Suggestions for road trip (portable) DC power supply?

[IDEngineer]

I travel a bit for work, usually doing hardware/software/firmware or some combination. I often end up having to work late into the evenings in the hotel room, and need a small, lightweight, quality DC power supply to power the projects. Fixed 12VDC is the minimum, but variable 0-15VDC would be a plus. One amp is plenty.

On a lark I picked up a 12V battery charger/maintainer and altered it to yield a fixed output voltage. It's about the size of two stacked decks of cards, has an integrated AC power cord, and I replaced its lugs with alligator clips. It "works", sort of, but the noise on the output voltage is astonishing. This thing could double as a spark gap transmitter. Using it reminded me that most analog voltage regulators (such as the SOT89's and DPAK's on our projects) are great at rejecting low frequency noise but are happy to pass high frequency noise right through to the downstream circuitry!

Any recommendations for a toss-it-in-the-backpack development supply? Thanks!

[IanB]

Maybe one of these powered by a laptop power brick?

(I don't know what the specs are for noise and clean output, though. Might need some extra filtering on the output if this is critical.)

https://youtu.be/Cw2AjcczHg4

[IDEngineer]

Maybe one of these powered by a laptop power brick?

That's a VERY interesting idea! Lots of those laptop supplies are ~19V, which since it's stepdown only would still allow up to ~15V out. Yes, power cleanliness is a question (particularly coming from a switcher brick) but it might be worth risking the $20 or so that they appear to cost.

If I give this a try, I'll report back. Thanks!

[CustomEngineerer]

Not sure about the quality of the power output, but several different companies make this form factor: http://tekpower.us/power-supply/tp3016m.html

Edit: Several threads discussing similar from the last couple of years
https://www.eevblog.com/forum/testgear/tekpower-tp3016m-handheld-power-suppy/
https://www.eevblog.com/forum/testgear/komerci-ps2002h-anyone-tried-or-tested-one-of-these/

Edit2: One more suggesting maybe not great
https://www.eevblog.com/forum/testgear/tekpower-tp3016m-power-supply-output-funkiness/

[IDEngineer]

I've ordered one of the little modules. They ship from China so it will be a while but I'll report back on how it works out. I have a 19V 4A switcher from some old laptop that has ~600mVPP of switching noise at zero load, we'll see how the little module deals with that. I'm thinking a quickly crafted 3D enclosure with some cooling slots, input connector, and two banana jacks and it could be a very effective road supply.

[IDEngineer]

The little DPS3003 power supply module arrived today, and onto the bench it went!

On the good side, it actually cleans up the noisy output from the Toshiba laptop power supply I am intending to use with it. In all photos, the top trace is input voltage and the bottom trace is output voltage, both AC coupled. In the first photo (3V_1A24.png), the output voltage has been set to 3.0V into an incandescent lamp load, yielding output current of 1.24A. In the second photo (12V_2A48.png) the output voltage has been increased to 12V with the current going to 2.48A. While the output noise does rise a bit, it's nowhere near linear - the 3V DC output shows 4.38Vpp of noise (!!!) while the 12V DC output shows "only" 5.43Vpp of noise.

3VNoise.png shows a closer view of the relationship between input and output noise, and 3VNoiseZoomed.png zooms in on one of those spikes. The little supply module does what it can, and it actually does cut the input noise to some degree, but the high frequency content is too much for it so a lot gets through.

Next up: How it looks with a clean(er) source of input power. If it looks better I might be able to use it for my backpack power supply, but if it's still noisy the hunt will have to resume. I really want to like this little module... it's almost perfect from a functionality standpoint, and with a quick 3D printed enclosure it would be very travel friendly. But it's got to have clean DC out....

UPDATE: Tried it with a GPE-4323 bench supply. Cleaner, but it's evident that the noise in the photos is not just laptop switcher noise, but also this little switcher module's noise - at the input and the output. I am revising my earlier opinion that the noise at the input is from the laptop switcher; I'm now convinced it's input-coupled noise from this little module. At the input, you can literally see the PWM duty cycle vary as the voltage is varied. At the output, the noise is pretty much the same as with the laptop switcher. At 12VDC out, the noise is just shy of 4Vpp at ~65KHz. Pretty nasty for a DC power supply, so the hunt continues.

[IanB]

That's not very pretty, but it's possible an LC filter on the output could clean things up a bit?

[DaJMasta]

Also have you tried loaded?  It could be that the noise really ramps up when the load is at its lowest.


May be worth the LC filter either way, as for the regulators passing HF noise, would something like a couple turns through a ferrite on the power cable be sufficient?

[IDEngineer]

Also have you tried loaded?  It could be that the noise really ramps up when the load is at its lowest.

Yes, the supply is rated for 3A output and the 12V_2A48.png screenshot is at 2.48A, ~83% of its rated current output.

[IDEngineer]

That's not very pretty, but it's possible an LC filter on the output could clean things up a bit?

I'm going to give that a try today. Though adding an LC filter will slow the supply's response time to changes in the load (if you think of the supply as an amplifier). Not that I expect incredible performance from this little module for ~$20 {grin}.

I'll report back on the LC experiments.

UPDATE: I have a small selection of beefy toroidal inductors lying around (22-33uH @ 10+ amps) so I played with adding LC, Pi, and T filters on the output. Results were mixed. Adding ANY capacitance, in any filter configuration, increased the peak-to-peak noise level. I'd blame LC resonance but it didn't matter what value was used. Inductors by themselves yielded improvement but the noise never got below ~1.2Vpp, even if the output voltage was set to 1VDC. Still pretty noisy for development work. It's too bad, because I really like this little module.

[IDEngineer]

I really want this little module to work, so I've kept at it. But I'm not making much progress.

The biggest spikes in the noise photos above are at about 100MHz. I built an LC tank circuit that resonates at that frequency, on the theory that an ideal tank circuit has infinite resistance to its resonant frequency. All that happened was the amplitude of the noise got worse.

Next, I have some Fair-Rite ferrite beads from another SMPS project, used to suppress conducted switching noise. Sounds perfect, right? Tried that and the noise got a LOT worse, roughly 4X bigger peak to peak, regardless of load.

Best results I've found so far come from putting a beefy 12-33uH toroidal inductor in series with both output leads. This suppresses the noise down to under 2Vpp. Under normal circumstances I'd laugh at a "power supply" with that much output noise, but unless I use a linear supply (with its volume and weight, both of which aren't desireable in a backpack supply) that may be the best this little module can deliver.

[IanB]

The biggest spikes in the noise photos above are at about 100MHz. I built an LC tank circuit that resonates at that frequency, on the theory that an ideal tank circuit has infinite resistance to its resonant frequency. All that happened was the amplitude of the noise got worse.

Tacoma narrows bridge? I do not think resonance is good when you want to damp down noise?

[IDEngineer]

Tacoma narrows bridge? I do not think resonance is good when you want to damp down noise?

The experiment was based on "an ideal tank circuit has infinite resistance to its resonant frequency". I wondered if putting in a series tank circuit, that had ~infinite resistance at the frequency in question, might suppress that frequency. I had to build up the tank circuit from non-ideal components but even if it was close I would have expected to see it absorb some of the energy. Instead, it got worse... just like almost everything else I've tried.

[IDEngineer]

Finally received some ferrite beads from DigiKey. These are NiZn ferrites, Kemet ESD-R-14S, whose curves show the highest impedance I could find at the target frequency (100MHz) within a reasonable size.

Best configuration I've found is a ferrite in both the + and - output leads, three turns through each bead. This yields noise of ~350mVpp at 10V / 2.5A output when powered by a Toshiba laptop brick at 19V. Still not lab quality but tolerable for a lot of on-the-road development. The ferrites are 60 cents each, so that's USD 1.20 to quiet this supply WAY down. Remember, noise was almost 3Vpp when this started! That's an 88% reduction for just over a buck.

I've designed up a 3D printable enclosure for it. Waiting for more filament to arrive (tomorrow), then I'll post photos and a BOM and make the STL file available for anyone else who wants to craft up a nice little backpack supply. I've got a technical road trip in a couple of weeks and this little guy is going with me if final testing goes well.

[rf+tech]

IDEngineer,

A parallel-resonant tank theoretically has infinite impedance at resonance. A series-resonant tank theoretically has zero impedance at resonance. When placed in parallel with the output, a series-resonant tank effectively shunts the unwanted frequency to ground. Residual resistance, both DC and AC (skin effect) limit how close to zero (or infinite) one can get in practice.

RF+ Tech

[IDEngineer]

A parallel-resonant tank theoretically has infinite impedance at resonance. A series-resonant tank theoretically has zero impedance at resonance.

Agreed. That's why I built the tank circuit with its inductor and capacitor in parallel with each other. Then, one end of that combination was connected to the output of the power supply and the other end of that combination was connected to the load. At DC, the inductor should (theoretically) have zero impedance so the DC power should (theoretically) be unaffected. But at the resonant frequency, this parallel tank circuit should (theoretically) have infinite impedance "through" it, and thus dissipate the energy of the resonant frequency in the two paralleled passives while permitting the DC to pass (theoretically) unchanged.

I did not try configuring it as a series LC and using it as a shunt to ground (paralleled across the output). It took some careful surgery to build the parallel version from the SMD parts I had lying around, not sure I'll get around to safely disassembling that and reassembling it in a series configuration. If I do, I'll report the results.

Thanks!

[rf+tech]

IDEngineer,

After a good night's sleep, I realized the need for some clarification - mainly for those less practiced.

In both series and parallel resonant tanks at resonance, the impedance realized is 2 pi F L = 1 over 2 pi F C. To obtain an near zero impedance shunt, one would use a high ratio of C to L.

Conversely, to obtain a high impedance series block (as I now understand was your approach), one would use a high ratio of L to C.

In both cases, real-world component Q imposes practical limits on how close to zero or infinite one can approach.

For the specific problem at hand, I would go for a combination of the self-resonant shunt capacitor approach, and ferrite beads for series choking.

Kemet has a handy online capacitor self-resonance calculator: http://ksim.kemet.com/Ceramic/CeramicCapSelection.aspx

Select C0603, C0G, 1.8 nF, 10%, 25 V. On the plot page, change the Stop Frequency from 100 kHz to 1 GHz. The simulation shows a nice broad ESR minimum, with SRF occurring at 120.2 MHz. Keep in mind that trace inductance and physical placement of the capacitor affects the final outcome, especially at VHF and up.

RF+ Tech

[IDEngineer]

Kemet has a handy online capacitor self-resonance calculator: http://ksim.kemet.com/Ceramic/CeramicCapSelection.aspx Keep in mind that trace inductance and physical placement of the capacitor affects the final outcome, especially at VHF and up.

That's a handy webpage, thanks! I was plotting my own results in Excel. Which, as you note, is only a hand-wave at the end result at higher frequencies. Gets you started, but there's no substitute for the real world.

[IDEngineer]

Here's the final, assembled result.

The 3D printed enclosure includes vents across the top (including over the heat sink) and on the lower rear panel for convective cooling. All connections come in the rear panel to keep the overall size small. A notch on either side mates with the locking ears on the power supply module, and little slots on either side allow a flat blade screwdriver to depress those ears for easy removal. The spacing of the rear panel connectors allow the output ferrites to stand vertically on either side, out of the way of the connectors and the vents.

I put heat shrink around each ferrite to protect the cores and their wires. Then it's just a matter of wiring things up. All wire, including in the ferrites, is 18ga high strand count silicone insulated test lead wire.

The electrical performance of the assembled unit is reasonable for a backpack supply. Final noise figures are <500mVpp at 10V / 2.5A output. At the moment I have it powering a CAN based project, supplying 13V at about 100mA, and the noise is below 100mV.

Besides the module, here's the BOM (all part numbers are DigiKey):

Q1 Red banana panel jack, #J151-ND, $0.65 each
Q1 Black banana panel jack, #J152-ND, $0.65 each
Q1 5.5mm x 2.5mm DC power jack, #SC1048-ND, $2.71 each
Q2 6.3mm NiZn ferrite cores, #399-10871-ND, $0.60 each
18ga test lead wire, your choice of colors
3D filament, your choice of material and color

Finished price under $30 plus an old laptop power brick I had lying around.

Hope this helps someone... have fun!

[bicycleguy]

Here's the final, assembled result.
...

Impressive, but no shunt capacitor?

[IDEngineer]

As I mentioned earlier in the thread, no matter what value, or how I arranged it in the output circuit, the addition of a capacitor worsened the noise.

I know, I know... it bugs me too. But I had to move on and get some real, paying work done! {grin}

An investigation for some future time. Or perhaps someone else will duplicate this and then dig in deeper...?