HV PSU design based on the TPS40210

[Jan Rychter]

Hi,

I finished my HV PSU design and decided to share it as open-source hardware, for anyone to use freely: http://jan.rychter.com/high-voltage-power-supply-for-nixie-tube-projects

The supply generates up to 220V from a 12V input. In addition to that, it also provides 2*Vout (so, up to 440V, for dekatrons), and two outputs for powering digital logic: 5V and 3.3V. The primary HV boost circuit reaches 88% efficiency when going from 12V to 185V at 55mA, with a 3% output ripple.

I designed it because I couldn't find anything that would make sense for my Nixie projects. There are plenty of tiny power supply modules available on eBay, but most of them end up being impractical: no 3.3V (for my microcontroller) and 5V (for my 74141 nixie drivers), no mounting holes, no >400V output for powering dekatrons. Some supplies make a token gesture towards practicality by sticking a 7805 on the same board, but you quickly find out that the current draw of 6x74141 is enough to require a large heat sink on a 12V-powered 7805 (one 74141 consumes 12.5mA!). This means that instead of a single-board power supply you end up routing your input power all over the place, implementing your power supply in several places.

The version I'm posting online is not perfect, but works quite well in a number of my projects. I decided I'd rather publish it as it is now rather than keep it locked forever.

The design is based on the TPS40210 from Texas Instruments, as I've grown tired of the MAX1771. I just couldn't get it to work reliably, didn't like the pricing, didn't like the availability.

Hope it helps someone, at least to understand boost converters better (I certainly learned a lot while building this!).

--J.

[John Rehwinkel]

I finished my HV PSU design and decided to share it as open-source hardware, for anyone to use freely: http://jan.rychter.com/high-voltage-power-supply-for-nixie-tube-projects

That's a nice bit of work, thank you for sharing it!  I'll just offer one tip: when supporting multiple package

variants and not wanting your schematic to look wonky, make a custom Eagle part that has a footprint

with the package variants, and use that.  A side advantage is that you don't have to try to overlap parts

to get the arrangement you want, avoiding design rule check errors.

The supply generates up to 220V from a 12V input. In addition to that, it also provides 2*Vout (so, up to 440V, for dekatrons), and two outputs for powering digital logic: 5V and 3.3V. The primary HV boost circuit reaches 88% efficiency when going from 12V to 185V at 55mA, with a 3% output ripple.

Those are some useful parameters.  That would also make a nice tube/CRT supply.  The 5V output could be

adjusted to 6.3V for heaters, and the doubler could be extended to a tripler (or more) for CRTs that want more voltage.  The regulation, efficiency, and available current would all be less, but CRTs don't need much current.

I really appreciate details like mounting holes!

The version I'm posting online is not perfect, but works quite well in a number of my projects. I decided I'd rather publish it as it is now rather than keep it locked forever.

Good thinking.  Now people can pick up where you left off and make their own improvements (and hopefully

share them).

The design is based on the TPS40210 from Texas Instruments, as I've grown tired of the MAX1771. I just couldn't get it to work reliably, didn't like the pricing, didn't like the availability.

Yeah, pretty much everything out there uses the 555, MAX1771, or MC34063.  It's nice to see another approach!

Hope it helps someone, at least to understand boost converters better (I certainly learned a lot while building this!).

You mention some of the stuff you learned, and offer a nice writeup on CCM vs DCM, which I appreciate.

I'd be intereseted if you care to share more details on loop stability or the parasitic ringing on the switching node and

how to tame it with a snubber.

Well done.

John

[Dave]

Excellent writeup.

There is a lot here that I will be digesting for awhile.

Thanks for sharing.
By any chance, did you price any options for assembly of the smd components?
What method do you use to bake your own?

[threeneurons]

Thank you for sharing.

Its a very reasonably priced chip. Only ~$1.50 in "ones-n-two-zees" !

[Jan Rychter]

> By any chance, did you price any options for assembly of the smd components?
> What method do you use to bake your own?

I just assemble the boards myself, on the cheap, so I don't have any pricing for assembly. What I do is:

* order the boards from iteadstudio,
* clean the board with isopropyl alcohol,
* apply solder paste from a syringe, using a needle,
* place components using tweezers,
* dump the whole thing into a cheap mini-oven,
* press the button on my MSP430 Launchpad.

The Launchpad board controls an SSR which PWMs power to the oven (just 4 times a second). It also has a tiny board connected to it, hosting a MAX31855 thermocouple interface. The thermocouple gets inserted into the oven, and my software does the PID control. It was just a quick hack.

This contraption reflows boards surprisingly well (see here for the curve I normally get:http://jan.rychter.com/enblog/home-made-reflow-oven-for-smd). The only issue is with solder paste application: it takes time and effort, and I usually end up applying too much. I recently ordered some stencils from OSH Stencils, we'll see how that goes.

The photos show varied results, you might notice some components are slightly shifted, and there is some flux residue. This is mostly because of the "too much paste" problem: I ended up reworking the main chips using a hot air station.

Still, overall, the process works so well that I have no fear of SMD and in fact I prefer it to pulling wires through holes.

--J.

[Jan Rychter]

On 5 maj 2014, at 19:27, John Rehwinkel <jre...@mac.com> wrote:

>> I finished my HV PSU design and decided to share it as open-source hardware, for anyone to use freely:http://jan.rychter.com/high-voltage-power-supply-for-nixie-tube-projects
>
> That's a nice bit of work, thank you for sharing it! I'll just offer one tip: when supporting multiple package
> variants and not wanting your schematic to look wonky, make a custom Eagle part that has a footprint
> with the package variants, and use that. A side advantage is that you don't have to try to overlap parts
> to get the arrangement you want, avoiding design rule check errors.

Yes, that's definitely the better approach. But I'm wary of investing too much time into Eagle, as I think it is on its way out (looking forward to switching to KiCad), so I don't want to spend too much time tweaking libraries.

>> The supply generates up to 220V from a 12V input. In addition to that, it also provides 2*Vout (so, up to 440V, for dekatrons), and two outputs for powering digital logic: 5V and 3.3V. The primary HV boost circuit reaches 88% efficiency when going from 12V to 185V at 55mA, with a 3% output ripple.
>
> Those are some useful parameters. That would also make a nice tube/CRT supply. The 5V output could be
> adjusted to 6.3V for heaters, and the doubler could be extended to a tripler (or more) for CRTs that want more voltage. The regulation, efficiency, and available current would all be less, but CRTs don't need much current.

And this is why making a design public makes sense — I would never have known!

> I really appreciate details like mounting holes!

Yes, I'm tired of devices that don't have them, and as a result are only useful as a bench toy. I don't know how people mount the other PSUs I've seen — cast them in resin? Make rails? Use clips?

> [...]

> I'd be intereseted if you care to share more details on loop stability or the parasitic ringing on the switching node and
> how to tame it with a snubber.

I'd love to, but unfortunately there isn't much to share. The loop stability calculations were done using SwitcherPro. It did require some tweaking, as it wasn't really meant to design a supply like that. I managed to get a decent margin, but I can't be certain of the actual behavior, because:

* the output cap plays a role, but isn't easily controllable, especially its ESR,
* the design is intended to be used with various inductors and caps.

And since I don't have access to a network analyzer, I can't really verify the calculations. All I can see is that I get no oscillations or unstable behavior, and that load transient behavior is nice and clean, in the circuits I've built.

As to the snubber, I just went with TI recommendations again — I don't think I can properly design a snubber with my 100MHz scope.

TI has good literature on the topic:
http://focus.ti.com/general/docs/litabsmultiplefilelist.tsp?literatureNumber=slup100
http://focus.ti.com/general/docs/litabsmultiplefilelist.tsp?literatureNumber=slva255

The test points placed on the board should be enough for someone to connect a network analyzer, so perhaps some day this will happen :-)

--J.

[Nick]

On Tuesday, 6 May 2014 23:25:11 UTC+1, Jan Rychter wrote:

... But I'm wary of investing too much time into Eagle, as I think it is on its way out (looking forward to switching to KiCad), so I don't want to spend too much time tweaking libraries.

Hi Jan - Nice project - I've used this chip for EL supplies, but not for nixies - Ti's SMPS chips are very nice & simple to use...

I was wondering why you feel Eagle is on its way out - Farnell/ELement14 have invested a huge amount of time & money in bringing it on and I see praise for it in the various forums I frequent - just wondering...

Cheers

Nick