I wish to put a Bluetooth receiver into an old hi fi amplifier. There are +/- 33V rails available and the Bluetooth device needs 3.9V and about 100mA maximum.

I was going to use an LM317 but I think it's getting close to it's input limit at these voltages or it will need some serious heatsinking.

Any ideas on how to drop the volts reasonably without allowing noise from the Bluetooth device back onto the 33V rails?

I thought of either a resistor divider prior to the LM317 or maybe a suitable (9V) zener?

Any help is appreciated

  • 3
    \$\begingroup\$ Buck converter! \$\endgroup\$ – winny Jan 22 '17 at 19:32
  • \$\begingroup\$ A notorious source of noise, no? \$\endgroup\$ – briscaF1 Jan 22 '17 at 19:39
  • 4
    \$\begingroup\$ Only if you don't don't it right. \$\endgroup\$ – Matt Young Jan 22 '17 at 19:41
  • \$\begingroup\$ Assuming I use a LM2596 module, what is the best noise reduction technique? \$\endgroup\$ – briscaF1 Jan 22 '17 at 19:45
  • 1
    \$\begingroup\$ Quiet depends on ground current effects on pickup and radiated noise pulses , which at 100kHz can be done easily with care. Efficiency of >80% is easy with SMPS while linear < 10% so your requirements are inconsistent. \$\endgroup\$ – Sunnyskyguy EE75 Jan 22 '17 at 21:03

Since it is an old hifi, I suggest to use a linear regulator and not a switching device / buck converter to avoid noise pickup into the old hifi circuits. You need to be able to loose 3W of energy in the regulator on the other hand. Find a high input voltage regulator in a TO220 enclosure and apply a proper heatsink. To avoid noise from the bluetooth circuit add a LC filter (eg. 330uH and 1000uF) in front of the regulator and keep the antenna as far away as possible from other circuits.

  • \$\begingroup\$ Perfect. I did some calculations earlier and also came up with approx 3W to dissipate so this gives me some confidence. I think a heatsink should not be an issue given the amount of space in the case and the filter sounds just the job. Thanks. \$\endgroup\$ – briscaF1 Jan 22 '17 at 20:15
  • \$\begingroup\$ @briscaF1 Most LM317's have output ranges up to from 1.25V to 35-37V, so 33V as input should be no problem. I also agree that linear may be the best way to go due to noise. Only problem I see is the dissipation of 3.3W. \$\endgroup\$ – StainlessSteelRat Jan 23 '17 at 17:11

This is a classical case for a step-down / buck converter.

Yes, switching something generates noise, but as everything, we can determine that and technically limit it to a level below significance.

  • Use a modern step down converter. TI's SimpleSwitcher modules are probably a good, no-worries, everything included package.
  • Make sure the switching frequency is well above what the signals relevant to your device are designed to pick up
  • Use appropriate noise rejection: Proper decoupling capacitors (as you'd need them on a linear supply, too), and ferrite beads designed to get rid of higher frequencies are probably sufficient to deal with the noise that still comes out of well-designed switch-mode regulators (which is, in fact, relatively well-controlled!)
  • Since you probably don't want your bluetooth device to directly couple into your most sensitive amplifier circuitry, you'd probably move it furthest away from the input signal section, anyway. That's a good place for a power supply, too. Generally: close to the device's main power supply should work!

I'm a lazy person. I simply went to the TI website, clicked myself through to the power modules (step-down), sorted by price, selected the cheapest module that would do my job (26–42V in, 3–5V out, 0.5A), and then clicked on the "interactive design" button. That led me to another option, which would require me to buy an inductor, but would be cheaper.

If you're really worried about noise:

You often see circuitry that first steps down using a switch-mode buck converter for the "large" voltage difference, and then does the remaining 0.5 V to 1.5 V difference with a LDO (low dropout) regulator. You'll find that in multi-supply RF machinery, for example, where noise is a lot more important than here.

However, you're, in the end, driving a device with that that probably has some internal means to deal with small voltage deviations as well as it's digital anyway, so I think you might be overthinking this – simply go for the switch-mode supply.

  • \$\begingroup\$ +1 for the lazy person and LDO followed by switching circuit \$\endgroup\$ – Umar Jan 23 '17 at 13:58

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.