Hi Everyone
I am thinking through which options I have to supply power to my next Nixie
project.
Most common seems to use a 10-15 VDC wall wart, but I don't like that the
connectors are not at all standardized.
It would be far more elegant to use a micro-USB connector, which I can also
use to update the firmware of the unit.
Of course I understand that there are some problems associated with such
solution that needs to be solved:

1, Input voltage is only 5 VDC. This would require a conversion ration of
40 which is pretty extreme. With a school book boost converter the duty
cycle would approach 98%.
I was thinking of using a good old MAX1771; do you guys think that it will
work?
I have considered to put a simple Cockroft Walton multiplier at the output
(thanks Mike for great inspiration!) but I don't really grasp how that will
affect the regulator...

2, USB only specifies at most 500 mA of current after negotiation. This is
no problem since I don't need to light the tubes while the unit is
connected to the computer.
Many phone chargers can provide 2000 mA to support certain products with
fruits on them.

3, Is it sane to connect an SMPS to the output of another SMPS? Can I
expect strange resonances?
Perhaps this is possible to solve with a simple (read: cheap and small)
input filter.

The specs I need is rougly 200 VDC at 10 mA with a mean error that at least
is approaching 0.

I would highly appreciate any comments to my idea of driving Nixie projects
from cell phone chargers.

Kind Regards
Anton

The spec sheet for John Taylor's 1363 module specifies input down to 2VDC,
so that should be fine for your purpose. Your required ouptput power is
well within spec too. Efficiency is given as 80% typ at 5-16 VDC in, so
with some care you could likely run the tubes while connected to a standard
USB port - might get away with it if the rest of your circuit is frugal and
you run the nixies at say 190V rather than 200V.

Also, they're tiny, cheap and beautifully made. So unless you feel you have
to roll your own, that looks like a no-brainer.
http://www.tayloredge.com/storefront/index.html

No connection with John, other than very happy repeat customer and admirer
of a job well done!

Jon.

Ditto, I used John Taylor's 1363 module with my last clock project and a 5v
usb cell phone charger to power it.  The charger supplies 5v at 800mA
according to the label.  It works a treat.

Brian

However, even the best supplies are about 85% efficient (some slightly
better), which will mean 470mA or thereabouts on the input side - very
close to the 500mA limit (yes , I know that this is not being driven from a
PC), Note that for USB the 500mA (USB 2) or 600mA (USB 3) number is only
for a powered hub port after negotiation - unpowered hubs have only a
single unit load (100mA) available under all versions of the USB spec.

http://en.wikipedia.org/wiki/Universal_Serial_Bus#Power

Nick.

On 8/3/12 9:10 AM, Nick wrote:

Nick,

He's not proposing to power it from a computer, but from a charger.

Those make more power than a USB hub will provide.

--
David Forbes, Tucson AZ

PC"...

It was a general comment about the risks of using USB ports...

Nick

On 8/3/12 9:15 AM, Nick wrote:

Your point being that you'll find out whether your computer makes smoke
or not if its USB port is overloaded?

I remember reading that PC USB ports are supposed to automatically shut
down on overload, but I don't know how well that attribute is tested by
China Inc.

--
David Forbes, Tucson AZ

As most boards are made in Taiwan or China, I suspect that they are mostly
Ok - they tend to use polyswitches or similar to self-reset after a bit.

However, sticking an HV SMPS which may or may not protect its input onto a
computer port may have exciting consequences (or may be perfectly fine).

;-)

Nick

Hi Everyone
I have looked more into this during the day. The performance of John's
flyback converter is indeed impressive but since space is extremely limited
this time I will need to implement the SMPS on the main (only) PCB. I think
that it will be more fun that way too =)

A flyback transformer seems to give higher conversion ratios, but I am
puzzled about it being an isolated topology by nature.
I plan to use charlie-plexed cathode drivers driven by the uC directly and
I don't want to use transformers or optocouplers to isolate so many signals.
Admittedly I haven't looked too much into flyback converters, but if it is
not possible to connect the grounds on the primary and secondary side I
don't see how this path is feasible.

Some weeks ago I stumbled over some interresting app-notes from Maxim and
Linear that described a simple modification that can be made to the simple
boost topology.

http://www.maxim-ic.com/app-notes/index.mvp/id/1109
http://www.edn.com/design/power-management/4317951/Tapped-inductor-bo...

A "tapped inductor boost converter" might be just what I need. It is very
elegant really, split the inductor in two and connect the switch to the
center tap. Sometimes such inductor is called 'autotransformer' since only
one winding acting on itself. When the energy of the magnetic field is
dumped during the discharge period there will be a substantial voltage gain
since the number of turns in the second part of the inductor is greater.

1, Does anyone have experience in this or a similar topology?
2, There does not seem to be that many inductors available seems limited at
best.
Maybe I am just using the wrong phrases when I search. Any ideas on
suitable inductors are appreciated.

Best Regards
Anton

On 12-08-03 12:23 PM, David Forbes wrote:

the easiest way for the Computer maker to do this is just to connect
directly from the 5V and many amps output of the Computer supply to the
USB port, and program the BIOS to just say yes and otherwise ignore the
request for more power from the USB device.

I am not sure how you can ensure that someone does not plug your
"device" into a computer, even just to test it if I does not come on.  A
half amp fuse in the 5V input would be a a great idea for starters.

--
Charles MacDonald                 Stittsville Ontario
cm...@zeusprune.ca              Just Beyond the Fringe
http://users.trytel.com/~cmacd/tubes.html
No Microsoft Products were used in sending this e-mail.

If you're going charlieplexed, then you'll be lighting only one cathode at
a time. Depending, on your overall design, that might mean one cathode ON,
at any time, period. What kind of nixie are you using ? You may actually,
need much less than that 10mA, you originally specified. If so, you have a
little more wiggle room, in your design.

Yeah, a transformer can give you an isolated output, but then voltage feedback is more complicated.  Also, transformers are
harder to find than plain inductors, and have more parameters to play with (for example, most flyback converters use transformers
with a gapped core).

Right.

Quite true.  You have four basic choices here.  One is to try to find a stock transformer that does what you want - as you've
seen, this can be tricky.  Another is to take a basic inductor, and add some turns on top of the winding, and use those turns
as the primary - there's a lot of trial and error involved, and some learning on magnetics*.  The third step is to wind your own
transformer from the ground up - having a coil winding machine is a help (our own David Forbes has taken this approach).
The fourth one, which I am in the process of investigating, is to use a "universal" transformer - these have six identical
windings, which can be connected in series and parallel in various combinations in order to implement a variety of
transformer functions.  One example is Digikey 732-2449-1-ND.

One approach I use is to look up the example circuits put up by the chip manufacturers and examine the part numbers
and manufacturers of the transformers used there.  While a "Pulse Engineering gobblydegook" may not be a line item
at Digikey, it's worth asking Pulse Engineering about them.

* Wurth Electronics used to offer a dandy set of books, among them "ABCs of Transformers" (Digikey 732-1415-ND)
for $18 and that explained a lot of useful concepts.  They seem to have discontinued that one, and have combined them,
along with some new material, into a single book called "Trilogy of Inductors" (Digikey 732-1414-ND), which is also
discontinued, and replaced it with "Trilogy of Magnetics" (Digikey 732-2511-ND), that covers a lot more ground, but is $72.

- John

One more thing, that I'm always repeating.... If you find yourself needing
an obscure item from an obscure manufacturer that is not sold by digikey or
mouser, don't be afraid to call up the manufacturer and ask for a sample.
They won't sell it to you directly (probably.. unless you want 50,000 of
them anyways) but will almost always send you a couple of them if you tell
them that you're prototyping a new design.

-Adam

Well yes, the tubes will be multiplexed. My plan is to make the converter
fairly general to support most medium sized tubes. In the first batch I
will most likely use IN-8's.
The nominal current is ca 2.5 mA so with a 4 tube clock I should need 10 mA
to get the same brightness. I do think that it might be (far) too bright
for use during the night, but I reckon that I can control that with a lower
duty cycle for either the anode or cathode drivers.

I haven't completed the design yet, but if it turns out to be too difficult
to get a full 10 mA at 200 V then it is probably not a big deal.

/Anton

The more I read about SMPS variants that use a tapped inductor I get more
convinced that it will work out.
Not only is the voltage gain improved by an order or so, but the voltage
stress in the switch will be very much limited as well.
This should allow me to choose a cheaper, and in many senses better MOSFET,
maybe even settle for a regulator with internal switch.

In the end it all depends on which inductors I find. A very last resort is
to wind it myself but I do see it as an option albeit I don't look forward
to it.
Autotransformers for these frequencies seems very unusual, if they exist at
all.
The "generic" flyback transformers seems to be very flexible but I am a bit
concerned about how much leakage inductance I will get sine the secondary
winding is split several times.
My best hope is to find an ordinary flyback transformer with a suitable
turns ratio, saturation current and inductance. Of course there will be
some leakage due to the forced midpoint interconnection on the PCB, but it
should be manageable.

/Anton

Hi Everyone
Here is a small update on progress.

Last couple of months have been busy at work, but I have been able to spend
some time thinking about this project.
The best source of multi winding inductors I have found so far is the
Versa-Pac series from Cooper Busman. They come with 6 identical windings
and properly connected they will give me a 1:5 autotransformer.
In theory this will reduce the duty cycle from a whopping 97.5% to the
somewhat more manageable 86.7%.

I have spent a considerable time reading about different kinds of snubbers.
Just when I thought that I had found an optimal way of designing RC- and
RCD-snubbers for the switch and the rectifier I got notice of far more
modern alternatives: active clamps.
But so far I have only found one such controller that can be driven from a
5V supply, but the carrier is not very easy to hand-solder, and in any case
the maximum duty cycle is just 74% due to current-mode implementation.
I guess that a good old voltage-mode controller will have to do. My
requirements on the quality of the regulated output is not very high anyway.

When I look at the few exemples of tapped inductor boost converter that
provide any kind of schematic I have not seen any kind of snubber at all,
except for the secondary side rectifier diode.
Is this just a simplification for sake of presentation clarity, or can it
really be omitted?

My plan, so far, is to draw snubber networks for the diode and switch just
in case, and just don't put them on the PCB if they prove redundant.
With some strategic test points I can  measure the leakage inductance of
the autotransformer and this should be a good starting point for an almost
optimal design.
It seems however that the leakage can be measured in different ways. How
would the members of the group proceed with such measurement?

This became quite some texts, but I guess that my main questions at this
point boild down to:

clamp and active rectification?

Best Regards
Anton

It seems to me that if you wired it as an autotransformer, you'd get a 1:6 unit.

Würth Elektronik offers similar configurable units in their "WE-Flex" series, also with 6 identical windings, but they seem a little more expensive.
They're available in both forward converter and flyback (gapped) versions.

Are you thinking of operating in flyback mode?

From context, it sounds like this is a feature built into switching controllers, and I hadn't heard about it either, so I got curious.
I found this data sheet, showing the active clamp circuitry picked out in a different color:

http://www.intersil.com/content/dam/Intersil/documents/fn76/fn7654.pdf

Looks like all it involves is a couple of capacitors, a FET, and a diode.  Should be something you could add to an existing design, although it
might be a technique that only applies to forward converters (I need to do more reading, apparently).

There's a nice discussion of snubbers here:

http://softsolder.com/2009/03/06/rc-snubber-resonant-design/

Apparently, they're to prevent ringing, but I'm unsure why ringing would be a problem.  It could be because the ringing introduces an AC voltage
on top of the supply voltage at the primary, and exceeds the safe voltage for the switching transistor.  Again, I'd have to do some more reading.
But if that's the case, you may have little to worry about, with a 5 volt supply and a reasonably sturdy switching transistor.

Good idea - you could also add the stuff for an active clamp too, I suppose.

The link above suggests this app note for CCFL drivers - as Ed puts it, "Pay no attention to the CCFL stuff, as snubber fundamentals are the same wherever you go."

I'm sure there is such a point (as such techniques exist and are used), but I don't know how to determine that point.  It seems to me like
an RC snubber is simpler and probably quite adequate to your use, but I'm totally guessing here.

Alas, I'm not much help here, either.

- John

Hi everyone

 Hmm you're probably right, since the inductor discharges through both the
primary and secondary winding.

*Are you thinking of operating in flyback mode? *

Yes, this is based on a modified boost converter, so energy will be
transferred to the secondary only during (1-D).
It seems that forward converters are more common than flyback converters
but I would like to avoid the extra inductor and rectifier.

*Looks like all it involves is a couple of capacitors, a FET, and a diode.

Active clamps can be used in any topology that use inductors with more than
one winding, but the controllers with built-in driver for the auxiliary FET
are meant for forward converters according to their data sheets.
This is actually something I have been wondering about:
*Is there a substantial difference in controllers meant for flyback or
forward conversion? Or can such controller be used for in any topology?
*
It is indeed possible to implement the auxiliary driver with discrete
components, but I don't think it is a good idea, since the dead-band in
between must be controlled extremely carefully.
Otherwise the dissipation in the switch will be extreme.

*but I'm unsure why ringing would be a problem.*

When the switch closes the current in the leakage inductance (which is in
series with the switch) don't have anywhere to go.
Instead the voltage will increase which just might destroy the switch
completely, or at least shorten its life time.
A more detailed analysis reveals that the leakage inductance will actually
resonate with the output capacitance of the (closed) switch.
By adding an RC snubber network in parallel the ringing can be damped in
several different ways (under-, critical- and overdamped).
This type of snubber will however consume som energy which will reduce
efficiency.

Another bad effect of the ringing is that the transfer of energy from the
magnetizing inductance to the secondary will not start until the potential
at the midpoint of the coupled inductor (transformer) has reached the
reflected output voltage.
In simpler words this means that the energy will not be transferred
efficiently until the ringing has ended.

Also the ringing will be a good source of EMI. This might not be a huge
issue in hobby electronics, but now that I know about it I can just not
forget about it =)
*
*

 This was a great reference, and that is the way I would proceed with the
design of an RC snubber network, at least when it comes to the measurements.
I do think that the text lacks a bit of the details, and I would probably
do an analysis in the laplace-domain, see how the poles move when R and C
is changed (root locus method) and choose their values based on that.
If the RC snubber don't catch the peak voltage I might need to put an RCD
snubber across the primary winding, but I think simulation in Spice will
reveal that.
As you see this is actually a quite advanced procedure; with active clamp
there is just one clamping capacitor that I need to choose. I also think
that the size of the clamp capacitor is not as dependent on stray
capacitance either.

Best Regards
Anton