Push-pull driving cathodes

[Jasper C.]

I'm still new when it comes to nixie tubes, and was reading this page, in particular the "partial glow" that occurs if the prebias voltage of an "off" cathode is too low.  Wouldn't that imply that the ideal solution would be to pull the cathodes to the supply voltage of the anode when their respective segment is off?

Only in this EEVblog video, starting at 14:25, he does just that and it doesn't go well.  I get that the anode voltage here is lower than the HV supply because of the 22k resistor, though I don't understand why that then has the effect that it does.  Is the segment shorted to HV acting as an anode? (Which might explain the higher current he's seeing.)

Would a totem pole circuit like that work if the anode resistor was removed and each cathode connected to the totem pole output by it's own resistor?

[David Pye]

Yes, that's it.  If the other segments are high, they just work as more anodes.

The best thing imho is just to leave them floating, but they will then float up to the anode voltage. So provided you have drive circuitry that can switch the full anode voltage, that is ok.

If your drive circuitry cannot handle the full anode voltage, you need to either keep at least one cathode on at all times (ie not use cathode blanking) or have a pull down resistor that holds them at say, 70v so they don't glow, and use the drive circuitry to put them from that holding voltage to ground.

David

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/144bce53-142d-44be-9a9b-fae08fafab7b%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[gregebert]

For direct-drive, you can just let the cathodes float, and they will be completely dark unless your driver has leakage. I've done 5 full designs with direct-drive that floats the cathode, and none has any issues. All told, there are 52 nixie devices in operation with no hint of failure or leakage. 24 of those tubes have been in service 24/7 for the past 8 years.

There are some variations in the design, mainly for cathode and/or anode-current limiting but they all float the 'off' cathodes.

Admittedly, I've never done a multiplexed design with nixies, only with LEDs, and unless the nixie tube is specifically designed for multiplex operation I would stay with direct-drive. The reason I follow is that multiplexed digits need a higher pulsed current to maintain visible brightness, and higher current leads to tube wearout.

[David Forbes]

The voltage applied to the elements of a Nixie tube is not as straightforward as you may think. Keep in mind that there is a voltage drop from anode to an 'on' cathode of ~130-140 volts, due to the ionized gas. Also, the capacitance of the tube elements and wiring has an effect on the switching behavior.

I have successfully made several varieties of multiplexed Nixie clocks and watches using 60V or lower cathode switches. I used printed circuit boards to reduce capacitance, and adjusted the timing to allow the recently turned off cathodes to drift up to their natural 'off' voltage before turning on the next digit's anode.

My Nixie watch design uses 50V cathode switch arrays. It also has no anode resistor. This is done by enabling only one cathode at a time, leaving the other tube with no cathodes enabled. 

The power supply has a resistor current sense and feedback mechanism to regulate the total current as well as the open-circuit voltage. Blanking is achieved by reducing the anode voltage to 100V, with a digitally controlled shift to the regulator feedback. 

So you can get good results with very little circuitry, if you apply a bit of cleverness to the problem.

http://www.cathodecorner.com/nixiewatch/firmware/nwrf-schem.gif

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/e168a850-f3d7-4e88-adc8-1b3bd32b1902%40googlegroups.com.

[John Smout]

I am currently working on a set of multi-digit panaplex displays. One is a 16 digit, 16 segment alphanumeric display and the other is 6 digits and 7 segments. Both were originally intended to be multiplexed across all the digits. There is a 'keep alive’ cathode and an 80v mid rail to keep the display just below firing level. A consequence is that both displays like some off-time before a new selection is made, otherwise I am seeing a very ‘hot’ display with arcing over to supposedly unlit segments. It’s a bit of a fine balance to get right, but software can cope with this.

John S

[Jasper C.]

On Sunday, 3 March 2019 20:53:04 UTC+8, David Pye wrote:

 

Yes, that's it.  If the other segments are high, they just work as more anodes.

The best thing imho is just to leave them floating, but they will then float up to the anode voltage. So provided you have drive circuitry that can switch the full anode voltage, that is ok.

I'm a little confused there, about it being OK to switch the full anode voltage.  Do you mean that the low side switch can handle the full HV, or that I switch my cathodes to HV?  The latter would contradict the "working as more anodes" issue.

[Jasper C.]

So I was originally looking at the Microchip HV513.  It's a 250V totem pole driver with 5V input logic and a latch.  Was going to do a quick sketch of what I had in mind for the output, but haven't had the time to get round to it.

I was planning to not multiplex my nixies, probably leaving all cathodes connected directly to the HV line, and placing current liming resistors on the cathodes.  Was interested to see if I'd need to tweak any of the individual resistor values uniform current/brightness.

If I implemented only a low side switch, I was looking at something like the MMBTA42, available from a few manufacturers and is good for 300V, but I haven't yet found any information about leakage currents when the device is off.  Alternatively I think I found a mosfet or two with sufficient V(DS) and logic level switching, but I need to recheck the D-S leakage.

[Jasper C.]

On Monday, 4 March 2019 04:12:55 UTC+8, nixiebunny wrote:

The voltage applied to the elements of a Nixie tube is not as straightforward as you may think. Keep in mind that there is a voltage drop from anode to an 'on' cathode of ~130-140 volts, due to the ionized gas. Also, the capacitance of the tube elements and wiring has an effect on the switching behavior.

I have successfully made several varieties of multiplexed Nixie clocks and watches using 60V or lower cathode switches. I used printed circuit boards to reduce capacitance, and adjusted the timing to allow the recently turned off cathodes to drift up to their natural 'off' voltage before turning on the next digit's anode.

My Nixie watch design uses 50V cathode switch arrays. It also has no anode resistor. This is done by enabling only one cathode at a time, leaving the other tube with no cathodes enabled. 

The power supply has a resistor current sense and feedback mechanism to regulate the total current as well as the open-circuit voltage. Blanking is achieved by reducing the anode voltage to 100V, with a digitally controlled shift to the regulator feedback. 

So you can get good results with very little circuitry, if you apply a bit of cleverness to the problem.

http://www.cathodecorner.com/nixiewatch/firmware/nwrf-schem.gif

The feedback regulated power supply is a very elegant solution, and I'd love to implement something like that, but my knowledge of switching power supplies isn't good enough to pull it off.  I envy you.

[gregebert]

The required voltage rating of a cathode driver has been debated in this forum a few times. There are 2 opinions that I know of

• The driver needs to be rated at the full anode-supply voltage
• The driver only needs to be rated at (Anode_supply_voltage - Voltage_drop_across_nixie_tube)

Based on my knowledge of semiconductor physics, it's fine to use a bipolar (NPN) cathode driver that is rated at a lower voltage than the anode supply because the breakdown mechanism is non-destructive as long as the current is limited. When an NPN is off, presumably when it's base terminal is grounded (NOT open...), there will not be any current gain from collector leakage current, so it will stay off. I suspect an OPEN base could result in some visible glow, depending upon the collector leakage current Ico. This is why legacy drivers like the K155 or 7441 can drive a nixie that requires about 160V even though the IC itself can handle only 50 V.

If you are using a MOSFET as your segment driver, which is typical in IC's, then you need to use a driver rated at the full anode supply voltage because a MOSFET's construction is sensitive to excess voltage (oxide breakdown) as well as current. Long-term overstress of the oxide will cause reliability problems and lead to failure.

[David Forbes]

Another important fact about Nixie tube cathodes: turning on one cathode by pulling it to zero volts causes it to steer all the available current away from the other cathodes, causing them to be dark. This is why you only need a 50V switch on each cathode. 

The caveat is that if no cathodes are pulled to zero volts, then there will be leakage current flowing through the tube that will destroy a 50V transistor. This is why I used the TD62083 with its set of commutation diodes, and connected the diode common anode pin to a 50V source in my Nixie watch circuit.

It's also why I made the blanking mode that drops the anode voltage to 100V, so that the cathodes all remain dark when blanked. 

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/c0c015b5-15e9-4387-b7f7-911193e174a4%40googlegroups.com.

[David Forbes]

Jasper,

I didn't use any fancy switching power supply knowledge - in fact, I
copied the switching power supply schematic straight out of the LT1308
data sheet!

The clever part was figuring out that I could do things with the
feedback resistor network to achieve current regulation and blanking
voltage control. The power supply adjusts the output voltage as needed
to make 1.2V appear at the feedback node. The rest is just simple
resistor networks, as taught in any intro to electronics class.

[Jasper C.]

I was planning to not multiplex my nixies, probably leaving all cathodes connected directly to the HV line, and placing current liming resistors on the cathodes.  Was interested to see if I'd need to tweak any of the individual resistor values uniform current/brightness.

I only just reread what I wrote there, and realise I made a typo.  I should have said I was going to connect all the anodes to the HV line...

On Thursday, 7 March 2019 00:41:00 UTC+8, nixiebunny wrote:

Another important fact about Nixie tube cathodes: turning on one cathode by pulling it to zero volts causes it to steer all the available current away from the other cathodes, causing them to be dark. This is why you only need a 50V switch on each cathode. 

The caveat is that if no cathodes are pulled to zero volts, then there will be leakage current flowing through the tube that will destroy a 50V transistor. This is why I used the TD62083 with its set of commutation diodes, and connected the diode common anode pin to a 50V source in my Nixie watch circuit.

It's also why I made the blanking mode that drops the anode voltage to 100V, so that the cathodes all remain dark when blanked. 

Ah, well that comes back to the page I referenced in the opening: http://www.decodesystems.com/re-how-nixies-work.html.  Maybe that was written in reference to having leaky cathode transistors? (With no voltage clamp, as in Fig. 4.)  If I had say 300V+ transistors with sufficiently low leakage when off (< 10 uA) and a 170V power supply, would I be able to blank the tube safely by turning all transistors off?

I'll have another look at power supplies. :)  I was a little concerned about how to calculate the value of the loop compensation components.  For my first attempt I figured I'd avoid the problem altogether by using one of Yan's power supplies, and focus on the segment driving circuitry.

 

[David Pye]

Yep.  A lot of designs just use individual HV transistors for each cathode.

I recommend mpsa42 personally as they are ideal for cathode driving and not expensive.

They come in several different packages including sot23 SMD if you need small..

David

--

You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/7026df09-ef22-408c-9e92-7daa3f5f2db6%40googlegroups.com.

[newxito]

I was using the good old MC34063 on all my boards but I just did a test with the LT3757 (already mentioned in this group) using the high voltage flyback power supply datasheet schematic with a DA2032 transformer. I’m really impressed, 170V up to 69mA at 12V with around 86% efficiency, just a bit warm, no audible noise. 

I’m not an electronic engineer, I don’t really know what I’m doing :-) So, I’m sure that with a correct board design the efficiency could be improved.

 

[lokn...@gmail.com]

Check out this guys projects. He has an extremely efficient power supply that is built for 12V but will run on a 5v 1A USB Brick/Wall-Wart. It is an Open-source project with files on github.

https://surfncircuits.com/

Collin

On Mar 9, 2019, at 14:49, newxito <axt...@gmail.com> wrote:

I was using the good old MC34063 on all my boards but I just did a test with the LT3757 (already mentioned in this group) using the high voltage flyback power supply datasheet schematic with a DA2032 transformer. I’m really impressed, 170V up to 69mA at 12V with around 86% efficiency, just a bit warm, no audible noise. 

I’m not an electronic engineer, I don’t really know what I’m doing :-) So, I’m sure that with a correct board design the efficiency could be improved.

 

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/0f043704-27dd-4e04-aa56-78d8a9602c5f%40googlegroups.com.

[Paul Andrews]

Or check out mine. Also open source. 3V-18V: https://hackaday.io/project/27899-nixie-tube-power-supply

[Frank Bemelman]

L2, 60 ohm? Must be a mistake of some sort..

Op zo 10 mrt. 2019 om 13:10 schreef <lokn...@gmail.com>

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/B3EE01EA-4F9B-4465-90AA-8BE54834FDE2%40gmail.com.

For more options, visit https://groups.google.com/d/optout.

--

Vriendelijke groeten,

Frank Bemelman

Frank Techniek

Vennestraat 11X

2161LE  Lisse

Telefoon +31 252 769 178

[newxito]

After further testing, the use of LT3757 (voltage regulator), DA2032 (transformer),  CSD19533Q5A (MOSFET) and RFN2LAM4STR (diode) seems to be a good combination for a nixie PSU. Now trying to optimize the board design.

[SWISSNIXIE - Jonathan F.]

The LT3757 looks quite like a nice ic, i will try it with the transformer PA0367A from Pulse elctronics, its quite small. Since its only availabe from digikey, i have to wait until i have an order big enough to order at digikey :/

Probably alot of people are not into new supplies because mostly (like in any other electronics project) the new and good IC's don't come in SOIC or DIP anymore, so you need to know how to solder leadless packages.... In a few other hobby projects i run into the problem that the ic of choice was only available as BGA...

One thing about nixie power supplies is usually that the efficiency is alway given at like 50mA, but usually a normal nixie clock uses way less. Even 6 IN-18 would only require around 40mA supply. Smaller builds tend to use 25mA or even 15mA. Nearly every converter is inefficient at low currents, since the losses are in absoulte values not in percent. "loosing 5mA" at 50mA would give you give you 90%, but the same design probably also looses 5mA at 15mA output so that would be way more bad. Alot of people are fooled by this fact.

I saw a quite cool design by linear technologies that has around 85% efficiency at 20mA Ouput, which is enough for a IN-8 clock or so.

[Paul Andrews]

I use the LM3478. It seems to have similar specs and comes in a SOIC version too. I was going to evaluate several, but the LM3478 worked well enough for my purposes!

[newxito]

I think there is a new version called PA0367ANLT

Yes, soldering this ic is a major issue, I don't have a reflow oven. For me using very little amount of solder paste and a hot air gun works fine.

You're absolutely right, with 12V the test PSU has the best efficiency at 60mA (89%) but then it goes down (86% at 25mA and 81% at 10 mA). For low current designs seems to be better to use 9V or even 5V as input voltage.