Arduino controlled boost converter

[Arznei]

Hello together,

I am one of the many who have gotten their hands on some nixie-tubes and now I am (of course) struggling with the power supply. My idea was to power the circuit from a 9VDC** power supply and use a boost converter to generate the needed ~170VDC. Now since I would need a microcontroller anyways, I figured i might aswell use it to control the boost converter while I'am at it. You can see my current progress in the attached schematic.
The values up until now are C1=220uF, C2=10uF, R1=220R, L1=100uH (1A) with the transistors as indicated in the schematic.
I soldered this part onto a prototyping board and am controlling the "UC" pin with an Arduino Uno (5V). I am sensing the current voltage with a voltage divider (100k-1k) wired to an analog input and controlling the PWM duty cycle for the UC-pin with the AutoPID library provided by the arduino software. The PWM-frequency is set to 62.5kHz.

I do actually get the voltage up to 170V (although its does fluctuate quite some bit) but it seems like i can only get ~4mA out of this boost converter before the voltage starts dropping.

So i wondered if any of you could give me some comments on:
1) Wether you think it's a good idea to control a boost converter with an uC or wether i should go with a dedicated chip.
2) The selection of a proper inductance. I used this one*: https://www.reichelt.de/Fest-Induktivitaeten-radial/L-09HCP-100-/3/index.html?ACTION=3&LA=446&ARTICLE=138645&GROUPID=3180&artnr=L-09HCP+100%C2%B5&SEARCH=L-09HCP%2B100%25C2%25B5&trstct=pos_0
3) And on how to improve efficiency and maximum output current.

I don't have a lot of equipment on hand, though I can use an electronics lab at the university for some measurements and soldering.

Thansk a lot in advance for any help!

*) I hope the link is not regarded as advertisement. I will happily remove it if thats the case!

Edit:
**) This used to say 12VDC. I am actually using 9VDC at the moment because of the BC517/BC516 though I could change that to 12VDC if needed.

[pigrew]

I'm a bit worried about your transistor driving. I suggest running the circuit through a spice simulator. Most of your issues should be visible from simulation.

The IRF730 needs >5 V to turn on, but your scheme probably switches its gate between 1.2 and 3.8 V, so it won't turn the FET on and off well. You should use some other driver circuit that will transform the micro's 5V into 10V or 12V to drive the FET.

At what frequency do you drive the circuit? Do you have feedback, or just use 50%? (no feedback seems dangerous, you should have some method of limiting the output voltage. Using series Zener diodes may be the easiest?)

EDITs:

Which diode are you using? Schottky diodes that have breakdowns more than 200V are rare.

Adding a RC snubber network looks to be a good idea based on simulation.

[Buriedcode]

Although others have used MCU-based boost converters, I'm going to suggest you either use an IC specific for the task, or one of those 555 timer based ones for nixie tubes. 

The trouble with your schematic (as well as the 555 timer based ones) is they have no current sensing, so a short at the output effectively puts the inductor and diode across the input supply - this will happily kill the diode if there isn't some kind of current limiting.  Also, using an MCU for anything other than just providing a pulse train is that it is rather slow for a feedback loop.  It can be done - I'm not suggesting it can't work - just that if your load is changing often it may have trouble regulating.

Then there is the matter of software glitches leaving the switch on for too long, putting the switch and inductor again, across the input leading to huge current flow.

I have seen a number of threads on "nixie boost converters" about, so it may be worth seeing how others have approach it.  Again, I'm not trying to put you off here, there are a number of ways to go about this ranging from what you're doing with a MOSFET< diode and inductor and MCU, all the way to hand wound transformers and dedicated flyback IC's.  I always have this link handy as it explains boost converters, flybacks, and specifically stuff for nixie tubes, it is well worth reading the whole article!  http://www.dos4ever.com/flyback/flyback.html

so... 1) for reliability dedicated chip.  For tinkering, learning, experiementing - can't hurt to try it with your uC! (except you may blow components, but they're pennies).
2) the linked article has formulas and explanations
3) Efficiency is ultimately down to:  MOSFET on resistance - this generally increases as the peak Vds increases. Switching losses (charging and discharging the MOSFETS gate takes energy, so faster switching = more losses) and to a lesser extent diode reverse recovery and forward voltage drop - 0.7V lost out of 170V doesn't really matter. 

For these reasons for boost ratios greater than about 1:8 people use flyback transformers (which are just coupled inductors).  This reduces the peak voltage on the switch allowing for as lower on-resistance MOSFET to be used (and often one with smaller gate charge).

[ebclr]

Don't reivent the whell.....

https://www.ebay.com/itm/DC-DC-Booster-Converter-Step-Up-Power-Supply-Module-for-Nixie-Tube-Glow-Mag-W5G2/253006744307?epid=849511022&hash=item3ae8608af3:g:T-kAAOSw1WJZEZBY

[Arznei]

Thanks for all your comments!

@pigrew
I put the transistor driving circuit in place exactly to increase the gate voltage. From my measurements it does seem to pull the gate to a few hundred mV on LOW and nearly to the full 9V on HIGH. May I ask how you got to your values of 1.2V or 3.8V? I don't want to find out my circuit just so happens to work out of pure luck.
The frequency is the PWM frequency I talked about. It is currently at 62.5kHz and using the standard Arduino methods I can't really go higher than that and the next lower frequency would be around 7kHz, which i think may be a little low.
I do sense the output via an analog input and regulate the duty cycle on that input. It does oscillate but it stays decently around the target voltage.

@ebclr
I have seen these already and I might eventually end up buying one just to make things work, but currently I am trying to use this project more as a learning experience than to get it running. Appreciate the tip though!

@Buriedcode
That link looks really promising, thank you! On the inductor though: once I figured out which inductance (e.g. how much uH) i want and how much current it needs to handle, what else should I look out for? How do different materials and shapes of the core compare? Is the ferrite core in the inductance I selected okay for a switching supply?
Like you said one reason for choosing the uC was to have more control over the regulation so I could experiment some with it and learn from it. So I think for the time beeing I will try to stick with it and if it doesn't end up working I might upgrade to a dedicated chip. I think I have some 555s flying around though, so maybe I'll try that circuit just to see if the Inductor/Diode/FET "frontend" is enough to do handle the output current i need.

[pigrew]

Thanks for all your comments!

@pigrew
I put the transistor driving circuit in place exactly to increase the gate voltage. From my measurements it does seem to pull the gate to a few hundred mV on LOW and nearly to the full 9V on HIGH. May I ask how you got to your values of 1.2V or 3.8V? I don't want to find out my circuit just so happens to work out of pure luck.

I am assuming that you use 5V logic from your micro. Since they are Darlington pairs, you should get two Vbe drops over the transistors. When the base is 5 V, I assumed 0.6 V for each junction, so 5-2*0.6 = 3.8 V. Checking the simulation, the gate voltage is getting much closer to 0 and 5V. So, maybe I'm wrong...

[not1xor1]

Thanks for all your comments!

@pigrew
I put the transistor driving circuit in place exactly to increase the gate voltage. From my measurements it does seem to pull the gate to a few hundred mV on LOW and nearly to the full 9V on HIGH. May I ask how you got to your values of 1.2V or 3.8V? I don't want to find out my circuit just so happens to work out of pure luck.
The frequency is the PWM frequency I talked about. It is currently at 62.5kHz and using the standard Arduino methods I can't really go higher than that and the next lower frequency would be around 7kHz, which i think may be a little low.
I do sense the output via an analog input and regulate the duty cycle on that input. It does oscillate but it stays decently around the target voltage.

I guess something is wrong with your measurements.
There is no way an emitter follower (single BJT or darlington) can increase the gate voltage.
That configuration boosts just the current.

[Arznei]

@pigrew @not1xor1
Oh I see, maybe I didn't clarify that properly. The V+ symbol in the schematic is supposed to be the +9V rail, that is also used to drive the inductor. The micro uses 0-5V to drive the UC pin, but with the transistor stage it should then be increased nearly to V+ wich is 9V in my case, right?

[gcewing]

with the transistor stage it should then be increased nearly to V+ wich is 9V in my case, right?

No. Each base-emitter junction drops 0.6V, so with the UC output at 5V, the emitter of the upper darlington will only be at 5 - 1.2 = 3.8V.

There are a number of ways you could get a bigger output voltage swing. One way would be to connect a resistor from the bases of the darlingtons to +9V, and an NPN transistor with its emitter to ground and its collector to the darlington bases, and drive the base of that transistor from the UC through a resistor. That should give you a 0 to 9V swing at the darlington bases, and a 1.2 to 7.8V swing at the output.

[T3sl4co1l]

There's no current sense!

Tim