with all software) and there's lots of help and tutorials on the web.
You can if you want to start with a specific regulator in mind, but this may well lead you to a custom voltage transformer or a standard, but somewhat higher voltage one. I started with the OP's 25VA 9V transformer. From Rod's articles, the typical regulation of a 25VA transformer is about 15% from which I calculated .486 ohms for the transformer winding resistance (that accounts for both the secondary and primary winding's resistance). I set the source voltage to be 9VAC x 1.15 x 0.9: +15% for transformer regulation and -10% for low mains voltage).
Choosing different capacitor values we can see the ripple voltage and particularly the minimum regulator input voltage. I used regular diodes but schottky diodes would give a bit more headroom.
You can see that:
a) the transformer rms current and peak diode current vary very little with capacitor value.
b) The required transformer VA rating is on-load voltage x rms current - so in the 10,000uF case is 21.1. This is at minimum input voltage; at maximum (+10%) the rms current is 2.431A, thus the VA rating is (9V + 10%) * 2.431 = 24VA thus the transformer selected should be OK if it is of decent quality and properly rated.
c) The minimum voltages mean that a low dropout voltage regulator is required - a 7805 isn't properly specified at 1.5A but looks to require 7.75 to 8V minimum at 25C let alone over temperature. The LM317 is no better.
d) The ripple current rating of the filter capacitor is important - it needs to be 1.913A absolute minimum (at max mains voltage) and this may determine the minimum capacitor size. Note that many low ESR capacitors only specify the ripple current at 100kHz and have to be derated at 120Hz. I personally would not be happy to operate the capacitors at > 50% rated ripple current, if operated continuously, as the heat generated will reduce the lifetime.
e) The minimum capacitor value has to allow for the tolerance (typically +/- 20%) and for loss of capacitance over life. +50% might be reasonable?
Note that using +/-10% mains voltages probably understates the real variations especially in parts of the world with poor infrastucture or a lot of solar PV. I'd be interested to know what 'real' designers of commercial equipment would use for minimum and maximum values.
Finally, whilst simulators give you a good understanding of what should happen, they are no substitute for real-world testing - the mains voltage often isn't sinusoidal for example.
Linear Power Supply Design - Part 2: https://sound-au.com/articles/power-supply-4.htm
Beginners' Guide to Transformers: https://sound-au.com/xfmr.htm
They have lots of practical real-world advice and information without being heavy on the theory.
I'd strongly recommend you use a simulator to help get a good feel for how the component choices impact the results - especially the sensitivity of the output to changes in component/voltages/temperature values etc. I use LTSpice as it's free, fairly easy to use (but you have to get used to the quirks as with all software) and there's lots of help and tutorials on the web.
You can if you want to start with a specific regulator in mind, but this may well lead you to a custom voltage transformer or a standard, but somewhat higher voltage one. I started with the OP's 25VA 9V transformer. From Rod's articles, the typical regulation of a 25VA transformer is about 15% from which I calculated .486 ohms for the transformer winding resistance (that accounts for both the secondary and primary winding's resistance). I set the source voltage to be 9VAC x 1.15 x 0.9: +15% for transformer regulation and -10% for low mains voltage).
Choosing different capacitor values we can see the ripple voltage and particularly the minimum regulator input voltage. I used regular diodes but schottky diodes would give a bit more headroom.
This really isn't true in the case of smaller transformers where the transformar winding resistance is the dominant factor in the peak and rms currents. In reality, increasing the capacitor value by very large factors above the minimum required for any reasonable value of ripple, has very little impact on the rms current in the transformer. It's rms, not peak current that is relevant for the transformer rating. The diodes obviously have to be rated for the peak, as well as, the average current.
| Capacitor | ESR | Transformer current | Vmin | Vripple | Diode current | Cap ripple current |
| 2,700uF | 98.2 milli-Ohms | 2.287A (rms) | 6.17V | 3.00V (pk-pk) | 4.419A (pk) | 1.726A (rms) |
| 10,000uF | 26.5 milli-Ohms | 2.346A (rms) | 7.60V | .842V (pk-pk) | 4.647A (pk) | 1.804A (rms) |
| 100,000uF | 2.65 milli-Ohms | 2.361A (rms) | 8.04V | .085V (pk-pk) | 4.696A (pk) | 1.823A (rms) |
You can see that:
a) the transformer rms current and peak diode current vary very little with capacitor value.
b) The required transformer VA rating is on-load voltage x rms current - so in the 10,000uF case is 21.1. This is at minimum input voltage; at maximum (+10%) the rms current is 2.431A, thus the VA rating is (9V + 10%) * 2.431 = 24VA thus the transformer selected should be OK if it is of decent quality and properly rated.
c) The minimum voltages mean that a low dropout voltage regulator is required - a 7805 isn't properly specified at 1.5A but looks to require 7.75 to 8V minimum at 25C let alone over temperature. The LM317 is no better.
d) The ripple current rating of the filter capacitor is important - it needs to be 1.913A absolute minimum (at max mains voltage) and this may determine the minimum capacitor size. Note that many low ESR capacitors only specify the ripple current at 100kHz and have to be derated at 120Hz. I personally would not be happy to operate the capacitors at > 50% rated ripple current, if operated continuously, as the heat generated will reduce the lifetime.
e) The minimum capacitor value has to allow for the tolerance (typically +/- 20%) and for loss of capacitance over life. +50% might be reasonable?
No; it's the rms value that matters as this determines the heating in the windings.
Note that using +/-10% mains voltages probably understates the real variations especially in parts of the world with poor infrastucture or a lot of solar PV. I'd be interested to know what 'real' designers of commercial equipment would use for minimum and maximum values.
Finally, whilst simulators give you a good understanding of what should happen, they are no substitute for real-world testing - the mains voltage often isn't sinusoidal for example.
Some thumbnail calculations:
The easiest way is to work with wattage, starting at the load. 5V x 1.5A = 7.5W
Assuming a linear regulator like a 7805, you'd want maybe 7 or 8V from the rectifier/filter.
Assuming 8V, the regulator has to drop 3V at 1.5A, so it will dissipate 4.5W (you'll need to bolt it to a good heatsink to keep it from overheating).
So we're at 12W so far...
Assuming a bridge rectifier, you'll want to add in 2 diode drops of maybe 1V each. Starting at 8Vpeak delivered to the filter capacitors, to puts your voltage requirement at 10Vpeak from the transformer secondary. 10Vpeak x 0.707 ~= 7Vrms at 1.5A for the transformer requirement.
That puts the minimum VA requirement at 10.5VA at 7 Vrms (7Vx1.5A). A higher VA transformer (at the same voltage) would be advisable just so it isn't always running at its limit. Maybe 15VA or so...
Edit: Take into account the excellent advice above re: capacitor values and ripple current. It's golden.
Last edited:
That's ok for a very first approximation but the transformer output voltage is rated for a resistive load in which case the load current and voltage are sinusoidal (and in phase). The filter capacitor causes the current to be delivered by the transformer in short pulses with a peak current much higher than it would be with a resistive load. My simulation showed the peak diode current to be around 4.6A. This causes a greater voltage drop in the transformer's winding resistance - the secondary voltage will look like a sinusoid with a flattened top so the peak capacitor voltage will be less than you calculated.
Neither have you allowed for the variations in the mains voltage which is usually specified to be +/-10% of the nominal. There are additional (relatively small) voltage drops due to additional resistance in the circuit including fuses, PCB traces and the capacitor's ESR.
Then you have to allow for temperature effects - the transformer winding resistance will increase by 15% for a 40C temperature rise further reducing it's voltage. Fortunately the diodes Vf decreases with increasing temperature which helps.
My simulation above shows that a 9V transformer doesn't supply much headroom for the regulator - 7.6V minimum with a 10,000uF capacitor which would be too low for an LM317 (3V guaranteed maximum dropout voltage). Using Shottky diodes would provide an extra 1 to 1.4V margin.
Sorry, this is definitely wrong - the transformer AC rms current is much higher than the DC load current. That's because the current is peaky and the heating in the winding resistorstance is proportional to I^2R. A typical rule of thumb is that the transformer rms current will be around 1.8x the DC load. For small transformers, like this one, that reduces to approx 1.6x but can be over 2x for larger ones. My simulation above shows you would actually need a 9V, 21 to 24VA (depending on mains voltage) transformer.
Thanks. I get the feeling the ripple current rating is often overlooked but it is vitally important. For electrolytics below 50V you will typically find the ripple current rating is related to it's physical size only - not the capacitance or voltage - see Panasonic FR series for example (but note that they only specify ESR at 100kHz - at 120Hz it can be 3x larger - or more).
Maybe. It would almost certainly be suitable from a power capability assuming your transformers are properly rated (ie. for continuous operation at 25VA). You state that the maximum load is 1.5A, but we don't know if that could be continuous or intermittently. Personally I would be uncomfortable specifying a component knowing it could be operating close to it's rated capacity for extended periods. Long term experience of failure rates in the field will tell you if you have got it wrong, whether under or oversizing. A lot will come down to trust in your transformer supplier - or your purchasing dept not to cheap out buying no-name components from suppliers that have a tendency to be 'economical' with their specifications.
If it can operate at 1.5A for extended periods you might also need to consider the temperature rise of the transformer. Surprisingly many manufacturers don't specify it. Vigortronix specify +40C for their 30VA toroidals, but Talema state the design rise of their toroidals is 60 to 70C. Some state a maximum ambient of 40C. Only you know what is acceptable for your product. Choosing a higher rated transformer will reduce the heat dissipation of the transformer but it's likely better regulation would mean a higher average voltage into the regulator increasing it's power dissipation.
From a minimum voltage perspective my simulation showed about 7.6V when the mains input is 10% below nominal. Again only you know if that is sufficient - as I said before it isn't enough for a 7805 or LM317 to guarantee proper regulation, Oh I also forgot to mention brownouts - if the regulator only gets 7.6V and the mains drops out for a few cycles, as it is want to do from time to time, can your product continue to function properly? Customers tend to be a bit stuffy about marginal designs...
Be aware that linear power supply design might look simple superficially, but it actually isn't - especially determing the AC rms current the transfor has to deliver, which is why simulation is such a good start. Of course if your transformer is well oversized then yes, it can be fairly simple. I strongly suggest you skim through the Rod Elliott articles I linked in my first post - a lot of it won't be too important to you but hopefully you can spot those parts that are.