using a buck dc-dc convertor from an unregulated power supply to power Teensy

A

AlainD

Well-known member
Hi

I'm trying to add a smart automated close to a garage door. Open is "detected" with unregulated power 24-36V -max. 34V!- (and motor noise?) to a pair off wires, it can deliver 1A. While the motor is running there's also a classic bulb giving light (few W).

For me it would be a simple solution to power a teensy from a buck DC-DC convertor, starting from a 24-36V unregulated supply to 5V (3A max.).

The teensy has only to run when there's power supply (less than 1% of the time), so this would be an efficient solution.

Do I need to add protection to the 5V side? I was thinking a capacitor xF (?) and a Zener diode.

I would also add (parallel to the teensy) a short piece of 5V LED strip, up to about 5W.
 
I would use a buck to 7V-9V, and a robust linear regulator down to 5V, with datasheet-suggested capacitors before and after.

DC-DC converters can have issues like overshoot at very low loads or overshoot spikes at certain fault conditions. While you can design/choose a DC-DC converter without those issues, I prefer having the robustness of a linear regulator taking care of such things, plus it also gives me more leeway in the DC-DC converter selection (as I tend to use the modules; my skills aren't enough to design one from scratch).
 
Nominal Animal said:
I would use a buck to 7V-9V, and a robust linear regulator down to 5V, with datasheet-suggested capacitors before and after.

DC-DC converters can have issues like overshoot at very low loads or overshoot spikes at certain fault conditions. While you can design/choose a DC-DC converter without those issues, I prefer having the robustness of a linear regulator taking care of such things, plus it also gives me more leeway in the DC-DC converter selection (as I tend to use the modules; my skills aren't enough to design one from scratch).
Click to expand...
Yes, but the Teensy already has a linear regulator from 5V to 3.3V...
 
Note: I am not an EE so take this with a grain of salt!

In some of my earlier boards for playing around with robotics, I would use a NIMH or Lipo type battery to power
my board. With some of them I used some form of buck converter.

Some of the later ones I would use some converters from Pololu, which worked fine for me:
www.pololu.com

Pololu - Step-Down (Buck) Voltage Regulators

Switching buck converters reduce an input voltage to a lower, regulated voltage much more efficiently than linear regulators.
www.pololu.com www.pololu.com
I used others as well.

With some Teensy boards, I did run into some issues with some converters. For example, I know I had some issues with:
https://www.amazon.com/gp/product/B000MXAR12 not working with some teensy boards.
 
Hi

After consideration and some searching on the forum I think of using 2 LM2596S DC-DC convertors, bot on the same input:

AZ-Delivery LM2596S DC-DC with display
(The voltage display is handy, but not really needed, but I like the screw terminals.)

1 would go from 24-34V to 22.5V to power a 24V led strip with about 5-7W (no switch, always on when there's power) (a 24V led strip can be put on a longer wire and is more efficient.)

1 would go from 24-34V to 5V to power a teensy at reduced speed + max 2 small led's and very briefly an oktocoupler. If useful also a small resistor load.

Any remarks or suggestions
 
I used the 2596S for a few years to bring 28V down to 5V and could not figure out why I would have random failures, periodically destroying my Teensys. The 2596 is quite old technology that doesn't have a lot of the new advancements of newer regulators. I switched to the LMR51430, and a number of problems went away. For instance, the 2596 is susceptible to inrush current. So hot-plugging my power supply gave a fairly good chance of "bricking" my connected Teensy. The 51430 has a startup delay of a few milliseconds... just enough time to allow the current to stabilize. The other thing is because of the high frequency of the 51430, the external components are much smaller. The 2596 requires two large electrolytic capacitors, and the inductor is also quite large compared to the newer switching regulators. So far I am very happy with the 51430 bringing 28V down to 5V.

A huge help for me was the TI Web bench simulator. You enter your requirements such as input voltage, and it recommends a number of appropriate regulators, and even generates a simulated circuit with the recommended resistor and capacitor values.

WEBENCH-CIRCUIT-DESIGNER Design tool | TI.com

View the TI WEBENCH-CIRCUIT-DESIGNER Design tool downloads, description, features and supporting documentation and start designing.
www.ti.com

Robert
 
airpanther said:
I used the 2596S for a few years to bring 28V down to 5V and could not figure out why I would have random failures, periodically destroying my Teensys. The 2596 is quite old technology that doesn't have a lot of the new advancements of newer regulators. I switched to the LMR51430, and a number of problems went away. For instance, the 2596 is susceptible to inrush current. So hot-plugging my power supply gave a fairly good chance of "bricking" my connected Teensy. The 51430 has a startup delay of a few milliseconds... just enough time to allow the current to stabilize. The other thing is because of the high frequency of the 51430, the external components are much smaller. The 2596 requires two large electrolytic capacitors, and the inductor is also quite large compared to the newer switching regulators. So far I am very happy with the 51430 bringing 28V down to 5V.

A huge help for me was the TI Web bench simulator. You enter your requirements such as input voltage, and it recommends a number of appropriate regulators, and even generates a simulated circuit with the recommended resistor and capacitor values.

WEBENCH-CIRCUIT-DESIGNER Design tool | TI.com

View the TI WEBENCH-CIRCUIT-DESIGNER Design tool downloads, description, features and supporting documentation and start designing.
www.ti.com

Robert
Click to expand...
Thank you very much.

Unfortunately a quick search didn't find any ready made versions of it. Let alone in the EU.
 
Ah, I see what you mean. You are looking for a pre-made PCB, like the ones for the LM2596, not something you have to build a circuit for? If that's the case, I'm not sure which other ready-made voltage regulators are available. The LMR51430 would be my choice if you were building your own PCB.

Robert
 
AlainD said:
Yes, but the Teensy already has a linear regulator from 5V to 3.3V...
Click to expand...
But the TLV75733P on the Teensy has an absolute maximum input voltage of only 6V, so it is that linear regulator that may go bork if a voltage spike occurs.

Let's say you have buck regulator outputting 6V, with at least 4.7µF capacitance on the output (X7R ceramics if no any requirements), followed by a say MCP1826S-5002E/AB (5V low-drop linear regulator in TO-220-3, can deal with spikes in excess of 30V) preferably with a small heatsink although not required, with 10µF X7R or more on the output. It would have very little noise from the buck converter left –– even less if you added an inductor and second bulk capacitor on the input, making a passive CLC (or "pi") low-pass filter, and the linear regulator "eats" any voltage spikes left. I like this.

Background, and explanation:

Most isolated DC-DC converters are basically switchmode converters with an unregulated output. They typically require a 10% of the rated load minimum, or their output will be much higher. I've seen over 2× output voltage with 1mA load on ones rated 2W or more! The solution there is exactly the same: pick a converter with half a volt to volt higher output on typical loads, optionally a CLC filter (inductor in series, capacitors to ground in parallel both before and after; the inductor current capability must be at least the maximum load), and then a low drop linear regulator with good supply rejection ("low noise"), and finally suitable capacitors to smooth out any leftover ripple and noise.

I live in Finland, where class II wall wart power supplies are ubiquitous, mains sockets unpolarized –– we have no "neutral" and "live" at the sockets; both are equally "live" ––, and ground loops common (typically between properly grounded and class II wall wart powered circuits). So, I use isolated DC-DC converters and digital isolators a lot; so much so that I got mocked at EEVblog forums for it. I don't mind, though; I am only a hobbyist, and the situations I use them do require them (or using laboratory power supplies for the devices I use instead of their own supplies, which I find a bit ridiculous myself, making troubleshooting unreliable as many issues are power supply dependent).

I can tell you for example that an ISO7721 plus two 0.1µF=100nF X7R capacitors used to interface my Teensies to various SBC's UARTs (RX and TX; ISO7742 for RX+CTS+TX+RTS; ISO7741 for SPI) has made my life easier, and saved me quite a few Teensies, according to GND potential differences as measured by my multimeter. Not only do they work as isolators, but they also handle any voltage level (10mA or so supplied by the other device) between 2.25V and 5.5V or so. If the two ground potentials are the same, then these happily act as voltage level translators. For an occasional SBC that uses 1.8V logic levels, I use similar TI ISO67nm instead (n-m signals in one direction, m in the opposite direction).
 
Last edited:
Nominal Animal said:
But the TLV75733P on the Teensy has an absolute maximum input voltage of only 6V, so it is that linear regulator that may go bork if a voltage spike occurs.

Let's say you have buck regulator outputting 6V, with at least 4.7µF capacitance on the output (X7R ceramics if no any requirements), followed by a say MCP1826S-5002E/AB (5V low-drop linear regulator in TO-220-3, can deal with spikes in excess of 30V) preferably with a small heatsink although not required, with 10µF X7R or more on the output. It would have very little noise from the buck converter left –– even less if you added an inductor and second bulk capacitor on the input, making a passive CLC (or "pi") low-pass filter, and the linear regulator "eats" any voltage spikes left. I like this.
Click to expand...
Thanks, made things clear.

Now I'm considering :

A buck DC-DC convertor to 10 V.

Add a 5-10W 12V LED strip to it (yes undervolted on purpose)
and then a LDO circuit to 5V with a Teensy where I try to keep the watts low (lower Mhz and maybe snooze library). I expect it to be less than 50mA at 5V, short peaks maybe 70mA. When using the Snooze library I expect to get the average in the 1mA range.
 
MarkT said:
Why use an LDO to drop 10V to 5V? 5V drop is within any regulator's headroom.
Click to expand...
You're right a simple linear regulator will do. Maybe even just go from 20V to 5V, given the low load. (I prefer 24V LED strips to 12V ones.)
 
I agree with MarkT. If you use the larger packages like TO-220-3 with a suitable heatsink (but consider what the potential on the tab is, and whether you need an isolating "silpad" or mica sheet in between; Mouser/Digikey/etc. sell these and they're cheap), it is only a matter of getting rid of the extra heat in the linear regulator. For example, a 10V drop at 100mA = 10V × 0.1A = 1W (of waste heat).

For larger drops, the low-drop-out is not needed, so you might look at say Onsemi MC7805ACTG (0.5€ at Mouser in singles), which is stable even without capacitors (although they will give a bit better noise performance). From 24V to 5V, it wouldn't need a heatsink if the current draw did not exceed 50mA, but with a beefy heatsink like Wakefield Thermal 567-647-10ABEP for TO-220 (1.82€ at Mouser in singles) you can safely draw 200mA or so, with something like +40°C rise over ambient at the heatsink.

One of the things I like to do is to use silpads (thermally conductive electrically isolating sheets; silicone, mica, polyimide or such) and cast aluminium enclosures for my projects, mostly those made by Hammond (since I have an affordable brick-and-mortar importer nearby). That way the entire enclosure becomes the heatsink, plus aluminium being conductive (except possibly for the thin oxide layer outside), it nicely rejects any EMI too, like a Faraday cage.
 
I realize this is a complete direction change, but I have a similar system where I can monitor garage door state (open/close) and I can remotely open/close it.

My method of monitoring is an ultrasonic distance sensor mounted near the top near the motor. It senses distance to the top of the door from the motor < say 3 feet it's open, else it's closed
 
You must log in or register to reply here.
Back
Top