thoughts on LM power regulators vs. zener diodes

[rmunson]

Hello folks -
I'm working on a simple power supply for a project -
I've got a LM7812 providing 12 volts to the main section of the circuit, but I'd like to drop that regulated 12 volts down to six or nine for another part of the device --

Can I just send the regulated 12 volt current from the main section into another power regulator (ie LM7809 or 7806) and drop it down to 9 volts or 6 volts --- or would it be better to use a zener diode?

Is there a problem if you try to drop power too far with a linear power regulator? What is the recommended limit?
Thanks!
Reed

 

[mbz]

I would go with using the LM7809 and LM780... Using a zener will drag the whole rail down including the LM7812 output
(if I understand correctly). Refer to the regulators datasheet regarding max input voltage and supply current. While it's ok
to chain regulators you may wish to keep things "separate", that being, have the regulators in parallel, provided that you
don't exceed max inout voltage. Useful when looking at separate supplies to digital and analog stages.

 

[sregor]

You can use a zener with a resistor. Make sure it is low enough value to supply enough current to the circuit. Regulator will work good too. Be aware that regulators require a certain overhead (read the spec sheets - I guess 3 or more volts - so a 9 volt regulator may or may not work off of a 12 volt line. Most regulators upper limit is about 40 volts in, so not a real problem the other way....

 

[gadget73]

The regs will usually give better regulation. I'd agree on paralleling them though, run the unregulated input to both, and get your 12v and 9v output seperately. It will reduce the current load on the 12v and make sure there are no worries about insufficient voltage for the reg to work right.

 

[rmunson]

cool - thanks for the input - I'll drop in a LM7809 at a separate section -- I'm getting my power off a 12 volt unreg ac/dc power adapter - it's reading 16V off the adapter -
the first 7812 is powering a small preamp/amp that are rated for up to 18V
the secondary section of the device is a simple doorbell type ringer -
I'll set up a 7809 off the unreg V+ from the power adapter --

out of curiosity, I'd like to see what the zener does -- can I drop it off the regulated 12+ -- or should I drop it off the unregulated input?

Thanks!

 

[gadget73]

Depends how you do it. If you use the zener as the pass element, meaning the voltage flows through the zener to your bell, it will drop the input voltage by however many volts the zener is. If the input sags, the output does too. Running it from the regulator should keep the output fairly stable, but of course your 12v regulator will also see the extra load of the bell.

If you run it as a shunt regulator, meaning drop resistor from voltage to zener, then zener to ground, and the load taken from the junction of the zener and the resistor, the output will sag if there is too much current drawn through the resistor. The trick there is you have to run enough current to allow the resistor to be small enough that it won't limit the current to your bell too much, but not so large that it exceeds the wattage rating on the zener or the resistor. Its also going to constantly waste an amount of power by generating heat. Doing it this way is more of use to limit max voltage but not hold a fixed minimum voltage. Its also best for low current loads. If you build a power supply out of individual parts, using a zener set up like this as your reference voltage source is pretty typical.

 

[Powertech]

Personally I would use LM317 regulators. The control circuits are quite simple and they are much quieter than the 78xx series regulators or Zener circuits. Agreed about setting them up in parallel rather than daisy chaining with only 3 volts differential. For most regulators 3 volts diff is a minimum.

 

[rmunson]

Thanks Powertech - yes another friend recommended going this route -
I was a bit confused at first but eventually figured out how to breadboard the circuit - I was confused because it's not left to right as you look at it - but rather right to middle? This vid helped me:

I've got my 2ndary power feed sorted out!
Thanks for the help everyone!
Reed

 

[fred soop]

Personally I would use LM317 regulators. The control circuits are quite simple and they are much quieter than the 78xx series regulators or Zener circuits. Agreed about setting them up in parallel rather than daisy chaining with only 3 volts differential. For most regulators 3 volts diff is a minimum.

I'm doing a preamp redesign (entire amplifier section of a receiver actually) and went with zener and transistor for regulation of the ±15 V supplies for the op amps. I have also heard that the 78xx and similar are noisy, but several well respected audio people (Self and Rod Elliott) use them regularly with no comments about noise.

 

[Binkman]

What's the Project?

 

[fred soop]

The project is a rebuild of a Fisher 600-T receiver. One was restored with a few improvements, including parts of the power supply and the phono preamp. Thread is on the Fisher forum around 2014. The other is presently on the bench and is getting up to date replacements for all of the amplifier sections. The fragile power amp blew up years ago and was replaced with a redesign that was what I was capable of doing in 1974. That is being replaced again with a much better design based on information from such experts as Bob Cordell, Douglas Self, Rod Elliott, Andrew Russell, and the late W. Marshall Leach.

Here is the regulator section. The left has a 22 V transistor regulator (multiplex decoder), then a chain toward the left for 19 V, 14 V (IF) and 12 V (front end). On the right are 2 15 V regulators for the control preamp with heat sinks and a 15 V regulator for the phono preamp. The right half of this board is duplicated with opposite polarity for the negative 15 V supplies. That is mounted on standoffs using the 4 holes not in the corners.

The transistor circuit uses a resistor, zener, filter cap to feed all 3 transistors. The transistors have a dropping resistor on the collector and a filter cap on the output. Actually simpler than the 3 terminal regulator with all of its external protection diodes.

1831

1832

 

[rothwellaudio]

I have also heard that the 78xx and similar are noisy, but several well respected audio people (Self and Rod Elliott) use them regularly with no comments about noise.

Zener diodes are noisy - some very noisy. Staying close to 5V gives you the lowest noise and the noise increases for higher zener voltages.

Douglas Self has a lot to say about regulator chip noise in his Small Signal book. 78 series regulators are noisier than LM317s but a lot less noisy than zeners.

 

[fred soop]

I'll have to review the Small Signal book again. My regulators have 47 uF across the zener and 47 uF at the emitter of each output. Additionally, the control preamp boards have 47 uF at the power input and all the op amps have a 0.1 uF between + and -. Hopefully, we should be good. But I do have another potential source of noise that might be higher than necessary. Being a vintage unit, I'm stuck with the pots (tone controls, balance, and volume) that are very high resistance for this type of modern circuitry. Tone pots are over 100K, volume and balance are 50K. These should all be 10K in a modern circuit. So, there will be additional Johnson noise. In fact, even though the balance pot is of a special design with low resistance on the top half, I was met with a surprise attenuation when testing and finally determined that it was the top of the balance pot with the lower resistance being the rest of that pot, the volume pot, and the input impedance of the next stage, all in parallel.

 

[rothwellaudio]

I think you're right - noise in the final circuit has a variety of causes and power supply regulation is probably low on the list.

 

[Powertech]

One of the things that I would be careful of is any DC offset reaching the tone and volume control circuits. There will probably be line caps at the inputs and outputs of these circuits. These are usually 10uF to 47uF electrolytics. A lot of people replace these is small film caps which cannot fully eliminate LF voltages which the pots will kind of treat as DC or close to it. Even the slightest DC offset can cause scratchiness in these pots.

 

[BinaryMike]

A lot of people replace these is small film caps which cannot fully eliminate LF voltages which the pots will kind of treat as DC or close to it. Even the slightest DC offset can cause scratchiness in these pots.

Film caps are far superior to electrolytics with respect to leakage current --- by several orders of magnitude.

 

[Powertech]

Film caps are far superior to electrolytics with respect to leakage current --- by several orders of magnitude.

I'm not talking about capacitor leakage current. Very low frequencies very close to DC will require quite a large capacitor reservoir to filter them from the following stage. That is why electrolytics were fitted in a lot of original designs using either discrete class A or earlier OPamps, both of which tended to produce DC offsets and close to DC frequencies at their output..

 

[gadget73]

Also consider cost and physical size. A mfg isn't inclined to use an expensive film cap when a cheap as chips electrolytic will get it done. 10uf+ 50v film caps are way more spendy than an electrolytic is.

 

[ConradH]

I'm not really grasping the argument. The pot has to be exposed to the audio signal, and IMO reasonable cutoffs for audio circuits need to be in the single digit region to prevent phase shifts at 20 Hz. They're probably not audible, but a 3-5 Hz lower cutoff is common in audio circuits. Now, does that mean pots will inherently be scratchy if low frequencies are present? My guess is no, assuming no continuous DC is present, though I don't have a good reason why. A low leakage electrolytic should do fine, as should a film cap. If you want to use the best dielectric, it should be polypropylene or similar, so the size might be a huge issue, or the part may not be available at all. I'd rather have a large value electrolytic than a marginal value Mylar, as I don't think Mylar is the first choice in SQ if any audio voltage gets developed across it, filter applications in particular.

 

[fred soop]

.....
reasonable cutoffs for audio circuits need to be in the single digit region to prevent phase shifts at 20 Hz.
.....
I'd rather have a large value electrolytic
.....

Any electrolytics in the signal path (common for coupling stages) should be sized 10 times larger than needed for the desired low frequency cutoff. That reduces distortion to the noise floor. But, in a complex preamp with active filters, tone controls, input and output buffers, etc., if each stage is 1 dB down at 5 Hz, then 10 stages cascaded will be 10 dB down at the same 5 Hz. Eventually, we start infringing on 20 Hz audio.