9V to 48V DC-DC converter


Konstantin Dimitrov over at Electronics Lab writes:

This project is a 8-12Vdc to +48Vdc DC-DC converter based on MC34063 switching regulator.
It’s a simple project of a DC-DC converter to make a phantom power supply for professional microphones. It can deliver 15-20mA at 48VDC. It ‘s based on MC34063 DC-DC step-up, step-down and boost converter. Input is between 8-12V DC and the output +48VDC/10-20mA.

Join the Conversation


  1. There should really be power planes instead of traces, especially for the input capacitor/inductor/switch loop. I can’t imagine that having good stability/noise rejection, which I assume you need for a microphone.

    1. Relax, it’s an ancient but still popular chip that runs at 33kHz. The data sheet have PCB sample layouts that are one sided. No biggie, it’s an old chip anyhow. Still, I found it in one of my WiFi routers too. It’s a stock boost circuit from the data sheet, components adjusted. There is an LC filter at the output which the data sheet claims will limit ripple to 40mVpp. Is it smooth enough? I’m not an audiophile, I wouldn’t know.

      Good catch though by Philippe below. I’m amazed that his missed the max ratings even though he presumably have read the data sheet. Or does he have some new second-sourced part?

      1. The switching frequency of the internal switch is 33 kHz. What is more important is the actual frequency content on the input, switch and output nodes.

        One tip: Look at were the voltage feedback and the output filter are connected. As it is now, the feedback, which is a high ohm circuit with low current, connected to a high current trace where there might be noise. A better solution is to connect the feedback directly to the + on the output capacitor and route it separated from the diode-output capacitor trace. The same goes for the output filter. If you connect the filter choke to the output capacitor from the DC/DC you will get better noise filtering.

      2. Strange, I can’t see a reply link to Niklas’s comment. Anyway, personally I would prefer to pay attention to layout for much faster modern switching ICs, but for MC34063, I would just expect a certain amount of ripple. Whether its suitable for pro audio I have no opinion. It’s 33kHz typical, the switch is a darlington, the output load is 10-20mA. It’s not cutting edge. Almost any kind of layout is gonna work, since I doubt this is going for EMI testing. If I’m gonna pay that much attention to layout, then I might as well start with a ground plane. ;-)

      3. 10-20 mA on the output will be a lot more on the input: Pin=Pout*Efficiency. Also consider the ripple current with a sawtooth shape that is superimposed on the DC level. The switch transistor is turned on by the output voltage comparator and turned off by the peak current detector, controlled by R2 and the peak current limit (0.30 V = 0.33ohms*Imax). The energy storage coil must be able to handle the peak ripple current to not enter saturation.

        Regarding ground plane. The 3D view of the PCB shows a ground plane, but not a very efficient one. Have a look at the return path for the coil charge current, i.e. pin 2 to the input capacitor. To have any impact on the performance, a ground plane needs as much attention as the other routing, if not more. It is not something that is added afterwards by just adding a GND polygon pour.

      4. I appreciate your critique. I’ll just assume the part he picked has enough margin not to enter saturation. Analyzing return paths? For 33kHz? I dunno about you or Dimitrov, but I’m not gonna spend time drawing the most perfect PCB for a MC34063. It’s a 20 year old IC. Cheers ;-)

      5. “Analyzing return paths? For 33kHz?”

        Have a look at the actual current and voltage wave forms for a step up or a step down regulator. Is there anything close to a sinusoidal signal, except for ringing at discontinuos current mode? Once upon a time there was a man named Fourier…

        I doubt that you are going to find a 68 uH inductor that both fit the selected footprint and has a saturation current of 1 A or more. <=25 uH, no problems, but above that.. Above the saturation current the inductor stops acting as an inductor and becomes almost a short circuit. The sawtooth shaped current waveform will get spikes that can pose a risk to the electronics itself. The current limiter can trigger but is it fast enough to prevent damage to a saturated switching transistor?

        Select an inductor (Henry) instead of a choke (ohms@Freq) that has a self resonant frequency (SRF) above the switching frequency and high enough saturation current to handle the peak ripple current (even at Vin extremes and with -20% of inductance).

      6. Gee, I have read Maniktala cover to cover. I stand by my views. Still, I suppose you feel you must finish this thread with the upper hand. All right, you are right in everything you pointed out. Awesomely right. Brilliantly argued. Cheers ;-)

  2. Plus, those mics can usually work from a lower voltage too.
    I usually just use a few PP3 9V batteries in series. Last ages.

    The board above needs a 9V supply anyway, so either that is from a battery anyway (in which case why not just use two or three PP3 9V in series instead of that board), or is from a mains adapter (in which case why not just use a higher voltage output mains adapter).

    1. MC34063A is good for >=3.0V operation. However, its PWM driving waveform is about 6:1 fixed, so the max boost is a little under 6X I guess, which may account for the Vin spec. A 12V power brick would work, or pull 12V from a PC PSU.

      I suppose the IC didn’t blow up for him, well, not immediately anyway, but I would not inflict this on anyone. The max voltage issue can be fixed with an external NPN.

  3. I hate to nit-pick, but why do people seem to hate mounting holes so much these days? We need to start a public awareness campaign for these nearly extinct creatures.

Leave a comment

Your email address will not be published. Required fields are marked *

Notify me of followup comments via e-mail. You can also subscribe without commenting.