1
\$\begingroup\$

I've got a few questions regarding the power transformer used in typical linear power supplies (with full-bridge rectifier, filter capacitor, and voltage regulator).

  1. A typical transformer datasheet will specify a full-load voltage, which is the RMS secondary voltage measured when the secondary is delivering its rated RMS current. Is it correct to assume that this VA rating only applies when the secondary voltage and current are both sinusoidal? For instance, if the bridge rectifier causes the current draw in the secondary to be pulsed (but the RMS value of the pulsed current is still equal to the transformer's rated value), then can I expect the secondary voltage to be sinusoidal at its rated value? In other words, how does the shape of the current waveform affect the transformer V-I ratings?

  2. How would one go about determining the proper current rating for a linear power supply transformer? My immediate thought would be to approximate the current through the filter capacitor as a periodic sawtooth pulse and find the RMS value of the pulse waveform, then add maybe 50% to account for the fact that the transformer supplies current to the load while the diode is conducting and also to provide a decent safety factor. Is there a better way of doing this?

  3. If you draw, say, 5 A of peak current on the transformer secondary, then for a 10:1 transformer, you'd be drawing 50 A of peak current on the primary. How is standard grid power able to handle such high repetitive peak current spikes without blowing a breaker?

\$\endgroup\$
  • 3
    \$\begingroup\$ Normally, you'd use a 10:1 transformer to produce a smaller voltage on the secondary, for example, 12VAC from 120V mains. In this case, 5A on the secondary is only 500mA on the primary. \$\endgroup\$ – DoxyLover Jun 18 '17 at 3:37
  • \$\begingroup\$ 10:1 turns ratio means V ratio is 10:1 and Current 1:10 and Z = 100:1 So 5A pk_out is 0.5Apk_in. What are you specs ???? \$\endgroup\$ – Sunnyskyguy EE75 Jun 18 '17 at 14:23
  • \$\begingroup\$ Sorry, I don't know how I made such a rookie mistake. I am well aware that the primary current will be one tenth the secondary current for a 10:1 transformer. I have no idea what I was thinking when I wrote that. \$\endgroup\$ – pr871 Jun 18 '17 at 18:56
  • \$\begingroup\$ My specs are 5 V, 0.5 A. I know how to size the transformer for voltage (taking into account all the various voltage drops), but I wasn't sure how to determine the current rating. Can I conclude from your post that derating the VA by a total of 40% for 10% ripple (0.5 V) would be ideal? \$\endgroup\$ – pr871 Jun 18 '17 at 19:04
1
\$\begingroup\$

A typical transformer datasheet will specify a full-load voltage, which is the RMS secondary voltage measured when the secondary is delivering its rated RMS current. Is it correct to assume that this VA rating only applies when the secondary voltage and current are both sinusoidal?

Yes only sinewaves.

> In other words, how does the shape of the current waveform affect the transformer V-I ratings?

For pulse loads derate VA rating at least 30% due to increased RMS current (heating) per average output current which gives more rise in temperature. Often 10% duty cycle pulses for 10% Voltage ripple uses RC=T=16x/f (f= pulse rate , line rate is doubled after full bridge)

main Reason: the copper losses will be Pcu=Ipk^2*DCR, for a copper DC resistance DCR. Thus a large cap loaded DC bridge will increase the losses and rise in temp. more than a sine current load for same RMS output power.

  • for the advanced reader

2nd reason THe output caps charge just before peak voltage and stop at peak voltage. Thus with 10% ripple that means the current is an amplified sine pulse that only starts near 80 deg and stops at 90 deg at a peak level of 10x the decay current into the load, so it has a 10% pulse width with very fast rise time and 10% sine peak shape, with more than 10 harmonics of 2x the line frequency (even and odd) which for laminated steel with high mu, can result in eddy current losses. Load balance affects VAR utilization on tapped transformers.

3rd reason* if the output is center-tapped and 1 diodes used on each leg, AND only loaded on one leg, say V+ and not V- then this results in offset DC current in the shared core and reduces margin to saturation with DC current flowing thru the core secondary ( from half-wave loading) THus primary saturation can occur sooner if one assumed incorrectly they could get full VA on half wave rectified V+. The VA output must be shared with load balance on each winding to utilize the power.

There are some excitation losses from the primary inductance creating a reactive current about 10% of rated current in order to activate the mutual coupling.

How would one go about determining the proper current rating for a linear power supply transformer?

Use VA rating VA/V*70%=Imax dc max and derate further for lower than 50'C winding temp rise even if sine wave.

My immediate thought would be to approximate the current through the filter capacitor as a periodic sawtooth pulse and find the RMS value of the pulse waveform, then add maybe 50% to account for the fact that the transformer supplies current to the load while the diode is conducting and also to provide a decent safety factor. Is there a better way of doing this?

I once used 10% ripple = 16/f and derate VA by 40%. Now I use SMPS. .

If you draw, say, 5 A of peak current on the transformer secondary, then for a 10:1 transformer, you'd be drawing 50 A of peak current on the primary. How is standard grid power able to handle such high repetitive peak current spikes without blowing a breaker?

Breakers near trip Current rating respond in a minute or few. RC low pass filter T= line f/8 thus ~1/10 second so the pulse is too fast to thermally affect breaker or fuse.

Normally Temp rise is the limiting factor with several causes that causes. We expect Voltage output to rise 40% with no load due to sine pk/avg and another 10% due to secondary DCR losses so V_no load will be ~ 50% more than avg.DC at max permissable load. THis makes it very inefficient compared to HF SMPS due to the much lower duty cycle charge periods and long discharge periods, so much greater storage energy and size of XFMR, Cap are necessary.

THere are books by Dr./Prof Keith Billings with easy nomographs that make linear supply design and SMPS design as easy as a recipe for making Won Ton soup. I as a client at Burroughs Inc worked, with him when he was Eng Mgr of Hammond Power Supplies in Canada ca. 80's. He is now in Colorado I believe.

\$\endgroup\$
  • \$\begingroup\$ just check out your second paragraph there please, lest it confuse noobs. Pulse loading, although it causes current spikes and excess heating, doesn't saturate transformers, at least not in the magnetic core sense of saturation, that only occurs with excess input voltage, or DC bias. When educating noobs, and even non-noobs, transformer saturation remains one of the trickier concepts to understand. \$\endgroup\$ – Neil_UK Jun 18 '17 at 4:24
  • \$\begingroup\$ Without a few pages of writing and his studying learning , what's the point? He got turns ratio mixed up. When DC load V ripple is 10% , the cap,diode and XFMR peak current ripple is 1/10% approx or 10x the average current \$\endgroup\$ – Sunnyskyguy EE75 Jun 18 '17 at 14:24
  • \$\begingroup\$ Neil you may edit my answer and explain the efficiency of this +/-12V 1A half bridge transformer and how I derived my derating answer tinyurl.com/y9mhbn94 \$\endgroup\$ – Sunnyskyguy EE75 Jun 18 '17 at 14:32
  • \$\begingroup\$ a better one here with 4:1 ratio tinyurl.com/y7abm3gq still very inefficient W/VAR as you well know. resistors added to simulate loss \$\endgroup\$ – Sunnyskyguy EE75 Jun 18 '17 at 14:46
  • \$\begingroup\$ Thx for inviting me to edit your answer. Removed reference to saturation in 2nd para and put correct reason for derating under pulsed loads. \$\endgroup\$ – Neil_UK Jun 18 '17 at 14:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.