Alternative to MPSA42 for direct drive clocks.

[Deviantgeek]

I was looking at the Supertex drivers, mainly the HV5522. Its 220v vs
300v, would that still work? I am driving 6 IN-18 tubes. I was also
looking at the SN75468, and a few 50v drivers(TPIC6A596)which I think
is too low... I will be building several clocks, and I want good,
reliable, cheap drivers.

(this is just for the cathodes)

[David Forbes]

The voltage required depends on what sort of control you plan to use,
and how many volts you apply to the anodes. The important thing to
remember is that if one cathode is lit, then the other cathodes don't
need more than 50V to turn them off. That's why the 74141 works with its
55V Zener diode clamps on its outputs.

If you want to display one or two cathodes in every tube at all times
(no blanking, but possibly fading between numbers), then a 50V driver is
barely sufficient. Or you can use an anode driver to blank the tube. I
use the TD62083 in my Nixie watch, and it works fine.

If you want to blank the digit with the cathode, then a 100V driver
would be necessary and you'd need to use a power supply below 200V.

If you want to blank the tube and use a >200V supply, then you need a
higher voltage cathode driver or an anode blanking circuit.

I think most of the Supertex parts want 12V on their signal lines, so
they are not very easy to use since normal logic uses 5V levels.

--
David Forbes, Tucson AZ

[Jeff Thomas]

I chose the HV5530 way back in 2003, and never looked for anything
else.

They've powered Z568, Z570, IN-18 and GR414 monster nixies without a
single failure across a few thousand clocks.

Here's a schematic example, used on a range of four different models:
http://www.amug.org/~jthomas/nixisatsch.pdf

IIRC, these devices were originally recommended by Mike Harrison in
2002.


Regards, Jeff

[Ron Schuster]

Could you clarify something for me? Are you saying is that a 74141
should not be used for cathode-side blanking? What about the Russian
K155ID1, which I assume has similar specs to the 74141?

[David Forbes]

On 3/1/12 7:52 AM, Ron Schuster wrote:
> Could you clarify something for me? Are you saying is that a 74141
> should not be used for cathode-side blanking? What about the Russian
> K155ID1, which I assume has similar specs to the 74141?
>
>

The 74141 is not designed for blanking. The Russian chip is said to have
higher voltage Zeners, so it may work. You do need to use a fairly low
supply, like 180V. A regulated switcher is best for this.

I recommend trying it yourself to see how well you like the results.

[Adam Jacobs]

The 74141 or K155ID1 will both work fine for cathode side blanking,
provided that your HV supply is a normal value (180vdc). If you are
using something very high, like 250v, then you're going to start running
into the problem that the 74141/K155ID1 are not able to extinguish the
nixie. FYI, the K155ID1 has somewhat different voltage characteristics
than the 74141. I understand that the 74141 can sink something like 60v,
where the K155ID1 can sink 100v.
I think that most nixies extinguish at about 135vdc or so.

-Adam

[Tidak Ada]

Me is said that the Russian K155XXx series has an aberrant (metric) pitch of
2,5mm instead of 0.1" (2,54mm) You have to consider that in your lay-out.

Btw., I have here a condensed pdf of the K155XXx series datasheets (not as
extended as Western ones).

eric

-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.

[Deviantgeek]

I will be building 2 clocks in the near future, and putting board
space together and all, the HV5522 are better for me. They are the
same chip as the HV5530, and a little cheaper. From what I understand
from the datasheet, I can give them ~12v and 5v logic and they should
work fine. This seems to be how Jeff uses them in his clocks.
Thanks!

[Paul S]

I've had 2 74141 chips fail on my Ramos prototype within the span of a
couple of years. In each failure a digit would always stay on. I'm not
sure if this was my fault, but I tried to take every precaution with
my circuit design. I drove 4 in12's at 170v dc with 4 russian 74141
chips.

At this point, I don't want to use them anymore. I feel like there's
not point in relying on old soviet ic's when you can use modern solid-
state technology with a little extra effort. Granted, the 74141's are
so damn easy and convenient to use. The HV5530 look awesome,
especially with the built in shift registers. If I were making one-
off's i'd just use those. But you can only drive 3 tubes with one
chip, and with a 4 digit clock, the unused 2 channels are a waste,
especially looking at the high cost of the chip (if you're mass
producing clocks this is prohibitive).

On the production Nixie Ramos clocks I intend to use these:
http://www.nxp.com/documents/data_sheet/PMBTA42DS.pdf They are dual
NPN 300v transistors that are really cheap. The downside is that there
are only 2 transistors per chip, so that's 5 per digit, so they are
going to significantly increase your part count on your board. But
they seem ultra-reliable, and at the cost I think they justify the
added part count.

-Paul

[Cobra007]

>
> On the production Nixie Ramos clocks I intend to use these:http://www.nxp.com/documents/data_sheet/PMBTA42DS.pdfThey are dual

> NPN 300v transistors that are really cheap. The downside is that there
> are only 2 transistors per chip, so that's 5 per digit, so they are
> going to significantly increase your part count on your board. But
> they seem ultra-reliable, and at the cost I think they justify the
> added part count.
>
> -Paul
>

I had been looking at those PMBTA42DS dual NPN transistors for my
nixie watch but found that the layout of the 2 transistors is not very
convenient to cramp them onto a very small PCB. I prefer to have both
collectors at one side of the chip and both bases at the other side.
This is the case with the DMMT5551S. Only downside is that they are
180V models, so they are probably barely sufficient for your clock. I
know there's been a lot going around that 50V would be enough etc etc,
but if you really want the maximum performance of your clock/watch
including PWM, blanking etc etc it is still better to go for high
voltage drivers. I ended up using the BSS131 mosfets which are nearly
impossible to cramp onto a tiny board, but they still fit and it also
saved me a base resistor.

Michel

[Adam Jacobs]

Is it possible that those russian 74141's were both from the same bad batch? I've made a lot of clocks and only ever had one fail on me (it failed the same way as you describe). For about $2/chip and a failure rate (for me) of about 1 in 100.

-Adam

[threeneurons]

...This is the case with the DMMT5551S. Only downside is that they are

180V models, so they are probably barely sufficient for your clock. I
know there's been a lot going around that 50V would be enough etc etc,
but if you really want the maximum performance of your clock/watch

including PWM,.. 

Michel

180V on the cathode side is plenty good. If you want compact, then a multiplexed design, will yield the lowest part count. Also a mux'd design will allow you to do the blanking and PWM on the anode side.

You should really get some nixies in front of you. Power them and monitor them, in all kinds of circuits, to get comfortable in their behavior, before you commit to any particular circuit. Here's a link to a scope image of a nixie in a mock up mux'd circuit:

http://threeneurons.files.wordpress.com/2011/08/timing01.jpg 

 

Note, that no current conducts initially, and ionization takes some time.

We can discuss this all day long, but it seems to be "trying to teach a pig to sing". The only way for you to get a handle on it, is to get intimate with it. Play with some nixies in 150 different ways. "Characterize" their traits, and document it. Abuse them, to test their limits. Kill a few in the process. Nothing is learned unless you're willing to "smoke" a circuit.

[Cobra007]

Yes, of course 180V would have been enough for my circuit, but Paul
was after high voltage drivers (he mentioned 300V) and I wasn't sure
if 180V would be enough in his situation.

>
> We can discuss this all day long, but it seems to be "trying to teach a pig
> to sing".
>

I don't really see the need to make this remark, it is neither useful
nor funny. It seems like you think this is the first time I solder 2
wires together.

There is 1000 ways to design a circuit and there is my way to design a
circuit and I happen to choose my way no matter what other people
think of it. Maybe it ends up being a bit over complicated but that is
totally fine with me.

Michel

[threeneurons]

Even though 300V was mentioned, I believe the actual intent was driving nixies. Only,the original poster can say for sure.

I'm not questioning your competence, in electronics, in general, but with nixies specifically. I'd be a lot more diplomatic with a neophyte. Yes, you can design your circuits in a gazillion different ways, but you contradict yourself. Your initially stated intent was to get the smallest (physically) circuit possible, but now you want to add complexity. Complexity that may not add reliability.

Let's get back to my nixie timing picture. I refer to that because, that's data that I've never seen before. I can't find that data anywhere. I had to run that experiment myself, and share it with the group. Before that, people were discussing all kinds of nixie misinformation about nixie "turn-ON". The only way to find out the truth, was to run an experiment. This is what I want you to do. Don't believe us. I mean that sincerely. Build test circuits first. Then only use the circuit that suits your needs the best.  

[Cobra007]

Fair enough.

I did measure that I-V curve long before I started drafting the
circuit as I had to figure out how long it took before ionization
would start and how much dimming I could get out of these tubes. For
my concept watch I used ULN2803 drivers which work perfectly fine in
most cases, but not in low light situations. As I only drive them with
very low energy in low light, there is usually 1 tube that won't light
up for a very long time (could be a couple of seconds). For the tube
that doesn't light up I need a higher voltage, which I can't get to
with the ULN drivers as the current starts to flow through the tube
that was already on (because of the zener clamping diode).

I didn't choose for the DMMT (Vceo is 160V btw, not 180V as stated
before) dual NPN because I would have end up with more components than
the single mosfets, it was also easier to implement basic circuit
protection with the fets rather than NPN transistors (one of the
reasons is that Vebo << Vgs max).

I may not be from the nixie era but that doesn't mean I do not
understand how they work. Once you know the basic characteristics of
the tube, driving them becomes easy, I mean this is not rocket
science, right?

Michel

[David Forbes]

On 3/11/12 3:09 PM, Cobra007 wrote:
> Fair enough.
>
> I did measure that I-V curve long before I started drafting the
> circuit as I had to figure out how long it took before ionization
> would start and how much dimming I could get out of these tubes. For
> my concept watch I used ULN2803 drivers which work perfectly fine in
> most cases, but not in low light situations. As I only drive them with
> very low energy in low light, there is usually 1 tube that won't light
> up for a very long time (could be a couple of seconds). For the tube
> that doesn't light up I need a higher voltage, which I can't get to
> with the ULN drivers as the current starts to flow through the tube
> that was already on (because of the zener clamping diode).
>

> Michel
>

A fine way to achieve the PWM dimming you want with a low-voltage Nixie
driver is to have a current measurement device in the power supply's
feedback path, which is conveniently provided from the common emitter
terminal of the TD62083 (ULN2083).

You can detect when the tube begins to ionize by sensing the current
using a comparator such as an LM339 working across a small resistor to
make <1V drop from emitter to B-, and use this time to start the on-time
countdown in the PWM dimming code. You then have to reduce the OFF time
after that pulse by the measured ionization delay.

The advantage of this method is that it will produce a more uniform
brightness in your dimmed display, since you are guaranteed that the
ionization time is taken into account. Plus, you can use the small,
low-cost driver chips.

[Cobra007]

>
> A fine way to achieve the PWM dimming you want with a low-voltage Nixie
> driver is to have a current measurement device in the power supply's
> feedback path, which is conveniently provided from the common emitter
> terminal of the TD62083 (ULN2083).

I think this is what you have done in your nixie watch. It will
effectively increase the anode voltage until ionization has taken
place.

>
> You can detect when the tube begins to ionize by sensing the current
> using a comparator such as an LM339 working across a small resistor to
> make <1V drop from emitter to B-, and use this time to start the on-time
> countdown in the PWM dimming code. You then have to reduce the OFF time
> after that pulse by the measured ionization delay.

In this case I would need 2 comparators to distinguish through which
tube the current flows because the current will flow but it will flow
through the wrong tube. I can see my anode voltage rise to 170V but
the tube doesn't light up, as there if no reason the voltage rise
should stop at 170V, the only explanation is that the current starts
to flow through the wrong tube. I have measured that current can flow
through the nixie down to about 120V. Interestingly, 120V + 50V zener
is exactly the 170V that I measure at my anode voltage.

And then, after I have detected that the current flows through the
wrong tube, there is nothing I can do about it unless I add more
circuitry to be able to stop the current flowing through the wrong
tube.

>
> The advantage of this method is that it will produce a more uniform
> brightness in your dimmed display, since you are guaranteed that the
> ionization time is taken into account. Plus, you can use the small,
> low-cost driver chips.
>

If all these features are added, it will probably produce a more
uniform brightness, but I can see from the amount of PCB space that I
have left over that it is not going to fit :-).

Michel

[David Forbes]

On 3/11/12 4:18 PM, Cobra007 wrote:
>>
>> You can detect when the tube begins to ionize by sensing the current
>> using a comparator such as an LM339 working across a small resistor to
>> make<1V drop from emitter to B-, and use this time to start the on-time
>> countdown in the PWM dimming code. You then have to reduce the OFF time
>> after that pulse by the measured ionization delay.
>
> In this case I would need 2 comparators to distinguish through which
> tube the current flows because the current will flow but it will flow
> through the wrong tube. I can see my anode voltage rise to 170V but
> the tube doesn't light up, as there if no reason the voltage rise
> should stop at 170V, the only explanation is that the current starts
> to flow through the wrong tube. I have measured that current can flow
> through the nixie down to about 120V. Interestingly, 120V + 50V zener
> is exactly the 170V that I measure at my anode voltage.
>

Then don't turn on the other tube at that time! I turn on only one tube
at a time, which removes that problem.

[Adam Jacobs]

This reminds me of a famous interview question:
Q: Does complexity increase reliability or decrease reliability?
A: Complexity DECREASES reliability.

On 3/11/2012 12:27 PM, threeneurons wrote:
> ...but now you want to add complexity. Complexity that may not add
> reliability.
>
>

[Cobra007]

>
> Then don't turn on the other tube at that time! I turn on only one tube
> at a time, which removes that problem.
>

That is not completely true, it's not that I turn on both tubes at the
same time, it is because the current starts to flow through the wrong
tube because IT CAN. With a 170V anode voltage and a 50V clamping
zener, the minimum voltage on either of the tubes (on or off) is 120V.
A tube that has 120V, is switched off but was switched on just a
fraction earlier (due to multiplexing) will leak current from anode to
ground through the zener clamping diode. You can't stop that unless
you disconnect the anode from the HV power supply, through an extra
transistor or something else.

I agree that adding complexity will in most cases reduce the
reliability. In my case, it is not so much that the circuit becomes
more complex (replacing 8 transistors in 1 package by 8 discrete
transistors) it is more an increase in PCB complexity. So I am not
worried that it will decrease the reliability.

Michel

[David Forbes]

I am confused. If you are turning off one tube then turning on the other
tube, and operating at a low duty cycle (I assume you're still
discussing the dimmed case), then there will be a long delay between
turning off the first tube and turning on the second tube. So the first
tube should have time to deionize. Is the leakage strong enough to make
a glow? For how long?

My Nixie watch displays an interesting result when one tube is broken -
the other tube will light all cathodes faintly when the broken tube is
selected. That's because all the cathodes are clamped to 50V, but the
power supply makes about 200V to an open circuit. (I connect both anodes
directly to HV without resistors.) In that case, the tube is lighting
when it shouldn't. But there is a broken tube, so there's no reason to
make the display function properly in that case.

I provide display blanking in my watch, which is accomplished by feeding
an MPU output to the HV supply feedback node through a carefully-chosen
resistor, which makes the HV drop to 110V when the bit is set to 1. This
allows me to turn off the display and not excite the case described
above in normal operation. It's a very low cost solution to blanking, as
I share the MPU pin with the accelerometer enable.

The long and short of it is that you ought to be able to use a 50V
driver chip in a two-tube design with a bit of effort.

[Paul S]

I mentioned 300V because that was the rating of the transistors.
Cobra, the DMMT5551S would actually work very well for my IN12 driving
application, and I also appreciate the bases being on one side, thanks
for helping me discover this component!
-Paul

[Cobra007]

I think you can compare it very much to the result you see when one
tube is broken. If the second tube doesn't ionize, it is basically the
same as if it is not there. In your case, suppose you take out 1 tube,
you will most likely find that the anode voltage of the other tube
will not rise above 170V even though your power supply can go up to
200V. This is because the leakage currents from all the other cathodes
to ground through the zener diode, consume all the current that the HV
power supply can deliver. All voltages from anode to each cathode will
be around 120V, together with the 50V zener, you will find a 170V
anode voltage rather than 200V. Of course this depends on how much
current your HV power supply can deliver, but I assume it is not more
than 3mA otherwise the battery currents will be enormous. Suppose 3mA
over 10 cathodes, would be 300uA per cathode, enough to just light
them up.

My HV power supply is quite a bit different from your one, I vary the
supplied energy as well as the "on" time during multiplexing, they are
2 individual PWM signals. So there is not necessary a long delay
between switching left-tube and right-tube. In fact, the PWM signal
for multiplexing only varies between 40 and 50% I think.

Michel

[Cobra007]

Glad I could help Paul, it's good to hear somebody appreciates a
comment :-).

Michel

[David Forbes]

> My HV power supply is quite a bit different from your one, I vary the
> supplied energy as well as the "on" time during multiplexing, they are
> 2 individual PWM signals. So there is not necessary a long delay
> between switching left-tube and right-tube. In fact, the PWM signal
> for multiplexing only varies between 40 and 50% I think.
>

I don't think that's a good idea. Nixie tubes look much better when
driven by a low duty cycle with full current, rather than a high duty
cycle with very low current. The cathode won't light fully at lower
current, and the glow is more "fuzzy" and indistinct.

Try it and see.

[Cobra007]

>
> I don't think that's a good idea. Nixie tubes look much better when
> driven by a low duty cycle with full current, rather than a high duty
> cycle with very low current. The cathode won't light fully at lower
> current, and the glow is more "fuzzy" and indistinct.
>
> Try it and see.
>

That is correct, there is indeed a minimum current that they will
still light up entirely and when that limit is reached, the
multiplexing PWM will reduce it's duty cycle below the mentioned 40%,
thanks for reminding me.