I'm designing a +/-12V split rail supply for a eurorack (modular) sythesizer. I was adviced to put a pre-regulator block at a higher voltage a connecting to it several point-of-load regulators. Because of the transformer I have, I was required to use LDOs, and to save cost I tried to use only positive one's (the negative one's are three times as expensive, at least in through-hole version).

Here is the simplified schematic of what I've come up with:

Simplified schematic The pre-regulator provides two 13.5V floating pairs of nets, which are then reduce to 12V at the load, and both GND nets are put together at that point. I think it works fine when only one POL regulator is connected, but it starts to break down when several of them are connected in parallel. The problem I think is that each U4 will not necessarily provide exactly 12V at their output.

Is there a solution that would allow me to keep the same regulators (which I already have)?

  • \$\begingroup\$ Please crop your image to make its visibility better. \$\endgroup\$
    – Andy aka
    Feb 20 at 10:35
  • \$\begingroup\$ It is already done. This is glitch from stackexchange \$\endgroup\$ Feb 20 at 10:39
  • \$\begingroup\$ @JonasDaverio. No, the image was still with the same size, but most of it transparent. \$\endgroup\$
    – devnull
    Feb 20 at 10:41
  • 1
    \$\begingroup\$ I can also confirm that it wasn't a stack exchange glitch. I was in the process of editing the image when it was completed by devnull. The image posted was uncropped. No BS please @JonasDaverio \$\endgroup\$
    – Andy aka
    Feb 20 at 10:55
  • \$\begingroup\$ @Andyaka I have the image on my computer and it is cropped down perfectly, so there must have been a problem somewhere, I don't know. Anyway the problem is fixed now. Stop accusing me for no reason, this is ridiculous. \$\endgroup\$ Feb 20 at 10:57

2 Answers 2


This all looks over complicated.

Pro analogue mixing consoles from the 1970's to the present typically use a single high current PSU that provides +/-17V (for powering standard audio opamps like NE5532, TL07x or modern equivalents). They often used the "upside down" regulator trick as you have here to reduce parts count.

The +/-17V is then distributed to the modules and at the input of each module is a 10R resistor mounted standoff - this is basically a cheap fuse (hence the standoff).

The advantage here is that the voltage everywhere is the same. Having local regulators, you can have perhaps 0.25V difference between supplies at different places, if one op amp then clips and feeds another with a lower voltage you can get weird things going on.

It's also just simpler and has worked fine in probably tens of thousands of devices sold for four or five decades.

NOTE : your schema is maybe simplified, but don't neglect decoupling caps and protective diodes around your regulators. Consult the device datasheet and don't skimp on this.

NOTE 2 : If you want to save time and hassle, you could consider building your module into an API module frame - there are compatible ones from many manufacturers and pretty good value. Save you a lot of grunt work, as long as you have enough front panel space for your interconnects.

NOTE 3: the higher the supply voltage the better your maximum output level, and therefore headroom and/or operating level, which in turn helps get the best possible signal to noise. There is nothing magic about 12V unless you have a very good reason to use that voltage.

  • \$\begingroup\$ What do you mean "mounted standoff"? For you other point, I was just blindly following the advice (on EEVblog forum) of having a pre-regulator and then POL regulators. They told me it was a standard practice in the same application you mentionned. I didn't really questionned it as I have no experience in it. \$\endgroup\$ Feb 20 at 10:46
  • \$\begingroup\$ a standoff resistor is a through hole resistor, say 0.25W, mounted so the body sits above the board surface by about 5mm. This is so that if the module develops a fault and tries to short the rail the resistor will burn without burning the PCB. The smell and smoke also gives you a nice indication of where the problem lies, physically :-) \$\endgroup\$
    – danmcb
    Feb 20 at 10:53
  • \$\begingroup\$ all I can say is that I worked in the pro audio industry from the mid 1980s until 2000 or so, and this is the way it is done on countless desks made by Soundcraft, TAC, Soundtracs, Studer and others. The arrangement you've been recommended has no merit over the standard method that I can see. \$\endgroup\$
    – danmcb
    Feb 20 at 10:55
  • \$\begingroup\$ For your NOTE 3, the reason is simply compliance with eurorack standard. \$\endgroup\$ Feb 20 at 11:14

It would be useful to know the power levels in question, and whether this for a one-off unit, or for commercial production.

If it is for commercial use, you will almost certainly need features such as safety certification, variable voltage input (110-250V), low weight (so no transformers), high efficiency and a good power factor. Unless your volumes are very high, and margins low, the easiest approach is to use off-the-shelf SMPS. These units can either be external (with 12V supply to your kit), and internal mount, with a direct IEC input connection, and internal DC connections.

If it is for your own one-off use, then I would also go the route of an off the shelf unit. You can easily get something that will "just work" and allow you to focus on the rest of the system. You will also avoid all the wasted heat you are going to generate by using linear regulators direct from a transformer output.

If you are concerned about switching noise, there are very good SMPS out there, but you can just use the SMPS as a pre-converter to say +/-18V, then go linear for the last bit. This is the approach used in many commercial audio systems.

  • \$\begingroup\$ This is for personnal use, but is half pedagogical, hence not using off-the-shelf components. By the way, could you give me an example of "very good" SMPS? I'm curious for future projects. \$\endgroup\$ Feb 20 at 10:48
  • \$\begingroup\$ I'm reluctant to give a general "very good" SMPS as it depends on the use case. If you wanted a high current supply (say 12V 100A) I'd go with a modular server PSU, and have detailed how to adapt these in the past colintd.blogspot.com/2016/10/… For general systems, Meanwell strike a good balance between cost and performance, avoiding the worst problems of "no name" Chinese units. \$\endgroup\$
    – colintd
    Feb 20 at 11:01
  • \$\begingroup\$ using an SMPS in an audio application requires exceptionally low output switching noise. This is not easy to find in off the shelf parts, and the worst part is that you don't really know if you are "good enough" until your project is complete. For a one-off project I would stay with linear. Or SMPS followed by linear regulators if you want to save on heat and weight. (Actually this is what you see in most current designs.) \$\endgroup\$
    – danmcb
    Feb 20 at 11:04
  • \$\begingroup\$ Been there, done that, and agree there can be problems. However, there are a range of standard internal PSU form factors, and if you use on of those it is fairly easy to swap if there are issues. \$\endgroup\$
    – colintd
    Feb 20 at 11:07
  • \$\begingroup\$ @danmcb How does it work to put linear regulators after SMPS? Linear regulators are typically quite horrible at filtering high frequency noise from SMPS. \$\endgroup\$ Feb 20 at 11:12

Your Answer

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

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