Boost converter inductor sizing question

160 views
Skip to first unread message

ZY

unread,
May 17, 2016, 4:18:51 PM5/17/16
to neonixie-l
Hello. I'm trying to build a boost converter for my project to boost 12V to 150-220V. I'm looking at various converter circuits online, like this one: http://desmith.net/NMdS/Electronics/NixiePSU.html, and they use inductors on the order of 100uH to 150uH, and use drivers that switch at something like 300KHz. But when I look at the wikipedia equation for discontinuous mode output voltage, it's this: 

Vo = Vi * (1+Vi * D^2 * T)/(2*L*Io)

If I plug in Vi = 12V, D = 0.92, T = 1/300000, L = 0.0001, and Io = 0.025, I get something like 93V, which is far from what I want to be at. So I'm curious about what I'm doing wrong here.

Also, from what I've read so far, it seems like it's bad to switch between continuous and discontinuous modes due to stability? Since my digits would turn on and off, my current draw can swing wildly, so I can't really guarantee that I would stay in one mode right? Do modern controller ICs just handle this without us having to worry about it?


Also just an extra tag along question. I see some nixies use anode resistors when they already have cathode resistors. What is the point of the anode resistor? Wouldn't the cathode resistor already limit the current? I'm looking at the SP-101 Panaplex datasheet and they have a 2.2k anode resistor. Wouldn't this affect the brightness depending on how many segments are turned on?

Thanks in advance.

SWISSNIXIE - Jonathan F.

unread,
May 17, 2016, 4:32:54 PM5/17/16
to neonixie-l
Hi

Modern switchinh regulators don't 'just switch' they are more advanced, to determine a suitable inductor, you need to take a look at the datasheet of the IC.

The converter from the link you showed, works perfectly, i rebuild that one a few times. If you have stability problems, you can solder a cap, whats also described on the page.

For panaplex, a lot of panaplex display are specifically designed for multiplex operation, what results in a much higher current. Be sure to check if the current rating is for continous drive or multiplex drive.

gregebert

unread,
May 17, 2016, 5:36:54 PM5/17/16
to neonixie-l
You may want to consider other topologies, such as flyback.

I spent a *LOT* of time debugging the DC-DC converter for my wristwatch project, with simulations and bench-testing. In the end, I got it working but my simulations were way-off because I did not have proper modeling of my transformer. 

The main problem I had was saturation, because I used such a small transformer due to space-constraints. Lesson-learned: Use an oversized inductor and stay well-away from Isat. You'll know it's saturating because it will get warm, and even warmer while under load. 

Another area you need to be careful about is the kick-back when the switching device shuts off. Depending upon the topology, your switch should be rated at the max output of your converter, NOT the DC input supply voltage. If you go with a flyback transformer (I used a 10:1 ratio), the kickback is ideally lowered by the turns ratio; in reality it's substantially higher due to leakage inductance and other factors.

Be mindful of the ESR of the capacitor you are using for output filtering. Simulators, such as LTspice, will display the RMS current of device (a very handy feature).

Expect to spend a lot of time on the bench with a scope. Monitoring the current thru the inductor/switch with a scope can be tricky, because noise will couple into the scope's ground lead and make you think you have horrible ringing/overshoot when you really dont. There are some good techniques online, such as removing your scope's ground lead and probe-shroud, and connecting the ground at the exposed metal ring near the probe tip.

Lastly, you will find that DC-DC converters work beautifully under almost no load. Once you start loading them down, they dont work as nicely, but they definitely do work. Welcome to the real world.....

Most importantly, when you do get it working, congratulate yourself on your accomplishment.

Nick

unread,
May 18, 2016, 10:49:13 AM5/18/16
to neoni...@googlegroups.com
The design on that page has been built many 100s of time with no problems - where people have had issues is where they've chosen to ignore the very specific guidelines on component choice and layout :)

EDIT: I recently updated the page a bit to replace deprecated parts with currently available items.

Nick

ZY

unread,
May 18, 2016, 3:35:20 PM5/18/16
to neonixie-l
Thanks for the replies. I'm currently building one using the guidelines from the website, but I'm still curious why the Vo equation in theory is way off compared to the real life version?

threeneurons

unread,
May 18, 2016, 3:49:36 PM5/18/16
to neonixie-l
Is it regulating ? If overloaded, it can never get to that voltage.

As Nick stated, if you followed guidelines faithfully, there should be no issues. That's a pretty finicky chip, if misused. 

What's the rating on the coil ? Can you give us its part number. I did not see a current rating. The coil, is the most important part, of a boost supply.

ZY

unread,
May 18, 2016, 4:15:24 PM5/18/16
to neonixie-l
Oh nono, I haven't built it yet. I was just wondering why the equation from wikipedia doesn't work. More of a theory based question than a practical one.

since I already had the footprint for it in my Eagle library.

Quixotic Nixotic

unread,
May 18, 2016, 4:20:48 PM5/18/16
to neoni...@googlegroups.com
On 18 May 2016, at 15:49, Nick wrote:

> The design on that page has been built many 100s of time with no problems - where people have had issues is where they've chosen to ignore the very specific guidelines on component choice and layout :)
>
> Nick

That'll be me ignoring your guidelines then Nick. I can't get that picky circuit to play nicely as you know and it demands the manufacturing of proper PCBs and the relatively expensive components you specify, so I simply won't bother with it any more. I appreciate the time and effort you put into your research and the intellectual elegance of it all, I just got frustrated with it, me being a numpty and all.

John S

Nick

unread,
May 19, 2016, 4:15:06 PM5/19/16
to neonixie-l
I wasn't getting at you at all John :) The MAX1771 is a subtle beast and quick to anger - if you recall I had a look at your setup and couldn't see what was causing it to misbehave... it took me about 6 goes (and discussions with Maxim) before my attempts got to the final stage.

The 1771 is a bit of a Ferrari and my design is aimed at optimum performance, not cost-effectiveness - generally, if you stick exactly to what's on the page, it'll work every time, but moving it onto another board etc. can sometimes be problematic... 

Your brilliant, left-field, designs more than make up for that !

Cheers

Nick

Quixotic Nixotic

unread,
May 19, 2016, 4:41:35 PM5/19/16
to neoni...@googlegroups.com

On 19 May 2016, at 21:15, Nick wrote:

> I wasn't getting at you at all John :) The MAX1771 is a subtle beast and quick to anger - if you recall I had a look at your setup and couldn't see what was causing it to misbehave... it took me about 6 goes (and discussions with Maxim) before my attempts got to the final stage.

Well at least I got that one to work after I got home again. Not sure how, I just unsoldered and moved a few components around and it started to regulate.

I didn't think you were getting at me Nick. I think to use that Max 1771 it needs a properly laid out PCB, not my el-cheapo home-etch creations. Now I have done 4 boards at elecrow.com and they are so cheap to do I would go down that route in future.

John S

gregebert

unread,
May 19, 2016, 5:11:06 PM5/19/16
to neonixie-l
I suggest poking around with your scope to see if any unacceptable noise exists, and also running at different supply voltages. You dont want a marginal design.

The last switcher I did failed to work from a bench supply, but ran perfectly on a battery. Turns out the leads from my bench supply were too long (inductance).

Instrument Resources of America

unread,
May 19, 2016, 11:37:51 PM5/19/16
to neoni...@googlegroups.com

I'm confused here just a bit. You said. "The last switcher I did failed to work from a bench supply, but ran perfectly on a battery. Turns out the leads from my bench supply were too long (inductance)."   I assume that when you state 'bench supply', you mean a D.C. supply.  If so then I have to question, ""inductance"",  huh???   Ira.





On 5/19/2016 2:11 PM, gregebert wrote:
I suggest poking around with your scope to see if any unacceptable noise exists, and also running at different supply voltages. You dont want a marginal design.

The last switcher I did failed to work from a bench supply, but ran perfectly on a battery. Turns out the leads from my bench supply were too long (inductance).
--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/f509851b-d3b4-43a7-be40-14e99f7831d7%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

IRACOSALES.vcf

gregebert

unread,
May 20, 2016, 1:32:55 AM5/20/16
to neonixie-l
Oh yeah, test-lead inductance. 3 feet of Pomona test leads for Vcc and GND (6 feet round trip). The board I was debugging is a nixie wristwatch, and it has a small DC-DC converter that takes the 3.7 to 4.2V supply (typically a Li-ion battery) down to 3.25V to run the onboard FPGA and display drivers. Even though it's low current, say 100mA max, the DC-DC converter runs at 2-3Mhz, and has very sharp current spikes. I couldn't figure out why the converter would start, then shut down a few hundred usec later. There was all sorts of ringing on the power supply, and I had assumed it was bad probe-grounding, because I could alter it with ground clip location.

Only after I removed the ground clip, and connected to the metal sheath on the probe did it become clear I had noise on the power supply. When went to battery power (total wire length of about 1 inch), the converter ran perfectly. Using the same short leads with the bench supply (as on the battery), it also worked. From what I could tell, the noise was so bad that the DC-C converter IC became inoperable.

6 feet of testleads is about 2uH. The inductor in the converter is 1.5uH, so the testleads definitely have a major impact on the circuit.

David Forbes

unread,
May 20, 2016, 1:47:00 AM5/20/16
to neoni...@googlegroups.com
I put a big capacitor (100 uF) on the input to my Nixie watch's power
converter, and it doesn't have that issue. A large input capacitor is
recommended in the data sheet for every DC-DC converter chip I have ever
used.


On 5/19/16 10:32 PM, gregebert wrote:
> Oh yeah, test-lead inductance. 3 feet of Pomona test leads for Vcc and
> GND (6 feet round trip). The board I was debugging is a nixie
> wristwatch, and it has a small DC-DC converter that takes the 3.7 to
> 4.2V supply (typically a Li-ion battery) down to 3.25V to run the
> onboard FPGA and display drivers. Even though it's low current, say
> 100mA max, the DC-DC converter runs at 2-3Mhz, and has very sharp
> current spikes. I couldn't figure out why the converter would start,
> then shut down a few hundred usec later. There was all sorts of ringing
> on the power supply, and I had assumed it was bad probe-grounding,
> because I could alter it with ground clip location.


--
David Forbes, Tucson AZ

gregebert

unread,
May 20, 2016, 2:12:01 AM5/20/16
to neonixie-l
Well, a battery is essentially a really big capacitor if the parasitics are minimized.
Reply all
Reply to author
Forward
0 new messages