I mistakenly bought four ±30V AC Transformers of 300 VA capacity, for a 4 channel LM3886 project. I have ±42 V @ 4.15 A. The chips are rated for ±35 VDC. I've looked at linear regulators, LM317, 337 and at higher equivalent up to 5 A. Problem is they are not rated above 40 V max, and are usually not used above ±30 VDC. I was thinking of the LM317HV/LM337HV with a couple of current boosting transistors, 2N3055, and their PNP equivalent for the negative side. The LM337HV is out of production and cost about $25, and would need 4. The LM317HV is still available at around $3. Please, some alternatives.

I'm currently at a standstill until I can find a solution. One thought is run them at 42 VDC, which is their max in the data sheet, but not recommended by T.I. A real big heatsink with some fan cooling? SSR, and just drop the AC by 5 V at input. Buck converters only are good for the positive half I've been told, so that's off the table. Zener diodes are noisy. I'm a newbie, some kits and can read a schematic.

  • \$\begingroup\$ Are you willing to waste 30 W for each rail (60 W, total) just to get the voltage down? (I'm assuming linear as that's the simple way to handle it. You've plenty of overhead voltage -- 7 V on each rail -- so the linear circuit is on the well-worn path and pretty standard. Though a thoroughly well-considered circuit will require a lot more information about your loads.) \$\endgroup\$
    – jonk
    Dec 9, 2022 at 18:55
  • 2
    \$\begingroup\$ You might be able to unwind some turns on your secondaries. Post a picture. \$\endgroup\$
    – Andy aka
    Dec 9, 2022 at 19:18
  • 1
    \$\begingroup\$ @jonk as far as I can tell he won't need regulators if the voltage is lowered to be compatible with the amplifiers. The regulators are a band-aid. \$\endgroup\$
    – Andy aka
    Dec 9, 2022 at 19:22
  • 1
    \$\begingroup\$ Dumb things first: If you bought the transformers from a local store, can you simply return them and buy the correct ones? \$\endgroup\$
    – user253751
    Dec 9, 2022 at 22:16
  • 3
    \$\begingroup\$ 300V to +/-30V transformer? May I ask why 300V, your mains likely is not 300V, right? Where does the 300V come from? \$\endgroup\$
    – Justme
    Dec 9, 2022 at 23:07

6 Answers 6


If you have a toroidal transformer with an accessible center hole you can add a bucking winding to reduce the output voltage without disturbing the factory windings.

Use insulated wire rated for the full RMS current and preferably for mains voltage. Try looping maybe 5 turns first and measure the voltage. You will connect that in series with the factory secondary low-voltage winding. Depending on which way you connect it, it will add or subtract from the factory winding. Adjust the number of turns to get the desired total voltage.

You can test with thinner wire, the open circuit voltage will be the same, but for full load you should use the appropriate gauge wire.

When you are done you can wrap the winding with polyester tape to hold it in place and keep it from potentially vibrating.

  • 1
    \$\begingroup\$ Thankyou for your explanation, seems a lot of people here don't understand the I'm not an engineer, I'm a newbie and looking for just an easy solution if there is one. Yours seems to fit that right. I'll let you know, look for more coffee. Lol \$\endgroup\$ Dec 10, 2022 at 1:24

The answer is obvious, buy the correct transformers for the job.

You say that would cost about $200.00 with tax, but other solutions aren't free either.

  • Linear regulators:- waste about 20W each so will need large heat sinks.

  • Switching regulators:- tricky to build, could introduce noise.

  • Rewind transformers:- potential electrocution or fire risk if done wrong.

  • Auto-transformer(s):- expensive, bulky.

Every one of these kludges has potential for failure, costs time and money, and increases the size and complexity of the project. For $200 you can avoid all that. You may also be able to sell or trade in the original transformers to make the final cost lower, approaching the cost of alternative solutions with a lot less hassle and risk.


Unless you have some good reason to use linear regulators, there are plenty of switching regulators that will do what you need. Maxim, LT, and TI all make buck converters that are good for input voltages of 60 V or higher, many of which have internal switches so they need nothing more than a handful of capacitors and resistors and one good quality inductor.

For the negative side, you can do one of three things:

  • use a chip designed for that specifically, like LT's LT8709
  • use a standard boost converter but run it backwards. If you're interested in this option, I can elaborate more on it; just ask.
  • use an inverting converter to turn the input 40 V into an output -40 V. This would be kind of difficult, as the 80 V input-to-output differential would require some pretty high voltage ratings, but should be possible with the right controller IC.


  • There is a problem getting Vac supply and Z[ohm] speaker specs.

** 300VA is overkill for a 50W Audio IC with 50W loss. 30% more is normal or 130VA, so you could power two channels with one transformer if your input voltage matches your transformer rating, which is not rated in DC.

The only problem with 42V is added heat but if you have 8 ohm speakers that means your Amplifier is 50% efficient instead of 60% which is not a huge deal.

enter image description here


The IC's are not limited to 35V!
They are Wide Supply Range 20V - 94V and rated ( if properly heatsunk) to 135W peak. The PSSR supply noise rejection is 85 dB minimum.

With Vs = =/-42V the spec extrapolates to 50 % efficiency at 50W the maximum waste heat power level.

The Thermal resistance of the heat sink must be like an old PC CPU heatsink < 0.5 'C/W with a fan. The IC is 1'C/W and the sink is electrically isolated, so you only need a thin layer of silver arctic grease and proper screw torque mounted to heat sink.

The power supply voltage will be 10% higher DC when unloaded as the coil resistance loses 10% at the rated VA load for the transformer ( standard practice).

Thus with a total Thermal Resistance of 1.5 'C/W your internal junction at 50W will be 1.5 & 50 = 75 'C while the tab will be cooler. This I consider an acceptable maximum similar to CPU core temp. maximums for long-term reliability.

More details with a simulation to follow.

You must read the datasheet and understand everything to be implemented and avoided like large ground loops, high capacitance cables

enter image description here

enter image description here

With this design using surplus CPU heatsinks you can run the fans slow and keep them cool.


I've looked at linear regulators, LM317, 337 and at higher equivalent up to 5 A. Problem is they are not rated above 40 V max, and are usually not used above ±30 VDC.

NOT a problem.

Using the LM317 as a representative of the group, it is rated for an absolute max differential voltage of 40 V (1999 datasheet). That is the voltage from the input to the output, not the voltage from the input to GND. In your case, the differential is only 7 V.

The 5 A LM338 has an in-out differential voltage rating of 32 V (40 V absolute max.). Minus the 7 V you want to drop, you have 25 V of margin. That's good. With a 4.15 A load current, you have to remove 29 W of heat. That's hot.

It is entirely possible to use a standard (non-HV) LM338 to regulate a 200 V DC source down to 180 V.


Sometimes you just have to say Buck It :)

I'm assuming you do not have a death wish, so I'm assuming you bought isolating transformers.

So it's easy as this: you can put transformers in series. If you do, the voltage across the transformers sums... right?. If you took a 30VAC transformer and put it in series with a 5VAC transformer, you'd have 35 VAC across them... right?

Well, not necessarily. It depends how the tranformers are phased. If they are phased opposite, then you get 30 VAC - 5 VAC = 25 VAC.

Arranging transformers this way is called bucking. Each transformer needs to have the amp capacity for the amps being drawn through it. The required VA is the amps x the RMS volts.

What the heck is VA?

VA is how transformers are rated. It is analogous to "watts" only when every load behaves like a resistor, with current exactly proportional to instantaneous voltage. If a load uses current not proportional to voltage, then watts is smaller than VA. VA is the entire sinewave the delivery equipment must deliver. Watts is the part of the sinewave you actually use. Because of how thermal stuff works, it matters.

What's the deal with voltage, anyway?

I have ±42 V @ 4.15 A.

From your 30V transformer. It sounds to me like you are meeting RMS for the first time.

AC is not one voltage. It's a constantly-changing range of voltages. So how do we identify an AC voltage? Westinghouse had exactly that problem when they took over Edison's failed city electrifications. Since light bulbs were the "killer app" of the time, they defined AC voltage as the AC voltage that will heat up a resistor exactly the same as that DC voltage. So a resistive thing that works on 110 VDC would work on 110 VAC. Like, existing light bulbs!

P = I2R

Feed an AC waveform into that with trig and calculus, and what pops out the other side is for the equivalent of DC voltage V, the AC peak voltage is V * sqrt(2). 30 VAC x 1.414 = 42.4 volts, exactly what you're experiencing.

I gather you're doing a simple bridge rectifier + capacitor style simple DC power supply. That will draw current when the AC waveform is above the capactor's voltage, and will not at any other time. That means you'll only be using part of the sine wave. Remember the discussion about VA versus watts? Totally applicable here.

If it were me...

I would look for a switching regulator capable of taking your existing transformer's output and making the DC voltage you are after. Building your own power supply is a waste of your time, unless you really want to gobble up the theory involved (except transformer+bridge rectifier+smoothing cap power supplies are obsolete, so who cares).

If you are using isolating transformers and not step-down taps, then you can use a bridge rectifier to flip the sinewave into a bouncing-ball DC with a 42.4 volt peak (minus the voltage drop of the two diodes). So you don't need to mess with 84 volts.

  • \$\begingroup\$ What's all this stuff about 300VA to +/-30V what if the input is only 120V and the transformer was rate for 240V \$\endgroup\$ Dec 10, 2022 at 2:40
  • \$\begingroup\$ @Tony I don't think 120V has been mentioned. OP is a novice and is using non-standard expressions. VA is quasi-"watts" as I say. +/-30V seems to be OP's way of saying "30 VAC".The +/- business might reflect OP's lack of understanding of RMS. \$\endgroup\$ Dec 10, 2022 at 2:57
  • \$\begingroup\$ I agree but he should at last say he is using North American power or EU /AS power and where he got the $200 Transfo. \$\endgroup\$ Dec 10, 2022 at 3:15
  • \$\begingroup\$ My mistake in using 300 VA. I meant to say 300VAC. +30/-30 @5 AMPS SECONDARIES. Primary is 120V@60 HZ. 30+30 × 5= 300 VAC. After rectification and taking the diode loss into account, I'm looking at +40/-40 VDC @5 AMPS. Being that I explained my technical level, and the shortcomings in my ability to solve the problem, I find some of the suggestions and explanations to be thoughtful and patient. Although some, act as if it's a bother to respond. Please dont then. We all start somewhere, no matter the endeavor, I do personal training, everyone gets what they need. Teaching is the the best part. \$\endgroup\$ Jan 17, 2023 at 18:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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