I'm using a transformer rated at "0-9v, 0-9v, 0.85A" to power an AD1865 dac chip (which I assume consumes just milliamps) - how do I calculate the VA rating of the transformer?
(I'm assuming 50mA max is the most it will consume, which would be 50mA/240v AC = 12VA transformer required).
(I'm assuming 50mA max is the most it will consume, which would be 50mA/240v AC = 12VA transformer required).
aha! my missing link was the "power factor"
Thanks Tony.
The reason I am seeking the VA rating is to upgrade the smoothing caps. They are currently 2,200uF, but I have some 22,000uF ones laying around that should give better performance, however need to determine whether the transformer can handle the extra stress associated with the increased capacitance (if there is any). 20va should be fine, would you think?
Thanks Tony.
The reason I am seeking the VA rating is to upgrade the smoothing caps. They are currently 2,200uF, but I have some 22,000uF ones laying around that should give better performance, however need to determine whether the transformer can handle the extra stress associated with the increased capacitance (if there is any). 20va should be fine, would you think?
yes, the impedance of such small transformer will probably be enough to limit start-up currents...take note that with such high value filters, it will take a while for the voltages to ramp up to working voltages.....
the i/v stage of the dac uses a separate transformer (0-220,0-9) - so the tube section takes a while to warm up anyway
I might change out the caps (will increase the 220v from 150uF to 470uF) and then see just how the transformers perform under load for a few hours, check they don't run too hot.
I recall reading that after certain level (roughly 10,000uF) there is not much return value on increasing smoothing capacitance - I wonder why some preamps and dacs have > 100,000uF - is there any way of justifying this expense?
I might change out the caps (will increase the 220v from 150uF to 470uF) and then see just how the transformers perform under load for a few hours, check they don't run too hot.
I recall reading that after certain level (roughly 10,000uF) there is not much return value on increasing smoothing capacitance - I wonder why some preamps and dacs have > 100,000uF - is there any way of justifying this expense?
The quick and sort of correct answer is that the VA rating is the same as output current X volts, however (and this is a big however) that is only true if you put a pure sine wave in and get a pure sine wave out. In this age of digital devices, switch-mode power supplies, light dimmers, brown outs and other power line anomalies, that is rarely the case. There are others here who are certainly a lot smarter then me about this, but in your case that transformer should have plenty of capacity for that purpose. As for the actual method to calculate VA, maybe one of our better schooled members can chime in, I just don't know how to do that.
Mike
Mike
you can look here about capacitance multipliers : http://www.diyaudio.com/forums/pass-labs/209-capacitor-multiplier-supply-aleph-3-a.html
It really depends on the ripple voltage - a big power amp might need 100,000uF to keep the ripple voltage low enough. But in the case of dacs and preamps where the draw is normally very low I'd say 100,000uF is a waste of money yeah. If low ripple voltage is a target better to put the money into regulation circuitry to reduce it rather than huge caps which only decrease the power factor and hence increase transformer heating.
LOL, no you won't need 12VA or more transformer.
yes 10,000uF is absurd for a chip dac using 50mA. Transformer -> rectifier bridge -> 2 x 220uF capacitors (one per rail) -> positive & negative regulator -> 10 uF ceramic capacitor (one per rail) is all that is needed, plus decoupling near the chip pins.
The initial calculation was wrong, it's not 50mA @ 240VAC, it's 22mA/+5V rail, -23mA/-5V rail... 45mA * 5V = 225mW, which is also listed in the AD1865 spec sheet as the typical power consumption.
So, you have 45mA sum across both rails, so 0.045A * 9V instead of 5V = 0.405W
0.405W/0.7 = 0.6VA transformer needed.
yes 10,000uF is absurd for a chip dac using 50mA. Transformer -> rectifier bridge -> 2 x 220uF capacitors (one per rail) -> positive & negative regulator -> 10 uF ceramic capacitor (one per rail) is all that is needed, plus decoupling near the chip pins.
The initial calculation was wrong, it's not 50mA @ 240VAC, it's 22mA/+5V rail, -23mA/-5V rail... 45mA * 5V = 225mW, which is also listed in the AD1865 spec sheet as the typical power consumption.
So, you have 45mA sum across both rails, so 0.045A * 9V instead of 5V = 0.405W
0.405W/0.7 = 0.6VA transformer needed.
Last edited:
Yeah but do we really want to be using the trafo at its maximum operating temperature? I suggest we don't - the copper losses are higher when the wire's hotter. So how about a 1.2VA one?
^ True and agreed, it shouldn't be ran at full power, but I doubt we could even find an actual/existing 0.6VA 240VAC/+-9V CT transformer so for practical purposes the transformer used would probably be the smallest one available at a reasonable price.
Yes and for a DAC the smallest one which provides the power at a reasonable temperature also has the virtue of lowest capacitance to mains. Which normally translates to lowest mains-borne noise getting in which hopefully means better sound 😀
Well - certainly good to find that this has sparked discussion - in the time that I last posted and subsequently viewed these replies, my soldering iron has been at work and I have replaced four caps (2,200uF -> 22,000uF [x3] and 150uF -> 470uF).
The result - the dac is fired up and appears to be idling without any worries, yet to test the sound for any major differences (none expected).
So, in essence, what this means is I have a dac with over-sized transformers and a lot of smoothing capacitance - probably unnecessary, but might assist in the re-sale value!
The result - the dac is fired up and appears to be idling without any worries, yet to test the sound for any major differences (none expected).
So, in essence, what this means is I have a dac with over-sized transformers and a lot of smoothing capacitance - probably unnecessary, but might assist in the re-sale value!
Hmm, I seem to be a little sleepy and ignored AC-DC voltage conversion in my equation so it might be 0.7VA instead of 0.6VA before adding a margin of overspec to reduce heat... but suffice to say that only a very small transformer is needed.
PSU caps which are much bigger than necessary won't only heat up a transformer but also run the risk of making buzz worse because the charging pulses will have more high frequency content so magnetic induction into signal circuits will work better.
Use calculation and/or modelling to choose the right value. Then a variation of x2 or x3 either way won't make much difference - electrolytic tolerances will cover some of this range anyway. Beyond that you are creating difficulties. For a good clean supply err on the low side, then add extra smoothing and make sure the grounding is right.
Use calculation and/or modelling to choose the right value. Then a variation of x2 or x3 either way won't make much difference - electrolytic tolerances will cover some of this range anyway. Beyond that you are creating difficulties. For a good clean supply err on the low side, then add extra smoothing and make sure the grounding is right.
This is an interesting point - high frequency pulses are certainly not desirable in a DAC! A capacitance of 22,000uF per 5v rail on the AD1865, whilst on the superfluous side of things, should not be of too much concern (the transformer appears satisfactorily rated, but alas I do not have testing equipment to view the charging pulses - I've also put 0.1uF film caps on each diode to assist with issues like this).
But more to the point - what is the design intention behind something like this, with 130,000uF capacitance?!
there seems to be an emerging line of thought, that you can can not overbuild psu's....thus we see large numbers/values of caps.....
i remember an article written by walter jung about the subject many years ago....
i remember an article written by walter jung about the subject many years ago....
My guess is that its more a marketing intention that guided that design. They want to appeal to people who think 'one cap good, three caps better, 30 caps amazingly excellent!'. 😛
Some of Rod Elliott's articles would appear to suggest the same (well, at least to say, that if your transformer is rated accordingly, higher capacitance can't necessarily do much harm).
Have a look at figure 4 on this page:
Elliott Sound Products - Linear Power Supply Design
In particular:
"While it may seem that a higher capacitance should draw larger peak currents, it must be understood that the larger values of capacitance discharge less between charge pulses, and ultimately require exactly the same "top-up" energy as a smaller cap. This effect can be seen just by looking at the ripple voltage figures - with lower ripple voltage, there is less voltage change when the diodes conduct, so the peak current and waveform remain relatively constant."
Note that he suggests 4700uF is adequate for over 1A load, and 50000uF is perhaps a sensible upper limit. What was your current load?
I could not see where he specifies series resistance; this sets the peak current.
I could not see where he specifies series resistance; this sets the peak current.
- Status
- Not open for further replies.