Transformers in parallel?

[G6Tanuki]

I need to build a quick-and-dirty 12V@3A power supply. I have two RS transformers, one is a 504-587, the other a 504-262.

They both each have two 15V 10VA secondaries. But one has a single 240V primary, the other has two 115V primaries.

In times-past I would have followed the Nike "Just do it" approach and wired the primaries/secondaries in parallel. A

And it would probably have worked well-beyond any sort of warranty I'd have offered a customer.

But... as an alternative, I have some rather-well-built Mil-spec heatsinked-bridge-rectifiers [16A@500V] and so my mind has been wandering to the idea of putting the secondaries of the two transformers into separate bridges before combining them into a few tens-of-thousand-microfarads of smoothers.

What would you do?

I want this to be reliable - delivering 12.5V@3A continuously for years without worry.

[lesmw0sec]

Neither to be frank. I'd put the existing units on the shelf for another job and design a unit with a single transformer.

[kalee20]

I wouldn't connect the transformers in parallel. One has 230V primary, the other 240V... so supplied with the same voltage, whatever it is, the secondary voltages will be different. And that'll lead to circulating currents.

If you connect each to a bridge rectifier, again they'll have different voltages so the higher voltage will do all the work, while the lower voltage output will never even turn on its diodes.

Both of those completely ignore the fact that the transformers won't have perfect regulation, so there's a current-limiting / sharing mechanism there. But it's rather undefined.

If you want to do it 'properly,' you'll get a decent transformer. If needs must and you're really stuck, do as you suggest but get an order off to RS, and until it arrives, keep a fan blowing over those transformers!

[G6Tanuki]

Opinions noted... but I'm going to do the Nike-thing and 'just do it' ::

a "decent transformer" from RS will not fit within the rack-profile... which is narrow-but-deep [hence the stacking-two-transformers-end-to-end idea].

I'm going to do some tests, though: the two transformers/bridge-rects run at worst-case demand for 24H, then a "feel-test' [does it feel hot?]

At least I know that the rectifiers are massively-over-rated for the application.

[Cruisin Marine]

Can you not use some low ohmage resistors on the secondary to share/ balance the current?

[Chrispy57]

RS do actually have a 40VA torroidal with 15V secondaries, and the manufacturer's data sheet shows a height of 39mm, so under the 50mm height of your own transformers that will fit in that chassis, if you wish to meet your "I want this to be reliable - delivering 12.5V@3A continuously for years without worry" objective.
https://uk.rs-online.com/web/p/toroi...ormers/7529387

[Neutrino]

A transformer's A.C. current rating needs to be recalculated from the D.C. load current:
http://www.hammondmfg.com/pdf/5c007.pdf

According to this, for a ‘FULL WAVE BRIDGE Capacitor Input Load':

• I D.C. = 0.62 X Sec. I A.C

A 15 V 20 VA transformer is rated 1.33 amp a.c. This corresponds to about 0.83 amp with a full wave bridge and capacitor load.

David

[Lucien Nunes]

I wonder whether the secondaries are bifilar? If so, could you unwind the one with higher OCV to meet the other. The load sharing wouldn't be perfect as you would have two slightly different transformers with the same ratio but it would be close and the circulation could be almost eliminated. Clearly if the two secondaries are in different layers that would be more of a faff, as would unwinding the 240V primary to 230V.

I can't help noticing the considerable voltage overhead there, using 15V AC to make 12V DC. Some 40% of the input power (at nominal voltage) is lost in the regulator, which if it is linear is a bit of a nuisance. This adds to the argument that these are not the transformers you are looking for.

What would you do?

Use a high reliability 5A switcher, or even a parallel redundant pair, unless it needs to be linear for noise reasons. I can't remember the last time I powered anything 12V with 50Hz transformers. The only application where I habitually default to linear is analogue audio, but I'm guessing that's not the case here with a single 12V rail. Even some of my custom audio mixers run on switchers.\

I appreciate that in theory its much easier to achieve high MTBF in a simple linear supply than a line-powered switcher, and if you want endurance over 50k hours they look increasingly attractive, but I have tens of thousands of units out there powered 24/7 and failures are rare indeed.

[lesmw0sec]

The only objection I have regarding SMPS is that they seem to have only two modes: 1) operating perfectly, 2) exploding...

[barrymagrec]

#7

40VA is significantly under size for what you say you need.

[Lucien Nunes]

Good point, staring at us in the face especially when considering a linear regulator. Needs more like an 80VA transformer.

[ms660]

Originally Posted by G6Tanuki

Opinions noted... but I'm going to do the Nike-thing and 'just do it' ::

A good guide for VA rating:

VAC = 0.85*VDC + 2*Vdrop

IAC = 1.65*IDC

Vdrop being the DC RMS voltage across one diode.

Lawrence.

[Glowing Bits!]

Valve linear amplifiers have three separate secondaries on one transformer feeding four diodes and an electrolytic, that's 12 diodes and 3 capacitors combined to create 1200V for the plate.
When connecting secondaries together, they have to be wired in phase or a fire extinguisher is needed.

[Joe_Lorenz]

In addition to the already mentioned concerns: Many years ago I was taught "Only identically built transformers in parallel and only if there is no other choice"
Joe

[G6Tanuki]

As an experiment, I wired both transformers in parallel and then loaded them with a bunch of 12V car bulbs [21-Watt brake-light type and 10-watt sidelight type] then left things running for 36 hours.

At the end, both transformers were noticeably warm, but no more so than the mains-transformer of a typical valve radio would get after an afternoon's use [and I suspect modern wire insulation is rather better than the 60-year-old stuff used on something like a Pye P75's transformer].

My 'finger-test' showed no detectable difference in the temperatures of the cores. I'm happy. Yes the issues of peak currents when feeding rectifiers are noted; I will be using a bridge because it uses both halves of the secondary all the time and results in lower peak currents than the old-fasdhioned centre-tapped-secondary-and-two-diodes.

[Leon Crampin]

Why not wire the transformers in parallel with a miliammeter in series with one leg to see what the circulating current is, off load. If it's a very small percentage of the working load, I would suugest that your scheme is OK.

Leon.

[G6Tanuki]

There's an idea.

I suspect that the difference in currents would be pretty small.

Equally, eachg transformer has two secondary windings that can be wired in series or parallel; I doubt these are bifilar-wound, hence in a given transformer one secondary winding will be closer to the core [and have a shorter length of wire needed to get the required number of turns] than the other seconday winding - meaning that wiring the two secondaries in parallel will surely introduce some sort of current-imbalance between them under load...?

Perhaps I'm just overanalysing this, like when the audio-types measue the resistances of the two halves of a push-pull output transformer primary and go into a panic when they find they are different!

[broadgage]

I would avoid connecting transformers in parallel unless known to be identical.

For this application I might consider an "off the shelf" switched mode supply, these are available very cheaply from a number of suppliers. Primarily intended for operating LED lighting.

A SMPS will be more efficient, a factor of some importance at today's electricity prices if long hour use is expected.
As much as 40 watts might be saved by use of a SMPS, if used for 4,000 hours a year that 160 kwh saved. At today's power prices that could represent in the region of £50.

I have had several running 24/7 for years without a failure.

[G6Tanuki]

While a SMPS would be the 'obvious' way to go, I need this particular supply to be friendly and unobtrusive when operated alongside HF/VHF radio gear - so anything that raises the noise-floor is definitely a no-no.

While I know that SMPS *can* be made instrumentation-grade and RF-quiet, the hassle of buying one, finding it's noisy, and doing the whole sending-it-back-for-a-refund-and-buying-another-which-could-be-equally-noisy thing will invariably cost more of my time/raise my blood-pressure more than any potential energy-saving from SMPS over Linear.

[ms660]

The transformers have a double section bobbin according to the data sheets, parallel connections shown in the 1st data sheet:

https://docs.rs-online.com/3920/A700000008915798.pdf

https://docs.rs-online.com/b846/A700000008857732.pdf

Lawrence.