NE-2 lamps not fully blanking

110 views
Skip to first unread message

Ron Schuster

unread,
Sep 7, 2012, 10:39:02 AM9/7/12
to neoni...@googlegroups.com
I've built a 4-digit IN-17 nixie clock based on a 1x4 mux design. It uses 4 neon lamps, 2 for the colon, 1 for the PM indicator and 1 for the Alarm On indicator. I have a small problem where I see a dim flicker in the lamps that are supposed to be off. They are driven as part of the nixie multiplexing circuit. One side of each lamp is tied to one of the digit anode drivers. The other sides are tied together and pulled to ground through one common MPSA42 "cathode" driver. I think I have sufficient blanking time between the digits. Each digit is on for 1 ms and the blanking time between digits is 100 uSec. The nixies are showing no ghosting at all. Any help in figuring out why my lamps are not turning fully off would be greatly appreciated. Schematics are attached.
Schematics.zip

Frank Bemelman

unread,
Sep 7, 2012, 10:58:16 AM9/7/12
to neoni...@googlegroups.com
Hi Ron,
 
You could try two resistors of 47K in series, between HV and GND.
Connect a diode 1N4148 or something, cathode to this voltage divider.
Connect the anode to the collector of Q9.
 
Your nixies are fine, because the 74141 has internal clamping zener diodes.
 
Frank
--
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.
To view this discussion on the web, visit https://groups.google.com/d/msg/neonixie-l/-/RssVX0Tgn2kJ.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

dr pepper

unread,
Sep 7, 2012, 11:01:40 AM9/7/12
to neonixie-l
Neon lamps will light from a very small current, I suspect you have a
leakage issue.
Try connecting a 100k or so from q9's base to ground to make sure the
tranny is off.
Maybe reduce the 33k feeding q9's base to 10k as well so the logic low
from the micro helps to hold it off.
You also have the neons connected directly to the micro, you might
want to isolate the micro from the neons and use a similar anode
switching method as you have used with the nixies, one of the
protection diodes of the pic could be conducting causing the neon to
light.
>  Schematics.zip
> 93KViewDownload

Bill van Dijk

unread,
Sep 7, 2012, 11:04:28 AM9/7/12
to neoni...@googlegroups.com

Hi Ron,

 

Providing the MPSA transistor is OK, it should work. There is one thought I have, since the bulbs require very little current to glow, could it be there is enough parasitic capacity in your wiring or circuit board to make them light? What if you were (for a test) to place a resistor of say 1 meg across the bulb and see what happens? It just may bleed off enough current to turn them off.

 

I have also a selfish question. I see you are using a PIC and a DS1302 RTC. I happen to be working on a design (16 segment VFD) using the same parts. Do you program your PIC in assembly or something else? I use assembly, and if by chance you do also, I would be very grateful for a copy of the code section dealing with the DS1302.

 

Thanks,

Bill van Dijk

 

From: neoni...@googlegroups.com [mailto:neoni...@googlegroups.com] On Behalf Of Ron Schuster
Sent: Friday, September 07, 2012 10:39 AM
To: neoni...@googlegroups.com
Subject: [neonixie-l] NE-2 lamps not fully blanking

 

I've built a 4-digit IN-17 nixie clock based on a 1x4 mux design. It uses 4 neon lamps, 2 for the colon, 1 for the PM indicator and 1 for the Alarm On indicator. I have a small problem where I see a dim flicker in the lamps that are supposed to be off. They are driven as part of the nixie multiplexing circuit. One side of each lamp is tied to one of the digit anode drivers. The other sides are tied together and pulled to ground through one common MPSA42 "cathode" driver. I think I have sufficient blanking time between the digits. Each digit is on for 1 ms and the blanking time between digits is 100 uSec. The nixies are showing no ghosting at all. Any help in figuring out why my lamps are not turning fully off would be greatly appreciated. Schematics are attached.

--

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.
To view this discussion on the web, visit https://groups.google.com/d/msg/neonixie-l/-/RssVX0Tgn2kJ.
For more options, visit https://groups.google.com/groups/opt_out.
 
 


No virus found in this message.
Checked by AVG - www.avg.com
Version: 2012.0.2197 / Virus Database: 2437/5254 - Release Date: 09/07/12

Ron Schuster

unread,
Sep 7, 2012, 1:45:01 PM9/7/12
to neoni...@googlegroups.com
All of my PIC programming is in C or Pascal. Sorry, no assembler.

fixitsan

unread,
Sep 7, 2012, 4:09:48 PM9/7/12
to neoni...@googlegroups.com


On Friday, September 7, 2012 4:04:47 PM UTC+1, Bill v wrote:

I have also a selfish question. I see you are using a PIC and a DS1302 RTC. I happen to be working on a design (16 segment VFD) using the same parts. Do you program your PIC in assembly or something else? I use assembly, and if by chance you do also, I would be very grateful for a copy of the code section dealing with the DS1302.

 

Thanks,

Bill van Dijk

 



Just an FYI, I worked on a clock project recently and chose a pic wich has a built in RTC. i think the sacrifice made is limitied EEPROM in the same device. I hooked up a 32.768kHz crystal and the thing ran really well. There is the ability to make very fine adjustments to the clock accuracy too.

I'm currently using a 18F26K22 internal timer off a 32.768kHz crystal, when main power is disconnected the clock can run for very long periods from a supercap by only waking from low power sleep mode to service the 1 second interrupt and then sleeping again.

I think one of the pics with RTC also has an onboard temperature sensor so adding a form of temperature sensing is easy too.

Chris

Ron Schuster

unread,
Sep 7, 2012, 10:13:56 PM9/7/12
to neoni...@googlegroups.com
I've seen an arrangement something like that in some schematics I've seen on the web. Can you explain what that does?

Frank Bemelman

unread,
Sep 8, 2012, 3:25:14 AM9/8/12
to neoni...@googlegroups.com
Hi Ron,
 
Multiplexing was also often seen with LED displays (7 segment) and that worked
very well because the segments are diodes. No currents that flow in unexpected
ways.
 
Nixieclocks of the last decade also often use multiplexing. The ‘segments’ don’t behave
like diodes at all. Many unexpected current paths exist, depending on which anodes are
on, which cathodes are off. Be aware that identical digits of different tubes are connected
in series, because the cathodes of these identical digits are all wired together. Disconnecting
a cathode junction sets all these cathodes floating. There is always 1 active anode. From
that active anode it is not difficult to find two digits in series, which leads you to another
(turned off) anode, but that tube has at least one digit connected to ground. That is not a very
good climate to ensure digits will turn off rapidly. This is the situation if single transistors
are used as switching element. A 74141 has clamping zeners, and solves this problem by
preventing the cathodes going too high.
 
You could also lower the HV. Below 135 volt you will have no ghosting.
 
If you add the voltage divider + diode, the cathodes (of your little neons) are
not allowed to go higher than half HV. They are still in series between anode
drivers, but not enough voltage left over to get ghosting issues.
 
Frank
--
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.
To view this discussion on the web, visit https://groups.google.com/d/msg/neonixie-l/-/VQHzrwTr-qAJ.

threeneurons

unread,
Sep 8, 2012, 2:29:30 PM9/8/12
to neoni...@googlegroups.com
A simple way, is to bypass the leakage. Tack a 470K resistor across each neon bulb.

It only takes a few 10's of microamps, to make it glow a little. With a 470K resistor, there needs to be at least 140uA, before the voltage gets high enough (65V) to even begin to start glowing. If there still is some glowing, then it may be a ghosting issue, someone else mentioned. But, if its ghosting, you should also see it at the nixies.

Ron Schuster

unread,
Sep 10, 2012, 9:57:29 AM9/10/12
to neoni...@googlegroups.com
Great explanation. Thanks! So then I could actually just use a zener in place of that resistor/diode network, right?


On Saturday, September 8, 2012 3:25:41 AM UTC-4, Frank Bemelman wrote:
Hi Ron,
 

Frank Bemelman

unread,
Sep 10, 2012, 10:42:32 AM9/10/12
to neoni...@googlegroups.com
Sure, a zener will also work. It�s just that most junkboxes have so
few high volatge zeners in them ;-)

From: Ron Schuster
Sent: Monday, September 10, 2012 3:57 PM
To: neoni...@googlegroups.com
Subject: Re: [neonixie-l] NE-2 lamps not fully blanking

Great explanation. Thanks! So then I could actually just use a zener in
place of that resistor/diode network, right?

On Saturday, September 8, 2012 3:25:41 AM UTC-4, Frank Bemelman wrote:
Hi Ron,

Multiplexing was also often seen with LED displays (7 segment) and that
worked
very well because the segments are diodes. No currents that flow in
unexpected
ways.

Nixieclocks of the last decade also often use multiplexing. The �segments�
don�t behave
like diodes at all. Many unexpected current paths exist, depending on which
anodes are
on, which cathodes are off. Be aware that identical digits of different
tubes are connected
in series, because the cathodes of these identical digits are all wired
together. Disconnecting
a cathode junction sets all these cathodes floating. There is always 1
active anode. From
that active anode it is not difficult to find two digits in series, which
leads you to another
(turned off) anode, but that tube has at least one digit connected to
ground. That is not a very
good climate to ensure digits will turn off rapidly. This is the situation
if single transistors
are used as switching element. A 74141 has clamping zeners, and solves this
problem by
preventing the cathodes going too high.

You could also lower the HV. Below 135 volt you will have no ghosting.

If you add the voltage divider + diode, the cathodes (of your little neons)
are
not allowed to go higher than half HV. They are still in series between
anode
drivers, but not enough voltage left over to get ghosting issues.

Frank




--
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.
To view this discussion on the web, visit
https://groups.google.com/d/msg/neonixie-l/-/4nHVg5dMvCEJ.

Ron Schuster

unread,
Sep 11, 2012, 9:51:52 AM9/11/12
to neoni...@googlegroups.com
Thank you to everyone who replied to my post. This group is excellent!

The parallel resistor across each bulb is definitely simplest, and from what I have tried, the most effective. When I tried this the bulbs became completely dark when "off". However, this is not my favorite solution. I already have printed circuit boards made and every additional component added after the fact is a pain.

Ron Schuster

unread,
Sep 11, 2012, 10:07:49 AM9/11/12
to neoni...@googlegroups.com
True. Interestingly, mine has more than I had expected. I haven't seen some of this junk for decades. I have 180v, 240v, 300v. I'm not sure why. I've been collecting stuff since the 60's. Of course I had nothing in the needed range! So I strung together a bunch of 12s and 15s to get 66v. That should be close enough, right? Doesn't work. It makes it worse. I also tried the voltage divider with the 1N4148, which did the same thing. I checked the signal with my scope and I can see it clamping nicely at the zener voltage. At this point I realized I had never looked at these signals with a scope before at all. Without this clamping circuit I see what looks like ringing or oscillation riding along the top of the DC level when the transistor is off. It's about 20 or 30 volts p-p somewhere around 8khz. Strange?

Ron Schuster

unread,
Sep 11, 2012, 10:18:36 AM9/11/12
to neoni...@googlegroups.com


On Friday, September 7, 2012 11:01:40 AM UTC-4, dr pepper wrote:
Neon lamps will light from a very small current, I suspect you have a
leakage issue.
Try connecting a 100k or so from q9's base to ground to make sure the
tranny is off.
Maybe reduce the 33k feeding q9's base to 10k as well so the logic low
from the micro helps to hold it off.

I gave this suggestion a try. When I reduced the base resistor to 10k the lamps became erratic. Can't really remember exactly what was happening but it wasn't right. I increased it to 22k and that worked better. I also added the 100k from base to ground. This has reduced the problem significantly. It's not quite as effective as the parallel resistor across each bulb, but it only requires me to add one component.
 
You also have the neons connected directly to the micro, you might
want to isolate the micro from the neons and use a similar anode
switching method as you have used with the nixies, one of the
protection diodes of the pic could be conducting causing the neon to
light.

Not sure what you meant by this. I don't have any neons connected directly to the micro. The lamps are connected between the anode drivers and the Q9 transistor.
 


dr pepper

unread,
Sep 11, 2012, 3:30:24 PM9/11/12
to neoni...@googlegroups.com
Hi again,
 
Good to see the 100k improves things, leakage either is or contributing to the isse then, maybe the base of Q9 is capacitvely coupled to something, I dont know why the 10k messed things up, 10k isnt a low enough resistance to upset the pic, if you pull too much current from the pic and force its current limit into operation then you can have all kinds of issues with the good old read modify write problem. Maybe something else odd is going on there.
 
Also I've had issues with nixies and led's for that matter when I've modified the data on the port controlling the cathodes (segments) before I've switched anodes, my more recent programs blank the display then switch the anode, then write the new value for the cathodes.
 
I think I probably misunderstood your schematics, I cant access them now the forum has changed but from what you describe it sounds ok.
 
R.E. the 32.768kc crystal and timer 1, I have a hard drive pendulum clock on my desk now, the display is on the end of the pendulum, the accuracy is very good, 3 secs per week, using a 32kc xtal is a good idea, my code just polls the bit of timer 1 that changes at 1hz, good power supply decoupling is essential and so's the 30pf or so stability caps for the xtal.

dr pepper

unread,
Sep 11, 2012, 3:38:24 PM9/11/12
to neoni...@googlegroups.com


On Friday, 7 September 2012 15:39:02 UTC+1, Ron Schuster wrote:

Ron Schuster

unread,
Sep 12, 2012, 11:29:06 AM9/12/12
to neoni...@googlegroups.com
I found out why the 10k didn't work. This is being driven by pin 3 on the PIC (RA4) which is an open-drain output. I have a 10k pull-up on it. Reducing the base resistor to 10k loaded this down too much. When I reduce the pull-up to 2k, the 10k base resistor works.

Ron Schuster

unread,
Sep 12, 2012, 11:37:47 AM9/12/12
to neoni...@googlegroups.com
When I said this worked well, I had only tried this on one lamp, and it blanked that one nicely. Now I tried adding the resistors on all the lamps. This works fine for the lamps, but it's causing problems in the nixies; parts of the anodes and/or cathodes are glowing where they shouldn't be.

On Saturday, September 8, 2012 2:29:30 PM UTC-4, threeneurons wrote:

dr pepper

unread,
Sep 12, 2012, 1:00:39 PM9/12/12
to neonixie-l
I wonder if the resistors accross the ne2's is shunting current from
one nixie anode to the one next to it, as the resistors are in series
affectively from one anode to the next when q9 is off.
If you get the same number ghosting then this could be the case, ie if
one nixie is lit 2 and the one next to it it dimly lit 2.
The only solution that springs to mind immediately is having 4 'q9's',
one mpsa42 per neon then there wont be any shunt resistance taking
effect, in fact if you did this all the measures you've taken so far
probably wont be necessary, maybe you've reduced the component count
too much, at least you'd be able to drive 4 mpsa42's from the same o/p
on the pic.
> > its ghosting, you should also see it at the nixies.- Hide quoted text -
>
> - Show quoted text -
Reply all
Reply to author
Forward
0 new messages