Suggestions for a LiPo powered nixie power supply design?

[Paul Andrews]

One project I have on the back burner is a very small battery-powered nixie display. I thought that a variation of a power supply design that I had been using for everything else, would work fine. It turns out that the prototype of the power supply, which I had built a few years back, only worked because of a dry solder-joint somewhere on the mosfet (yes, I'm serious). The version I built specifically for this project quickly had everything overheating. When I went back and touched up the soldering on the prototype, it showed the same behavior. The culprit, BTW is pretty much down to the tiny 1:20 transformer. I have built variations of this design with bigger transformers that work very well.

There are a lot of variations of power supply design that I could mess with - obviously I have already scoured the internet on this topic - but that is the trouble. This project will never get finished if I have to run through multiple prototypes trying to find one that is small enough and that works. So I was wondering if anyone could just say 'use this design'.

The constraints are:

1. It has to fit on a circular PCB the same diameter as the tube or less (about 17mm).
2. It has to provide around 150V-160V regulated output, or maybe just 'limited' output.
3. It only has to provide 1.5mA to 2mA.
4. It has to use a LiPo as the power source, so it should work at voltages between around 3.5V and 4.5V.
5. It has to use parts I can get from digikey (so no sourcing transformers from old cameras that I can't find for example).

Surface mount components are fine...

[gregebert]

Sounds like your transformer is saturating; common problem with small magnetics.

I used an LPR6235-123 from Coilcraft in my wristwatch. They sell to end-users in small quantities, which is nice.

Even though it only ran for 7 seconds at a time, it still overheated at 7mA load but it's a possible option for you at 2mA.

[David Forbes]

I could send you a bunch of the 31105 transformer that I use in my Nixie watch with the LT1308B. 

They are available from Wurth in large quantity, but no longer a stocked item.

--
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/28bd2601-7483-4d25-a734-1bbd08ce19ce%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[SWISSNIXIE - Jonathan F.]

One Idea whould be trying to use the LT8580, there is an appnote about using it as a VFD Tube supply. Sure, you would need a second dcdc to reach 9V, but those are super tiny nowadays. Could possibly fit on a 17mm board

[Paul Andrews]

FYI: This was the transformer I was using. The saturation current is 0.7A.

[Tomasz Kowalczyk]

Hmm, datasheet of this transformer shows a pretty funky step up circuit. Does anyone have a link to description of this circuit, or just name of it?

https://www.coilcraft.com/pdfs/lpr6235.pdf

[gregebert]

I've done a fair amount of work with flyback converters, or minor variants. The other circuit looks like some kind of self-oscillating topology, and I have not done anything with those other than build a Radio Shack kit with one back in the early 1970's.

I really wish manufacturers would provide decent SPICE models for their transformers; I'm cursed with simulating these designs for weeks or months, only to find out they behave so much differently on the bench under load. It's humbling to experience how a converter works so well with no load, only to see it fall to pieces when you put it to work.

[gregebert]

I meant to say "humiliating"........

[Paul Andrews]

Hi David, PM sent...

You received this message because you are subscribed to a topic in the Google Groups "neonixie-l" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/neonixie-l/Tu4e5th5ZCk/unsubscribe.
To unsubscribe from this group and all its topics, 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/CAPbqtvfwb6CiOuktGAk_B4QiYCT1NMbh%3D2grSh-Y0-7zWJ85fQ%40mail.gmail.com.

[Paul Andrews]

Hi Greg. I took a look at the transformer you are using, but I opted for this one from the same family instead: LPR6235-253LMRB. It is 1:10 and has a much higher Isat than the one you are using. Amazingly it fit on the same footprint and it runs at a cool 100F at room temperature. Very happy.

[Richard Arndt]

I've made several PCB design attempts of a small USB 5v to 170v 1w power supply for 1x IN-18 or Dalibor R|Z568M using the PA0367 Transformer and LT3420.  Initial design would run away into the 300+ volts unloaded with no sign of regulation when loaded, even though it was set to 170.  After some email advice from Analog Devices looking over my PCB design, paying attention to ground loops, using copper poors for current wires, and vias connecting top/bottom ground planes,  my third attempt regulated @170v to 1ma, but required addition of snubber on secondary.  It would fall out of regulation quickly after 1ma.  I got frustrated, had not ideas left, tossed it into the scrap box...  

If anyone has ideas on how to resurrect this one, I'd work on it and share the design.  Maybe some of the other transformers would be a better choice.  I'd like a small SMD daughter board design costing maybe $10 to incorporate into my OneNixie clocks.  Of course I can buy them on EBay, but that's no fun :-)  I have been using HV supplies from VFDClock's for a few years.  They are quite good!

Rich

On Thursday, May 2, 2019 at 1:59:20 PM UTC-4, Paul Andrews wrote:

[gregebert]

Welcome to the club......I've never had a switching supply work exactly as initially conceived. It always took several iterations.

I did not spend much time going thru the datasheet because this device looks like it's meant for pulsed, rather than continuous, operation. That could have a large impact on how it behaves.

First, you will need some scope pictures to get a good idea what's happening in the circuit.

[newxito]

I did some tests with the PA0367 and LT3757 using the datasheet schematic (350V/flyback), regulation was ok but the efficiency was poor, 78% at 5ma/170V with 5V input voltage

[Paul Andrews]

I could've sworn I replied to this. Anyway, my design uses an LM3478 and an AO4294 - a combination I have used on many boards with different transformers - and fits on a 17mm diameter board. However driving an IN-18 at 155V/6mA the transformer gets very hot! This particular board doesn't do any snubbing on the primary side though.

I have experimented with quite a few different transformers and the design is really quite sensitive to the transformer being used. Interestingly, I have found that a lower winding ratio generally works better I.e. I get better performance with a 1:5 transformer than I do with a 1:10 or a 1:20. I suspect, but can't prove, that this has a lot to do with ringing on the primary, which is worse with the higher ratio transformers.

I also tried one of the LT isolated switches and got nowhere fast with it. One interesting thing about these is that they require a minimum output current draw, something I don't like for a Nixie tube application where you might want to turn the tubes off, and I include dimming in this, where you are turning the tubes off and on very rapidly.

[Richard Arndt]

Good Point Greg.  The LT3420 has Charge (CHG) pin which is essentially an enable/disable.  If CHG=HIGH, device charges to set voltage.  If CHG=HIGH and voltage is reached, device enters a refresh mode to regulate.  CHG=LOW is low power mode.  CHG can also be PWM'ed by micro to control input current to charger.  It has been a year since my last version, and I may resurrect working on it cause I'd rather win!  The LT3420 seem quite novel in how it measured current/voltage, but there may be a better choice in the end.

[Richard Arndt]

Interesting point Paul regarding turns ratio.  The PA0367 is 1:12, and even though ringing on primary and secondary were small, it was causing havoc with devices methods for measuring primary voltage and secondary current.

[Bill Notfaded]

Mega props for this Paul... I've followed you trials and tribulations in hackaday and your posts here.  I've learned a lot from what you've documented.  I just want to say thanks and also that it's really appreciated.  I wouldn't understand nearly what I do now if it hadn't been for your posts.  There's a lot of knowledge out there but if it wasn't for people like you that write it down and share what you've found it'd be a smaller world.

The transformers from old cameras comment was pretty funny.

B