If you are planning on creating some sort of Nixie tube display, you will undoubtedly need to find yourself a high voltage DC power supply. If you don’t want to add a transformer to your project, you can always opt to build a boost converter instead. [Andrew Moser] shows us just how easy it is to build one, discussing the theory behind simple boost converters along the way.
Boost converters are often driven by dedicated ICs, but in this case the PWM signal from an Arduino does the job just fine. [Andrew] covers the process of choosing the proper components for the circuit, discussing duty cycles and components to avoid lest your boost converter die an untimely death.
He shows us how to implement a feedback system to get a more precise output voltage, but as Lady Ada has shown us, an open loop works pretty well too.
For the beginners that want to just get things up and running, his instructions and code should be sufficient, but [Andrew] provides plenty of reference links for those looking to delve deeper into the subject.
Or you could just use a 555 timer :p
The venerable old pickit 1 from microchip contained a great example of using a micro to generate the signals for a boost controller. The same pic, a 16F745, was used to communicate via usb, program the pic that was being used and control the boost supply to provide the 12v programming voltage. They eventually wrote up the design as an app note
http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1824&appnote=en012103
Looking at the breadboard layout, switched mode power conversion circuits are not so happy operating with such long lead lengths in the high current switching loops. if ya like to reduce the inductor size by increasing the switching freq to still conservative few 100 khz, a better layout and attention to capacitor ESR will be needed. What ya will find is the ideal conversion ratio of 1/(1-D) will not hold true for duty above ~.9 This is most likely why the circuit is “most efficient” at 30khz too.
We are offering a super small form factor nixie power supoly that is perfect for the breadboard. It comes in a few flavors, with some simple resistor changes, you can have 100v to 150v 180v fixed or 150 to 200v. This was designed to be a removable module for a nixie clock project I am working on. It is a closed loop feedback power supply that has a very low ripple. We currently only have a few of these supplies on hand. Let us know if there is any interest in these supplies over at greskoindustries.com. I can set up a video of it in operation over at you tube as well.
Have to put a plug in here for Linear Tech’s freely downloadable LTSpiceIV simulator.
If you’re going to be playing with switching converters, it’s helpful to simulate them first (so as not to let the smoke out). You can load any PSPICE model you want into it, though it comes with all the LT devices and sample circuits preloaded.
And the simulator also works for pretty much any analog circuit you want to simulate…not just switchers: amps, 555 timers, whatever.
Freely downloadable off Linear’s website and designed to run on Linux under WINE.
You’ve got to be kidding. An arduino as a controller for a boost converted? Just use a dedicated IC, they are cheap and reliable. I honestly do not see the hack value of this project. As a follower of H-A-D for more than 2 years, I am sad to find that lately Hackaday has become almost purely arduino-a-day…
Note: disagreement with my opinion is no grounds for removal of my post… Only disrespectful comments, eh?
Using an Arduino *just* to make a single boost supply is a bit wasteful. But if you’re driving nixie tubes, you’ll probably need an MCU of some sort anyway, in order to turn the segments on/off in a meaningful way; and having it perform *both* duties is a good thing.
This is meant as a tutorial for beginners, who more likely than not, have an Arduino on hand. Whether it is the most efficient use of a microcontroller is another matter. As Chris points out, driving Nixies is typically just one part of a project, and the Arduino can do other things while also driving the boost supply.
That said, if you feel that there are too many Arduino-related posts, please feel free to submit your own work, using other microcontrollers. We publish things that are sent to us as well as items we find online. Like it or not, the simple fact is that tons of people are using Arduinos to build things. We work with what we are given, so show us something different if you would like to see non Arduino-related projects.
It clearly says this in the article, and it’s even underlined:
“This guide is intended for educational purposes.”
It’s not about some finished and polished, repeatable project. So, in a way, it is sort of disrespectful to go on a rant about something without even considering what the article is saying.
What benefit is there to using a MOSFET driver? I see his warning about the FET’s gate capacitance…but it seems like you could accomplish the same isolation with just a 2nd (smaller) FET and a pull-up resistor. Am I missing something?
The statement about the MOSFET’s gate capacitance “resonating” with the Arduino is an oversimplification.
The faster you switch a MOSFET, the greater the gate capacitance becomes when switching (Miller effect); so the harder it becomes to switch.
That starts becoming significant when you switch in the tens of Khz. So you must either:
1) Make the pull-up resistor small. But then it will heat up when the smaller FET is on (since it’s effectively shorted).
2) Or use a more reasonably-valued resistor, but then it won’t be able to turn off the MOSFET very quickly. The MOSFET produces extra heat since it spends more time in a “half-on” resistive state.
Making a discrete MOSFET gate driver is easy. Making a *good* one takes some practice. There are many pitfalls, and ways to get weird results or inadvertently damage your MCU.
As Andrew said, “My guide was written to get the nooblet up and on his feet.” Using a driver IC works better for that goal.
If you really want to learn about discrete gate drivers, that’s better discussed separately. Check out “Design And Application Guide For High Speed MOSFET Gate Drive Circuits” by Laszlo Balogh; even if you don’t understand all the math, the sample circuits and explanations are still educational. And although specific to tesla coils, “Why MOSFETs fail in Solid State TC duty” by Richie Burnett is an excellent summary of every way a MOSFET can perform in unexpected ways. Both should come right up with a Google search.
Try simulating drivers in SPICE. You can perform a lot of experiments in a hurry, without real smoke or a real oscilloscope. I went as far as designing a triple 100Khz switching regulator with resettable soft-start. It only uses a single LM339 IC, and a reasonable amount of discrete components. It was a great learning experience! If I ever get around to building it, and verify it works as well in the real world as it does in SPICE, I’ll be sure to release it and send the link to HAD.
Thanks for the explanation!
I’ve thought about simulating circuits in SPICE before, but I never managed to figure it out…maybe the SPICE software I found was crappy. I see a plug for LTSpiceIV…is there a particular SPICE software that you recommend? Any good tutorials on how to get started with SPICE?
@matt: I’ve heard good things about LTSpice too, but haven’t used it yet. It’s probably worth checking out.
The only SPICE I have experience with is NI Circuit Design Suite (formerly Electronics Workbench), courtesy of my employer. It’s too pricey for personal use, but that won’t stop a sufficiently motivated person from acquiring it anyway if so inclined.
A good SPICE program needs few tutorials. Everything, even the virtual test equipment, is just like real world counterparts.
The one exception is convergence errors. Real world components and wires all have a bit of parasitic inductance, capacitance, and resistance (LCR) which isn’t properly accounted for in SPICE. Without that to damp signals, some simulated circuits can feed back and oscillate in ways that would never occur in the real world. This usually results in a convergence error; the circuit is inherently unstable and cannot be solved. To fix this, you might have to insert a few milliohms of resistance here, a few picofarads of capacitance there, until sufficient to damp the unrealistic oscillation. It’s more of an art than a science. There are tips and guides all over the ‘net, plus probably a section in the manual for any SPICE you choose. NI’s software can fix most convergence errors automatically by experimentally tweaking simulation parameters until the problem goes away, which is a big convenience.
Note that the reverse holds true as well – a circuit that works perfectly in SPICE may not work properly in the real world. Especially high frequency/power circuits on solderless breadboards; where the parasitic LCR can be quite large.
Still, SPICE is a very useful tool.
reason for the fet driver is for any sort of R pull up circuit to be fast (ie fast edges) with a significant capacitive load, the R needs to be small and thus will have high power dissipation. with a small “logic level” fet, ya might get by with directly driving from the uPC output. though I’d add a small series R and diode clams to ground and Vss to protect the uPC.
a giant step for the arduino community: pwm with analog feedback!
blinking leds with pwm becomes interesting if you don’t just set a pwm-value but use a sense resistor to actually control the current.
if you take this setup -or a similar one with a buck converter in the middle- and replace the voltage divider with a sense resistor in the leds path, it can be used for driving power leds.
Cool, I have a 9V to 56V bread-boarded converter driving 18 white LEDs in series with an op-amp relaxation oscillator (probably similar to [Chris]’s LM339 U1A) to switch it sitting right here on my desk. Just a little exercise in circuit design. It will boost up over 100V with no load, so as always, exercise caution. :-)
I am making a DC booster for myself, too.
I use ne555 directly drive a MOSFET
( didn’t use a MOSFET driver)
According to others’ result, I should get 500V but finally I got 340V(max).
then I change duty cycle and frequency(1k~8kHz), I still get 340~345V.
so my next test will add a mosfet driver,
I suspect ne555 can really output 200mA for short rise time( like 200~400 ns).
um….I just want to ask a question.
Many books taught us change duty cycle and then the output voltage change.
Is you result can see the change(about duty cycle)?