I want to build a power supply that has multiple 9v outputs. Ideally, the outputs will be isolated from each other (rather than just paralleled). This is for powering guitar effect pedals. The current draw on each output will generally be pretty low, likely in the 10s of milliamps. Commercial/off-the-shelf versions of these types of power supplies (example) seem to support up to 300 mA per output, so I think I'd like to match that just for the sake of headroom.
When I originally thought about this, I thought I'd just use a transformer with multiple secondaries. Specifically, the same number of secondaries as I wanted outputs. (Or, just use multiple transformers.) So I'd have mains AC, transformer(s), and then per-secondary rectification plus regulation circuits (e.g. lm7809). That seemed like a pretty easy/straightforward approach.
But then I got to thinking, what I'd really like is to have the option of AC power or battery power; in particular, high-density lithium ion batteries like the 18650.
I did a tiny bit of web research, and came across this article, which describes creating a 5v to 10v step-up converter using the MC34063 switch-mode regulator IC. This looks pretty straightforward to me. I could use the MC34063 as a step-up converter, where my source voltage is 3.7v nominal from the 18650, and my output voltage is 9v. Looks like the MC34063 is pretty cheap (<$1), and supporting components look to be standard (i.e. cheap) fare. It also has decent efficiency (presumably better than similarly-priced LDOs); since I'm on battery, I'd like to have reasonable efficiency.
So, ignoring the AC mains input for now, I'm thinking I could build this supply with one (or more, in parallel) 18650s as the voltage source. And then duplicate the MC34063 reference step-up circuit for each 9v tap I want. I don't believe this truly isolates each output, but for my purposes maybe it's good enough?
So the question is: is this a reasonable idea? Any gotchas or issues I'm overlooking? If this looks like a non-bad idea, I'll try to whip up a schematic and PCB in the next week or two. Summary goals are:
Thanks for any comments or feedback!
When I originally thought about this, I thought I'd just use a transformer with multiple secondaries. Specifically, the same number of secondaries as I wanted outputs. (Or, just use multiple transformers.) So I'd have mains AC, transformer(s), and then per-secondary rectification plus regulation circuits (e.g. lm7809). That seemed like a pretty easy/straightforward approach.
But then I got to thinking, what I'd really like is to have the option of AC power or battery power; in particular, high-density lithium ion batteries like the 18650.
I did a tiny bit of web research, and came across this article, which describes creating a 5v to 10v step-up converter using the MC34063 switch-mode regulator IC. This looks pretty straightforward to me. I could use the MC34063 as a step-up converter, where my source voltage is 3.7v nominal from the 18650, and my output voltage is 9v. Looks like the MC34063 is pretty cheap (<$1), and supporting components look to be standard (i.e. cheap) fare. It also has decent efficiency (presumably better than similarly-priced LDOs); since I'm on battery, I'd like to have reasonable efficiency.
So, ignoring the AC mains input for now, I'm thinking I could build this supply with one (or more, in parallel) 18650s as the voltage source. And then duplicate the MC34063 reference step-up circuit for each 9v tap I want. I don't believe this truly isolates each output, but for my purposes maybe it's good enough?
So the question is: is this a reasonable idea? Any gotchas or issues I'm overlooking? If this looks like a non-bad idea, I'll try to whip up a schematic and PCB in the next week or two. Summary goals are:
- Simple: to me, this means sticking mostly to datasheet reference circuits
- Low cost: uses only readily available, common parts
- As efficient as possible without violating simple/low cost goals
- Individual 9v outputs are isolated from each other
Thanks for any comments or feedback!
Last edited:
You can design for more 9V outputs but they will not be isolated in the general understanding of no galvanic connection. The converter will be able to handle some 1A at the input, thus up to 4W if the battery voltage can follow. That should leave you max. 400mA in total at the 9V outputs.
Last edited:
As Electronic engineer specialized in industrial electronic, I strongly advice against MC34063, mainly from old Motorola or newer On brand. It is a very poor internal design, has no compensation pin (So there is no way to make it a compensation network) and is very common that this device catches fire or short circuit the internal output transistor with catastrophic consequences. As a very better substitute, I suggest L4971 or similar from ST.
Last edited:
Yes, but it is not so easy. '063's ussualy are used as buck converter. Perhaps a UCC384x (CMOS version of UC3842) plus a small MOSFET is a very proved reliable solution.
What topology are you thinking it? (Flyback, forward, SEPIC?)
Consider the LT1072 series although a bit oldie.
What topology are you thinking it? (Flyback, forward, SEPIC?)
Consider the LT1072 series although a bit oldie.
Last edited:
Here is a large range of candidates:
Selection Table for Step-Up (Boost) Regulators | Parametric Search | Analog Devices
Selection Table for Step-Up (Boost) Regulators | Parametric Search | Analog Devices
An impressive listing of possibilities.
For permanent step-up operation, forward and SEPIC are hardly worth the increased complexity.
Flyback or boost? Flyback requires more windings on the core. The advantage of traditional boost is that a single output boost converter has only one winding and can use a standard choke.
Matt, we need your input here.
For permanent step-up operation, forward and SEPIC are hardly worth the increased complexity.
Flyback or boost? Flyback requires more windings on the core. The advantage of traditional boost is that a single output boost converter has only one winding and can use a standard choke.
Matt, we need your input here.
What input do you need?
I guess I don't have enough knowledge to know what I really need! Ultimately, I was hoping for something simple like a linear regulator (e.g. lm317, lt7809), but with greater efficiency to save battery power (and ideally not have to worry about heat). But also cheap. Power range: ideally, each 9v output would support up to 300mA (but that's mostly for headroom; in practice, each output will likely supply only 50mA or less). I haven't decided how many 9v taps I want, somewhere around four to six.
Frequency: should I care? Or maybe a better question, I assume there are tradeoffs with higher/lower frequency---if that's true, what are the tradeoffs? Why might I want higher or lower frequency?
Input voltage: planning on using either 26650 or 18650 cells (possibly in parallel to increase capacity). So input voltage range would be about 3v to 5v.
Isolation: how much complexity is added to have each output isolated from the other?
Space required: not too picky here. I'm more concerned with something fairly simple, meaning little to no deviation from the datasheet reference circuit. And that reference circuit should have a fairly small number (say 12 or fewer) supporting components. And those supporting components should be of the standard/common/readily available type. All that should imply a fairly small amount of space required.
Is there good design practice that can reduce the chance of this? E.g., maybe a fast-blow fuse in front of the Vin?
Obviously I haven't looked at every switch-mode IC, but based on the few others I've looked at, it appears the MC34063 might be an outlier in terms of (low) cost.
How about any of the following? I only took a quick glance at the datasheets, just picked a few from Mouser, sorted by price:
Unfortunately they all seem to be in the low end power-wise.
The max. input current any of them can handle is 1.5A.
With an input voltage of 3V and disregarding current-ripple and operational margins, 3V*1.5A=4.5W. But, you will need say six 9V outputs of 100mA to have at least some margin. 6*9V*100mA=5.4W.
Even if you go further to the edge the margins will be insufficient.
LM2587 can handle 5A (fine) but cannot operate reliably below 4V at the input (damned).
An RT9297 (TI TPS61087) should be able to do the job with a 3A switch. One possibility.
The max. input current any of them can handle is 1.5A.
With an input voltage of 3V and disregarding current-ripple and operational margins, 3V*1.5A=4.5W. But, you will need say six 9V outputs of 100mA to have at least some margin. 6*9V*100mA=5.4W.
Even if you go further to the edge the margins will be insufficient.
LM2587 can handle 5A (fine) but cannot operate reliably below 4V at the input (damned).
An RT9297 (TI TPS61087) should be able to do the job with a 3A switch. One possibility.
But if I have one regulator per output, then then 1.5A input current ought to be plenty, right? Then the 6 is dropped from your second formula, giving 9V*100mA = 0.9W. Or then I can go for my 300mA design target, 9V*300mA = 2.7W, still well under 4.5W.
The one regulator per output idea comes from my original plan of using multiple transformer secondaries (along with separate downstream components: rectifier, smoothing caps, LDO, output cap) to achieve per-output isolation. So all the discussion above assumed a similar design, where I still have one regulator per output, but those regulators are all ultimately powered by the same battery. Obviously this isn't isolation like independent secondaries would provide, but I feel like there should be some benefit to having multiple regulators. For example, consider I have two devices powered by two outputs: with multiple regulators, one device shouldn't affect the regulation of the other. But they are clearly not isolated, as they all share a common ground.
I suppose the simplest equivalent to the multiple secondary design is to literally have completely independent circuits, i.e. one battery per output. But that feels kind of clunky to me, I'd rather have one battery for all outputs.
So I guess I have to figure this isolation thing out for real. Then I either continue with the one regulator per output idea, isolating each regulator circuit from the battery. Or, I have one big higher-power regulator for all outputs, and have some kind of isolation scheme for each output. In the latter case, I'd be willing to spend more on a bigger regulator, because I only need one.
Sorry to ramble, kind of thinking out loud, hopefully that makes some sense! Do there exist switching DC-DC converters with multiple, isolated outputs? Am I getting in over my head here?
Edit: I suppose I could do something like this: battery -> inverter -> transformer w/multiple secondaries -> regulator per output. But surely that inverter step adds a significant amount of inefficiency? Not to mention cost!
Last edited:
You are fully right, individual converter chips for each output (max. 2.7W) means you can use the "small" ICs and each output will have its own regulation and not disturb the other outputs. Further advantage - you can use standard chokes.
Yes, you can feed them from the same battery.
Sorry if I misunderstood your intention of having a single converter chip for many outputs. Disadvantage - you will need a custom-made transformer/choke.
As analyzed above, you can use one single but more powerful converter IC for plural outputs. In this case, I will suggest you a flyback design with multiple output windings. The outputs can be isolated but only one will be regulated and the other relying on the coupling to the regulated output.
Yes, you can feed them from the same battery.
Sorry if I misunderstood your intention of having a single converter chip for many outputs. Disadvantage - you will need a custom-made transformer/choke.
As analyzed above, you can use one single but more powerful converter IC for plural outputs. In this case, I will suggest you a flyback design with multiple output windings. The outputs can be isolated but only one will be regulated and the other relying on the coupling to the regulated output.
Last edited:
You do not want to hand-build a whole lot of switchers, transformers, rectifiers, and regulators.
A well-built switcher is self-regulated. 100mA for $6.
I am assuming the 3.65V max on this series is the actual working voltage of a "3.7V" battery.
https://www.mouser.com/datasheet/2/281/kdc_meu1-267600.pdf
A well-built switcher is self-regulated. 100mA for $6.
I am assuming the 3.65V max on this series is the actual working voltage of a "3.7V" battery.
https://www.mouser.com/datasheet/2/281/kdc_meu1-267600.pdf
Attachments
Although the apparent advantage of having many regulators can lead to an easy adjustment, it has a strong disadvantage. Unless you can synchronize all of them between them or to a master clock, heterodyne between them will generate lots of spurious signals conducted and radiated by wires.. Still worse, in the case you can synchronize them, minimum differences in duty cycle has different Fourier components that also can heterodyne between them. This is why this solution is rarely used in standard products.
The 18650 battery is 3.7V nominal, but when fully charged is around 4.2V, and around 3.6V or so is when it needs to be recharged.
The PDF you linked shows 3.3v or 5v input for that DC-DC converter; that's definitely out of the range of the 18650 cell. The 26650 might work, as it's 3.3v nominal. Do those converters expect a regulated input voltage? The datasheet didn't say... if so, that suggests I'd need a regulator between the battery and the converter. I did look at some of those isolated DC-DC converters on Mouser, I had the same question for all of them: it appears they expect a regulated input voltage.
But maybe that's another legitimate design possibility: battery -> 3.3v or 5v regulator -> multiple isolated DC-DC converters (one per output)?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.