Stuck on the usual dillema plus a SMPS question

[Ryan McDonald]

Ok. I've decided that my 6-tube clock will either do a direct drive
using six 74141's and three shift/latch chips *OR* a 2x3 mux. I'm
kinda leaning towards the direct drive because it won't sing or
flicker. Questions:

Will the 2x3 mux really be noticeably dimmer? I know folks say that a
1x6 really sings. Can anyone comment on whether a 2x3 sings/flickers
and how noticeable it is?

I'm concerned about tube life with direct drive since the duty cycle
is 3 times higher over a 2x3 mux. Can the tubes be dimmed by
increasing the anode resistor WITHOUT stressing the tubes (ie. is it
unhealthy to "underdrive" the tubes assuming you at least make the
strike voltage)?

I know most folks here on the list prefer the max1771 based supply
over the common 555 based one due to efficiency. I have read though
that the 1771 ps is *very* finicky to things like board placement,
lead lengths, and component flavors. The 555 seems simpler and more
reliable at the expense of efficiency. Firstly is this true? And
secondly, would the usual 555-based design be adequate to drive six
IN-14's?

-Ryan

[DennisNAS]

Ryan,

I have both mux and direct drive clocks here and I have never heard
the singing some report. I would think it's just my ears going away
but the wife would surely say something. (Maybe she's deaf too). As
far as flicker goes I have noticed that if the HV drops much below
170v you can see it.

As far as tube life goes, I have a direct drive IN14 clock that's been
running 24/7 for about five years now with no problems. Pretty short
span as I know many on hear have been running them longer than that.
Personally I wouldn't recommend increasing the anode resistor to
increase dimming. You should do that through PWM. Decreasing the anode
voltage can lead to poisening.

I like both the 555 and the 1771 and the component layout can be
tricky on either one. As far as I'm concerned the 555 is pretty tough.
I have used up a few 1771s accidently so I really triple check my work
now before powering up a new supply. Either one will easily power 6
IN14s. I use the 555 if I'm not concerned about efficiency. But it
seams that the 1771 adjusts much better than the 555.
Dennis

[threeneurons]

I that 555 supply is to power 6 IN-18s direct drive, then it ain't
gonna cut it. IN-18s need at least 4mA, minimum. X6 that's 24mA.
You're really pushing the 555 designs. Yes, the MAX1771 can be
unstable. If you're going to have a PCB etched, then just copy how
nick did his, and life should be okay. There is always the middle
option, of using the MC34063. Just make sure you use the design that
has a pull-down transistor (MK1.5), or push-pull transistor pair (MKII
two). Just having a 330 ohm resistor trying, to dump the FETs charge
(MKI one), will give poorer performance than the 555. Efficiency,
determines which is best. A 555 will get you 65% at best. That's 35%
going up in heat, or 2.33W of heat. With a MK1.5 or MKII design,
efficiency can be as good as 80% (75% typical), so that will be 1.44W
of waste heat (assuming 75%). The MAX1771 should give you at least 85%
efficiency, or 0.76W of waste heat. This is all base on 4.32W going to
your nixies (and anode resistors).

How bright those tubes need to be depend on where this thing will be
sitting. If its in the TV room, or any other room in your house, other
than the kitchen, or workshop, then the ambient light will be pretty
low. I have 1x4 mux'd clocks, and all of those are still pretty
bright, in a normal home environment. Maybe too bright. Best thing to
do is to light up a single tube and stick in the room that's intended
as the home for this clock. If you want to be thorough, pulse it on
and off at varying duty cycles, to mimic multiplexing.

To avoid flicker, make the full frame rate faster than 60Hz. Maybe a
lot fast if you have young eyes. For example: 1x6 muxing. the full
frame (all 6 tubes) must be accessed in 16.7mS or less. 16.7/6=2.78mS
max per tube. Again you can try this out with a simple pulse circuit
on one tube, and see when your eyeballs no longer can see the flicker.
For my old eyes, that's a month and a half. For most normal people,
that's somewhere over 25Hz (40mS).

Don't run the tubes under their specified minimum current. that's just
as bad as overdriving them. You'll get cathode poisoning over time,
and some numerals won't illuminate fully, maybe in a month or two. You
can adjust brightness, by varying the duty cycle. Say its a 1x6 mux'd
clock, and each tube gets 2.5mS of ON time. During this period, vary
the actual ON time from say 200uS to 2.5mS, in 100uS increments. Let
all tubes stay off for the remainder of that 2.5mS. BTW: nixies are
kinda slow, as display devices go. They take about 50uS to turn on,
after the power is applied. Its not an LED, that can respond in 10s of
nanoseconds.

To see if it will sing, again run an experiment. Pulse each of your
six tubes individually, and see at which frequency, if any, that those
tubes will sing at. Avoid those frequencies, when you pick your
multiplexing frequency.

As you can see a lot of this will be determined by experiment, which
is how it should be. If there's one thing I keep ranting about, is
don't take anything as dogma. Try it out, and see what works for you.

[Ryan McDonald]

Threeneurons,

I will be driving six IN-14's (not IN-18's). Think the 555 will do it
in direct drive?

-Ryan

[Adam Jacobs]

My $0.02 to add to Dennis' comments:

My hearing is slightly diminished.. I've never heard any "singing" whatsoever from my 1x6 muxed clocks.
My son has super-hearing, though (he can hear a CRT being turned on from across the house). I had him put his ear up to various 1x6 muxed clocks in the house (all IN-12, though) and he says he doesn't hear anything. I don't know if my results are atypical. Also, although I can see a faint flicker, if I look at the tubes with the corner of my eye, I can't see a flicker when I look at them dead on. I don't know if I have atypical vision.. I know I don't wear glasses. Your mileage may vary. ;)
I would think a 2x3 mux would be just fine, provided you feed them with 180v or so. Don't try 170v with multiplexed nixies.

Yes, they do get dimmer when you multiplex them. You can drive them harder to make up for that if you want (with corresponding tube life reduction). Again, I think a 2x3 mux run at normal current (2.5ma in the case of IN-14) should be just fine as far as brightness and you'll get serious serious tube life. With that said, I have a direct drive clock that has run 24-7 for 8 years now. It just lost its first tube (an IN-14) about a month ago, I still haven't gotten around to replacing it. Interestingly, the IN-17's which have a shorter lifespan in their datasheet, are still glowing fine. I have been running these all at their rated current (2.5ma for the IN-14, 1.5ma for the IN-17).

Never "underdrive" a nixie tube. It will fail fairly rapidly due to cathode poisoning.

I've never tried the 555 or Max1771, I always use Mike's MC34063 mk1.5 circuit. I've found that it doesn't really care how I lay it out. I'm happy with it. It would definitely run a 2x3 mux. Mike could tell you if it will also push direct-drive for 6 tubes (15ma) I assume that it would, though. This brings me to another topic, though. a 2x3 (or 1x6) mux will use less electricity and require a smaller power supply (wall-wart). I like to buy 100ma 12v supplies for my 1x6 clocks. A smaller supply is cheaper and smaller. I've got a 6 tube (IN-8) direct-drive clock that requires 1a @ 9v... yuck.
Oh, and multiplexing really is less wiring. for a 1x6 mux, you need a single 74141, which is run to all of the tubes in series. Then your little anode drivers (2 transistors, a handful of resistors) x 6, and that is about it. Only requires 10 I/O pins. Direct drive will need 24 pins and 6 74141's. (Oh, did I mention that the 74141 uses a ton of current?)

-Adam

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To post to this group, send an email to neoni...@googlegroups.com.
To unsubscribe from this group, send email to neonixie-l+...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/neonixie-l?hl=en-GB.

[threeneurons]

IN-14's. Like these:

http://www.flickr.com/photos/20801462@N00/3704107559/

Use to be in Wittlich, Germany, from 2004 until 2009. Now its in
Covina, CA, so I can see it regularly. My brother moved, along with
the clock. IN-14s are as clear as the day they where 1st installed.
Its also powered by a 555 supply, which also has to power that 6802
dekatron. +450V in addition to the +200V for the nixies. Its 4x1
mux'd. The uC is an older mega8, and it uses the cpu clock to time the
AC line to see if there's 50Hz, 60Hz, or no AC (DC) coming into the
wallcube jack. Also has a DS1307 RTC, and chimes hourly. A decaying
sinewave, which sounds like a real ding, as opposed to a beep. The
schematics and HEX code can be found on my Skydrive:

https://skydrive.live.com/?cid=f9db37b8211ce831

under Electronics Projects / Nixie Clock #3

[MrNixie (UK)]

I have built a number of mux'ed and direct drive designs. I find that,
on balance, a direct drive display appears brighter to my eyes than a
multiplexed one FOR THE SAME AVERAGE CURRENT.

On some mux designs, I HAVE had singing tubes - particularly larger
IN18 types, when you can hear resonances and harmonics of the
multiplex frequency. Higher mux frequencies are better for display
stability, but where singing IS an issue, the tubes get louder with
higher mux frequencies. Hardly enough to keep the cat awake, though.

I am not a great fan of 555 SMPSs, but it's not the 555 that's
switching the current, but the MOSFET. Choose the right one (low Rds,
low gate capacitance) and you should be able to get all the current
you could reasonably need.

ALL of my SMPS designs use UC3843's, Cheap and bomb-proof and I have
never had any stability issues with them. I copy the designs from the
data sheets, with a few extra bits thrown in from time to time.

In any small simple SMPS design, efficiency varies a lot, depending on
that output MOSFET and the inductor.

If you want to try direct drive designs with lower chip counts, look
at the serial>parallel high voltage shift registers from the likes of
Supertex. Can't solder those buggers into a PCB with your plumbers'
blow torch, tho!

Experiment, experiment, experiment!

[jb-electronics]

Hi,

> If you want to try direct drive designs with lower chip counts, look
> at the serial>parallel high voltage shift registers from the likes of
> Supertex. Can't solder those buggers into a PCB with your plumbers'
> blow torch, tho!

There are PLCC sockets with .1" spacing, they work just fine with the 44
pin types and can be soldered quite easily.

Regards,
Jens

[Adam Jacobs]

Yep, I love those PLCC sockets. Also, sparkfun makes some nice SOIC,
SSOP & SOT23 breakout boards for those of us that prefer vector board to
PCB.

-Adam

[Eric1180]

Hi I am a new member but I think I have done a pretty good job with my first clock. I have mine set up 1x4 and I run the 74141 and 4x anode drivers off a single shift register. At first I got a loud whine that changed frequency with different number, but after putting a 33uf 200v cap it stoped ALL noise :D its so simple.
so mine is set up like this ... Nixie <- current resistor <- cap going to gound <- power supply

[Adam Jacobs]

Do you get any ghosting with this arrangement?

-Adam

[Eric1180]

No, No ghosting everything works great. I have a lot of controls over my clock with the program I made. Why do you asks about ghosting? I would love to learn more.

[Adam Jacobs]

Well, I was thinking that a capacitor between the nixie's anode and ground (especially a big one like 33uf) is going to make that waveform more DC-like. A multiplexed anode is going to look like (roughly) a pulse waveform, there will be a short period of high followed by a longer period of low. If it is 4 tube mux, then the ratio of high to low would be 1:4, same as if you punched 25% duty cycle square-wave into a function generator.
I think that a capacitor in that configuration would smooth the edges of that pulse. The capacitor will begin to charge when the pulse goes high, and the capacitor will begin to discharge when the pulse goes low.
It's not something I would have thought of, but if it works, then it works! That's why everyone is always railing about how important experimentation is (including me).

I'd be interested to see a scope capture of a single tube's anode line both before & after the cap were installed.
It would be pretty easy to model mathematically, though.

-Adam

On Sat, Aug 6, 2011 at 2:52 AM, Eric1180 <ericmer...@gmail.com> wrote:

No, No ghosting everything works great. I have a lot of controls over my clock with the program I made. Why do you asks about ghosting?  I would love to learn more.

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.

To view this discussion on the web, visit https://groups.google.com/d/msg/neonixie-l/-/jxox80WeWBkJ.

[Eric1180]

oh I see what your taking about, that if you have a capacitor on the anode output the capacitor has to charging every time the anode is turned on. My clock isn't set up that way.
The power supply is constantly on and the one capacitor is almost alway full. Then 4 anode drivers all connect to the power supply.
So the capacitor Is in-between the power supply and the anode drivers.

So I guess the reason this removes all that nosie is because in that brief period when one tube is switched on the cap takes up the slack and prevents the strain on the power supply that makes the noise ... I guess capacitors are like bungy cords. And there is no ghosting because the capacitor is almost constantly full. If I could figure out how to put up pictures I would love to show you.

How do you get rid of the noise on your clocks?

[Ryan McDonald]

Well, I was going to build the 555 using the plans at
www.ledsales.com.au/kits/nixie_supply.pdf which calls for a IRF740. I
noticed that the MAX1771 plans always call for a IRF830 or a
IRF644PBF. If I build the 555 would I be better off swapping the
IRF740 for a IFR830 or IRF644PBF? Or are the 555 plans tuned to the
IRF740. I'm absolutely not against building with the 1771, but I
guess Im scared of reports of its instablity if not done *exactly*
right. I do understand the fundamentals of a switchmode supply but I
do not understand them well enough to debug and tune them with a
scope. The 555 seems more "forgiving". Does anyone know what the
output specs are of the 555 design are anyway? I've never seen them
stated.

Is it correct to say that the only way to purposely dim the tubes is
to use PWM on the tubes themselves (and not use PWM in the HVPS to
reduce V or C)? I totally get uC's and digital logic but V/C
considerations are not my strength. Am I correct in saying that a
tube will attempt to draw too much current if directly connected to
the HVPS (even if the HVPS is running at the correct V) and the anode
resistor prevent this (while also requiring a V adjustment to get the
V *and* C correct simultaneously?)


I completely understand the IN-14 datasheet when it says the current
is 2ma for the tubes in direct mode, but can someone check my head
about what the datasheet means when it says "Average current for
digits 0.7 - 1.5mA" and "Pulse current for digits 7 – 13mA". I take
this to mean literally what it says (during the "on time" of a mux,
current is 7-13 and the average current is .7 to 1.5). In mux mode,
how would I go about measuring the current using a scope (I have a
fairly decent 20mhz analog scope I use for logic analysis) and/or
multimeter. Why is it ok to drive a mux at 7 to 13 ma "pulse
current", when the tube is rated at no more than 3.5ma? I do
understand the duty cycle aspect here but it still seems that during
the "on time" of the mux, we are nailing the tube VERY hard (well
above it's direct drive specs). I'm concerned that my ignorance about
V/C is going to cause me to run my tubes too hard and cause sputter/
poisoning.

I would really appreciate some of you veterans helping about a
newb. :) BTW, I will post my BASCOM-WWVB code soon. It turned out
really nice!

-Ryan

[Adam Jacobs]

Why does it have to be 555 or MAX1771? Nobody ever gives the MC34063 any love. ;)

Calculate your anode resistor with a single tube on, direct drive. (Make the software just turn on one tube and leave it on).. Calculate that current to be 2.5ma
Now, put the same anode resistor on all the IN-14's. :) Simple.
You _can_ put more current into nixies when they are multiplexed, it makes them brighter but reduces their lifespan.

-Adam

--

You received this message because you are subscribed to the Google Groups "neonixie-l" group.

[Adam Jacobs]

Ok, I think I misunderstood. I was thinking you had the capacitor between the anode driver and the nixie.
Sounds like you have the capacitor between the HVPS and the anode driver. Sounds good.
I don't do anything to lower the sound on my nixies (aside from acrylic cases.. lol)... My nixies don't make noise (I guess because I haven't used anything bigger than an IN-8/IN-14/IN-12) and my PS makes a _little_ noise [this is mechanical noise, the coil tightening and loosening on the ferrite many times a second] but the case quietens it just fine. I do have a smoothing cap after the HVPS as well, to make the power more DC-like otherwise the output looks like a sawtooth on my boost switching supplies.

-Adam

--

You received this message because you are subscribed to the Google Groups "neonixie-l" group.

To view this discussion on the web, visit https://groups.google.com/d/msg/neonixie-l/-/YWlt8brZLtsJ.

[Eric1180]

Do you cut your own acrylic cases? My clock had a acrylic top and then another frosted decorative layer that is cut out with a design. But i am so so picky about everything being Perrrfect.  But Acrylic is a tricky material. 

[Adam Jacobs]

Nah, I let tap plastics cut mine for me. If you give them a stencil,
they will do all the cutting you like. But, to be honest, I'm not one of
those that feel like the case is 50% of the clock. I respect those (Mike
et al) that do, but I'm not one of them. If I'm feeling lazy, I have tap
cut me 6 pieces (saw-cut acrylic.. like $5 total) and acrylic adhesive
them into a clear case. it'll hold basically anything that I put in it,
from synthesizer to transceiver to nixie clock to whatever other thing.
If you want laser cut from them, you pay.. I think it is $150 for any
sized job.

-Adam

> tricky material. --

> You received this message because you are subscribed to the Google
> Groups "neonixie-l" group.
> To view this discussion on the web, visit

> https://groups.google.com/d/msg/neonixie-l/-/-Dl-OCerG9oJ.

[John Rehwinkel]

> If you want laser cut from them, you pay.. I think it is $150 for any sized job.

Yeah, that's why I plonked down $600 for a laser cutter of my own.

- John KG4L

[neutron spin]

Is that laser cutter similar to the ones that are selling on E-
bay?....Hard to believe the price. How does it work with 1/8 inch
acrylic cutting?....thanks...

Robert W8UUU

[Mich...@aol.com]

What laser cutter is $600.

I am interested in getting one for that as well - would be a nice little toy.

Kind of like the egg-bot.

 

Michail

[John Rehwinkel]

>>> If you want laser cut from them, you pay.. I think it is $150 for any sized job.
>>
>> Yeah, that's why I plonked down $600 for a laser cutter of my own.

> Is that laser cutter similar to the ones that are selling on E-
> bay?....Hard to believe the price. How does it work with 1/8 inch
> acrylic cutting?....thanks...

Yup, bought mine from eBay. Granted, these aren't the plug-and-play units advertised, but I expected that. I was basically buying a laser tube and power supply, enclosure, steppers, gantry, and optics. The controller board that came with it had electrolytics that were already leaking, but since I never intended to run it with MS-DOS via a parallel port anyway, I just
replaced it with an EiBot board and talk to it via USB.

Cuts through 1/8 acrylic just fine, but for that kind of work, you really want to add air assist (the air keeps the fumes off the optics and helps blow the melted plastic away from the cut so it won't just fuse itself back in place after the cut is made.

Pics and story on my blog:

http://bodger.dreamwidth.org/14594.html

I took this approach because I'm cheap and I like to tinker. If you want a nicer unit, with air assist and the work already done (and don't mind being stuck with MS-DOS to run it), you can buy somewhat nicer units from a US vendor for $1850 here:

http://fslaser.com/products/lasers/hobby-lasers/40wbasic

- John KG4L

[John Rehwinkel]

> What laser cutter is $600.

The list price is a little more these days, but now they throw in free shipping.

This is basically the one I bought, $760 with free shipping.

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=390337162466

It's made by Shenhui Laser, their model number K40iii. Essentially the same box is sold under a plethora of other vendor names.

A search for "CO2 engraver" on eBay will bring up many choices.

The vendor I went with is "love-happyshopping", which is apparently one of several storefronts for this outfit:

Bai Aoran
Building C62, No. 2795, CanAn Road, JiaDing District, ShangHai City
201812
China
Phone: 18964091226

> I am interested in getting one for that as well - would be a nice little toy.
> Kind of like the egg-bot.

As you'll see in my other reply, I replaced the controller in mine with an EggBot board, amusingly enough!

- John KG4L

[threeneurons]

| the 555 using the plans at:
|
| http://www.ledsales.com.au/kits/nixie_supply.pdf

| calls for a IRF740. I noticed that the MAX1771 plans
| always call for a IRF830 or a IRF644PBF.

The ABCs on power FETs:

Three parameters, that you need to look at:

1) Vdss: or how high of a voltage it can handle. The max output of
your switcher, without resorting to doublers or triplers. Higher
usually better, but you usually pay a penalty, in ON resistance.

2) On Resistance (Ron): Lower the better. This times the current
flowing thru it, determines how hot it gets. Remember, efficiency, is
how much is going to the load, and NOT generating heat. Usually goes
up with higher voltage rated FETs.

3) Gate Charge (Qg): The gate (to source) of a FET looks like a
capacitor, though not a linear one, so over all charge, as opposed to
capacitance, is more helpful. Whatever you use to drive the FET has to
move this charge around. This is often the "Big Lie" in driving FETs.
They have a high DC resistance, but if you have to switch them fast,
then moving that charge around, makes the impedance (AC resistance) go
down with switching frequency. See "Capacitive Reactance". A lower
gate charge is better.

Now look at those three FETs:

IRF830: Vdss=500V, Ron=1.40ohm, Qg=24nC
IRF740: Vdss=400V, Ron=0.55ohm, Qg=63nC
IRF644: Vdss=250V, Ron=0.28ohm, Qg=63nC

Any of the three can handle the voltage needed for your common nixie
supply (180V-200V). So they're all okay in that respect.

Ron: IRF644 rates the best, and the IRF840 is the worst.

Qg: IRF830 is the best, but here we need a qualifier. The MAX1771 and
MC34063 (MK1.5 & MKII) designs can easily move that gate charge
around, so it really isn't an issue. If you use the MK1.0 (passive
pulldown) version of the MC34063, then this parameter comes into play
big time.

In practice I'd use the IRF740 or IRF644 over the IRF830, for any
circuit. Of course, I won't use the MK1.0 version of the MC34063
designs.

[threeneurons]

| Well, I was thinking that a capacitor between the nixie's
| anode and ground (especially a big one like 33uf) is going
| to make that waveform more DC-like.

Er, NO !

The purpose of multiplexing is NOT (like in NO) to meter out the
power. Its ALL (like in everything) about TIMING (like everything at
EXACTLY the right time) !

The whole point about putting a switch on the anode is so that, and
only that, tube is ON when the data is presented to its cathode(s).
REMEMBER, ALL CATHODES, of ALL TUBES, are bussed together, in
multiplexing. All the K0's are tied together, all the K1's are tied
together. Only ONE (1xN multiplexing example) tube can be ON at any
one time. If you put a cap in the anode circuit, then it will several
tubes will be ON when the cathodes are switched to present another
tubes data. Ghosting will be a CERTAINTY.

In fact, that's why there is a resistor across the base and emitter of
the high side PNP (MPSA92). AND we tell (INSIST) on people not trying
to "Muntz" it out (wiki it). Its needed to make the transistor switch
OFF faster. To make sure you don't have ghosting:

(1) Add or keep a 4.7K to 22K resistor across the base-emitter
junction, of the anode switch (PNP, MPSA92). If its missing, the
transistor (due to charge across the BE junction; not linear) will be
enough so it will still be ON (for a short while) when the next tubes
cathode data is presented (at the cathode).

(2) Add blanking intervals between digits. Nominally, ~2X-5X the
tube's normal response time. Nixies take ~50uS to turn ON, so a
blanking interval between 100 to 250uS, is recommended. During
blanking, turn ALL anodes OFF. This is better than turning all
cathodes OFF. Why ? If you are using a 74141 (or similar) to drive the
cathodes, if all are off, the voltage at the cathodes, WILL exceed the
zener clamps, and current WILL flow. It will show on several cathodes,
and will look like an orange blur. Always keep at least ONE cathode
ON. Somewhere during that blanking interval, switch the ON cathode
from the value of the last digit, to the next one.

I've experienced all of the above first hand. All of it has been
confirmed by further scope images. As always, Don't believe me. Try it
out for yourself, both ways.

<mini-RANT>The essence of modern science, is not the body of knowledge
that it has acquired, but HOW that knowledge was acquired. By
experiment - TRY IT OUT - DO</mini-RANT>

[Eric1180]

Thank you i will have to check them out. 

[Nick]

On Aug 8, 6:06 pm, threeneurons <threeneur...@yahoo.com> wrote:
>... AND we tell (INSIST) on people not trying

> to "Muntz" it out (wiki it).

Cooo.... Madman Muntz mentioned in two threads on the same day! His
name liveth evermore...

FWIW, here's the link from my answer in the other thread:
http://en.wikipedia.org/wiki/Madman_Muntz

Nick

[Ryan McDonald]

threeneurons,

That is GREAT info. Thank you very much! Now, for the much harder
(more opinion-prone) question. Which of the three designs
(555,1771,34063) are *least* influenced by "incorrect" board
placement? I am about to start building, and my original plan was go
to with the 555 plans that many people seem to use because I have not
heard any warning about the placement being critical. I *have* seen
warning after warning about the 34063 and 1771. I plan to layout my
own ONE-SIDED through-hole board and let the auto-router make most of
the decisions about placement. With this construction technique would
the 34063 be a better plan than the 555? The most I'll be driving at
once are six IN-14's in direct drive. The 1771 plans scare the crap
outta me because of all the warnings about placement. Is the 555
design really more tolerant?

-Ryan

[John Rehwinkel]

> That is GREAT info. Thank you very much! Now, for the much harder
> (more opinion-prone) question. Which of the three designs
> (555,1771,34063) are *least* influenced by "incorrect" board
> placement?

The 555 designs are generally "brute force" designs, and tend to be less efficient but
more tolerant of suboptimal layout. By changing the operating frequency, you also can have
some control of layout critical-ness -- higher frequencies are more sensitive to layout details,
but lower frequencies require larger inductors and the like.

> I plan to layout my
> own ONE-SIDED through-hole board and let the auto-router make most of
> the decisions about placement.

This is a terrible idea with any SMPS (IMHO). The auto-router simply doesn't have any concept
of which traces are current carrying, which coupling is dangerous, and how bad a few corners can be. You really want to lay out SMPS designs by hand. We're talking a dozen or so parts here, and knowing which connections need to be low impedance is the key - keep those short and fat.

- John KG4L

[jb-electronics]

Ryan,

I do not think that autorouting a SMPS is a very good idea. A very good
German website told me to always keep in mind the "current loops"
(Stromschleifen), i.e. where the duty current (yellow coloured on the
website) and idle (cylcle) current (orange) flow. Maybe this site is
worth something when translated with Google? The images should be
self-explanatory.

http://www.lothar-miller.de/s9y/categories/40-Layout-Schaltregler

Regards,
Jens

[Ryan McDonald]

Thanks John,

I should have been more specific. I plan to let the auto-router make
most choices about component placement EXCEPT the hvps. I plan to
hand-place and hand-route the hvps. I was planning on copying the 555-
plan layout because it is the ONLY set of plans that I have seen
implemented as a non-SMT one-sided board. If I could find a
1771/34063 layout from a reputable source that was one-sided non-smt,
I'd jump all over it.

[Adam Jacobs]

I never have any problems with doing the MC34063 MK1.5 design on
vectorboard, just however I like to lay it out that day. I'll take some
pictures so you can see my general placement and understand how tolerant
this design must be for it to work the way I do it. :)

-Adam W7ATJ

[jb-electronics]

I used the MC34063 on several occasions, but I always had the problem of
finding a suitable inductor...

Do you guys know where to get them?

Jens

[Adam Jacobs]

I always use Mouser.. Do they not ship to Germany?
I think that Mike even calls out mouser part numbers at the bottom of
his schematic. That was my favorite thing about Mike's MK1.5 schematic,
actually.. he calls out real part numbers for the critical components.

-Adam

[jb-electronics]

I will give Mouser a try soon, and I did not know there were part
numbers in the circuits. So maybe I will have a working SMPS after all
;-) I will check it out.

Regards,
Jens

[threeneurons]

| finding a suitable inductor...

I currently like the Bourns SDR1806 series, for nixie supplies upto
40mA. Mouser has them, but so do most other distributors. The
Coiltronics UP4B series is also good if you need more output. They're
a tad beefier. They're both surface mount devices, but they're BIG
parts, so you can tack leads onto them, if you like. I do, for
prototypes.

[James]

I'm a fan of direct drive myself. My advice is to ditch the 74141s and
use a modern solution. Supertex makes a HV shift register, or you can
use the old 74HC595 and either MPSA42 transistors or octal transistor
arrays.

I've used the older MAX771 with success in quite a few clocks. These
days you can buy ready-made DC-DC converter modules from several
members here for about what it would cost you to build one yourself.
You might just pick up one of those modules, slap it in your design
and be done with it.

[Ryan McDonald]

James,

I have seen discussions of the Supertex parts but it seems like no one
has really settled on how to actually use one in a real design (I've
never seen a schematic that used one either). I like the idea of
combining the 74141 and the 595's into a single chip but without a
"approved by experts" design, I'm afraid if I built one, it would just
burn up in a month.

Also. I did manage to find a MK1.5 design at http://zapro.dk/public/Nixie_SMPS_V1_5/
so thanks! :) I just don't like the idea of some storebought module
sitting in a clock I designed myself. I like the fact that the mk1.5
is somewhat tolerant of component placement.

[coggs]

Well this thread has jumped around on a few topics, but because I am
trapped on the tarmac I thought i'd drop my $.02..

On SMPS designs, feel free to liberate my approach, which is to use
the PWM of a microcontroller to drive the same power mosfet + coil +
rectifier + cap arrangement as the 1771.

On HV Switching, my circuit uses the supertex HV513s which are serial
in parallel out hv drivers can act as either anode or cathode drivers.
Not the cheapest, but lowers parts count...

On Laser cut. I use ponoko.com; A typical case can be done for less
than $75. Far less in modest volume. I would buy my own laser, but i
am not sure a $700 laser would cut 3/8" acrylic at anything other than
a snail's pace. To say nothing of the coddling.

For my schematics go to www.cogwheelcircuitworks.com and go to the
downloads page. feel free to ask questions on the dedicated google
group (link is on home page) or just shoot me a private email.



cheers,
..c

[MrNixie (UK)]

Ryan, if you would like to understand what happens to a nixie tube
when presented with an unrestricted (by the anode resistor) power
supply, I suggest you pick an old unloved nixie tube, arrange a good
deal of open space, and try it! I also suggest you also have one of
those fire extinguishers designed for electric fires handy, too :).
You'll need it.

Once lit, nixie tubes are basically constant voltage devices. Think of
Zener diodes. An individual nixie tube might run at 135V with a
current of 2mA. Push that current to 4mA and the voltage across the
tube will rise a bit to say 137V. A different sample of the same type
might do this but at 139-141V, instead. So, this makes it very hard to
set a Vht to get the right current for each of six tubes. And if your
Vht is typically 180V, then the attached nixie tubes will draw
WHATEVER current they can get from the supply (tens of mA or more)
until something gives out in an explosive manner.

So basically, that's what the anode resistor is there for. In the
above examples, a value of 22K will ensure that the nixie tube ends up
running at about 2mA, regardless of tube-tube variations. Yes, you
lose a bit of power in the resistor, but that's better than losing
your precious nixie tubes...

You can dim nixie tubes to some degree by droping the Vht, but you
will run into other problems of part-lit segments. Even in my direct-
drive designs, I dim the displays by PWM'ing them - all on for 20% of
the time, then all off for 80%.

The 555 is a good general purpose timer, but it was never DESIGNED to
be an SMPS device - or more particularly, it was never designed to
switch power MOSFETS quickly and cleanly, in the way that most
specialised SMPS ICs are. And I dare say that it is as easy to design
an unstable 555 SMPS as it is easy to design a good MAX1771 one. Just
keep everything short, especially high current and feedback wires/
tracks.

To summarise the Three Neurons' advice, the ideal MOSFET would have
infinite V(ds), zero R(on) and zero Q(g). For reasons that escapes me,
the components industry doesn't seem to have come up with one of these
yet (!), and any current real device is a comprimise. Buy a couple of
each type, and experiment. I personally favour a range of mosfets in
plastic cases from Toshiba.

Also, remember inductor losses, too. A good quality inductor that can
handle the PEAK (not average!) current at the top of the ramp will be
required.

Laurence




On Aug 7, 7:44 am, Ryan McDonald <mcdonald.r...@gmail.com> wrote:
> Well, I was going to build the 555 using the plans atwww.ledsales.com.au/kits/nixie_supply.pdfwhich calls for a IRF740.  I

> noticed that the MAX1771 plans always call for a IRF830 or a

[threeneurons]

- Ryan, if you would like to understand what happens to a
- nixie tube when presented with an unrestricted (by the
- anode resistor) power supply, I suggest you pick an old
- unloved nixie tube, arrange a good deal of open space, and
- try it! I also suggest you also have one of those fire
- extinguishers designed for electric fires handy, too :).
- You'll need it.

If your going to do this, please, please, take a video of it, and post
it for all of us to see ! ... what ... you think I was going to warn
him about something on safety ? ... That takes all the fun out of it !

http://www.youtube.com/watch?v=9jDsNe_bmtE