Nixie Tube Schematic Review

[Asstroman]

Hello, this is my first post here in the NeoNixie group so I hope this is an acceptable thread topic.  I'm excited to find other people interested in this sort of thing to get some feedback if anyone has it to offer.  Attached is a preliminary schematic that I designed for my first go at a Nixie clock. (When I say I designed I mean pieced together from the internet and filled in some gaps). I'd greatly appreciate any feedback anyone has on it concerning better methods or any major problems that would prevent if from working. Eventually I'd like to add in more features like acquiring the time via WiFi or GPS, temperature sensing and better dimming control and such, but for now I'm keeping it relatively bare bones (I think). I'm using an Atmega328p as the uC which will be communicating via I2C with a MAX1771ESA+ RTC chip. Even though I know multiplexing shortens the lifespan of the tubes I've decided to go with using MUXing because I'm not so knowledgeable on SMPSupplies (yet) and don't quite know how to get the necessary current out of one to use Direct Drive. My scheme is to use a 3:8 encoder on the anode side to select which anode is on. And on the cathode side I'm using a bin-decimal converter to select which digit is displayed. Each digit of a tube is attached to the same digit on the other tubes so that when the bin-decimal converter turns on the base, say for the "1"s transistor, all the "1"s for each tube will turn on on the cathode side, but only the correct anode will be on, thus displaying only the correct number on the intended tube. I haven't seen another design that does exactly this in the same way so I am not 100% confident it will work. It seems pretty straight forward, but then again, as with all electronics projects it probably isn't. I will include in my code some measures to avoid cathode poisoning, however, I just learned of another issue which is blue spots appearing somewhere on the anode mesh... I think. I still need to do some research about this and how to avoid it but if anyone has any knowledge about this they'd like to share I would appreciate it as I'd like to avoid this from happening in my design. Thanks for your time and I look forward to any and all responses.

Shep

 

[gregebert]

Welcome to an addictive hobby!  

The anode-driver will need some rework because the NPNs (T1 thru T6) will only drive-out 4.3V (5v I/O from the ATMega, minus 0.7Vbe = 4.3V) as the circuit is shown. With multiplexing, you will need what's known as a 'high-side driver', and often that is done with PNP or PMOS devices.

On the schematic, you would swap collector & emitter pins and change the NPN to a PNP.

Next, you would need to add a base-resistor between the digital logic, and the base. This is to limit the base-current to a proper value. For a 170V supply, a maximum of 4mA per tube, and a rough-guess of transistor beta of 50, you would need 80uA of base current. That would be about 2Meg ohms for the base resistor. The exact value isn't critical.

Lastly, you will need a high-voltage logic driver to control the PNP transistors. When the PNP is off, the voltage seen by the logic-driver will 'see' about 170V, which means regular logic gates might get damaged. In reality, the ESD protection of the 74HC238 would sink the 80uA into the 5V supply and turn all anodes on.

You could use an HV5530 to drive the PNP transistors, as well as the tube cathodes. Which means you can get rid of T8 thru T17. A single HV5530 has 32 outputs.

Now, you dont need to display 0000 thru 9999 for a clock, so some cathodes can be left undriven. For example, on a clock only digits 1 & 3 need to display all 10 numerals. Digit 2 displays 0 thru 5, and digit 4 displays 0-2. So you could drive all required cathodes for a clock with only 29 pins on the HV5530, in other words, direct-drive. So you can get rid of the PNP devices as well and make the design much simpler.

[Asstroman]

Hi gregebert!

Thank you so much for your reply. I've taken a look at the datasheet for the HV5530 and it seems like it's a no brainer to use this instead of the transistors to simplify things so I will be taking some time to modify that. However, I do have a few follow up questions about some things you said that don't quite make sense to me that I'd like to understand for my own future benefit. What do you mean when you say the NPN transistors will only drive out 4.3V? The transistor I was planning on using was the MPSA42 which is a high voltage-power transistor. The datasheet says that Vce is about 0.5V, so disregarding the anode resistor, if I were to drive the Nixies with 170V supply, shouldn't the anode see about 165.5V. Also, using an NPN I thought that the 74HC238 would be sourcing the current rather than sinking it. Am I incorrect in this thinking? Thanks again for your reply and helping me to understand this stuff.

Shep

[David Forbes]

Shep,

There are two polarities of transistors, NPN and PNP. NPN requires the base to
always be 0.6V more positive than the emitter, and PNP requires the base to be
0.6V more negative than the emitter.

The anode must be controlled from the 180V point, not from the 0V point. The
most straight-forward way to do this is with an NPN transistor whose emitter is
connected to 180V. The base is controlled by a high voltage NPN transistor
through a resistive divider, to allow the on/off command to be sent across the
180V 'space'.

An example of a multiplexed anode drive circuit that works is shown in my old
Nixie clocks... page 17 has the schematic diagram.

http://www.cathodecorner.com/nc620akitman.pdf



On 3/29/2016 2:51 PM, Asstroman wrote:
> Hi gregebert!

> What do you mean when you say the NPN transistors will only drive out
> 4.3V? The transistor I was planning on using was the MPSA42 which is a high
> voltage-power transistor. The datasheet says that Vce is about 0.5V, so
> disregarding the anode resistor, if I were to drive the Nixies with 170V supply,
> shouldn't the anode see about 165.5V. Also, using an NPN I thought that
> the 74HC238 would be sourcing the current rather than sinking it. Am I incorrect
> in this thinking? Thanks again for your reply and helping me to understand this
> stuff.
>
> Shep

--
David Forbes, Tucson, AZ

[gregebert]

Circuit-wise, the NPN's T1 thru T6 are emitter-followers. In order to turn them on, the base-emitter junction needs to be forward-biased, which is about +0.7 volts. So, if the base is driven to +5v (the max possible from the 74HC device), the max-possible voltage at the emitter will be 5.0-0.7 = 4.3 volts. It makes no difference how high the collector voltage is (unless it's SO high that you get breakdown).

You are correct that the 74HC will source base current, but that will only happen as long as the base-emitter junction is forward-biased. For the sake of argument, let's imagine the base is at +5V and the emitter is around +160V. In this case, the base-emitter junction is reverse-biased, which means the transistor is off and no collector current (other than leakage) is possible. Stated another way, it's not possible for the circuit as drawn to produce more than +4.3V at the emitter.

[David Forbes]

Sorry, I meant to say a PNP transistor with emitter tied to 180V rail.

On 3/29/2016 3:23 PM, David Forbes wrote:
> Shep,
>
> There are two polarities of transistors, NPN and PNP. NPN requires the base to
> always be 0.6V more positive than the emitter, and PNP requires the base to be
> 0.6V more negative than the emitter.
>
> The anode must be controlled from the 180V point, not from the 0V point. The
> most straight-forward way to do this is with an NPN transistor whose emitter is
> connected to 180V. The base is controlled by a high voltage NPN transistor
> through a resistive divider, to allow the on/off command to be sent across the
> 180V 'space'.
>
> An example of a multiplexed anode drive circuit that works is shown in my old
> Nixie clocks... page 17 has the schematic diagram.
>
> http://www.cathodecorner.com/nc620akitman.pdf

[Asstroman]

David and gregebert,

Thanks for your replies, they were very helpful. It seems I have had a somewhat profound misunderstanding of the operation of transistors but I've spent the last few hours looking at the schematic David referenced as well as other articles and youtube videos and I'm beginning to understand a little better. I kept thinking that once the transistor was saturated it acted pretty much like a short circuit from collector to emitter. Thanks again for your input I appreciate it very much!

Shep  

[gregebert]

Yes, a saturated transistor behaves similar to a short-circuit, but in order to do so it must still be properly biased.

The basic rule is: a bipolar transistor will conduct collector current when the base-emitter junction is forward-biased.

For an NPN transistor, the base & collector both need to be positive with respect to the emitter

For PNP, base & collector need to be negative wrt emitter.

------------------------------------------------------------------------------------------

MOSFETs are "similar" to bipolar transistor, though trickier because they generally have reverse-diodes between source ("emitter") and drain ("collector"). For most applications, the reverse-diode is a godsend and has no undesireable effects. They dont require any static drive current (the gate is synonymous with the base of a bipolar), but they do require several volts to turn-on rather than ~0.7V for a bipolar device.

I've used bipolar and MOSFET drivers for nixies; which one to use depends upon the application.

[Jonathan F.]

What parts do you intend to use for the power supply?

If you want to save parts, use a 7805 regulator for the 5V, you don't need R1, R2,PC2 then.

About Multiplexing, if you have a 1:6 Multiplex, then every tube is only one sixth of the time on, resulting in lower brightness. A IN-18 is specified to have 4-6mA per tube. If you want the same brightness that direct drive would have, you have to drive the IN-18 with six times the rated current! But, personally, for home usage i think most clocks are too bright at rated current, especially at night. So a lower brightness may be nice.

You could replace the NPN Transistors with mosfets, they can be driven directly by 5V, no more parts needed.

[Asstroman]

Hello Jonathan,

I may eventually end up using a fixed voltage regulator but the reason I have an LM317 in this schematic is that I already have these and necessary resistors so it was mainly just to keep cost down.  And that's an excellent point about multiplexing that I hadn't really though of so I thank you for that. After this and the previous comments I think I may just use direct drive. I'm not sure why but I was originally thinking that multiplexing would be more feasible because I wasn't sure if I could get enough power out the power supply. But after I took another look at it I think the power supply will have no problem driving the necessary ~38mA. I'm still weighing the pros and cons as I figure them out.

Shep

On Tuesday, March 29, 2016 at 1:04:19 PM UTC-6, Asstroman wrote:

[Jonathan F.]

Hello Shep


Some words about the power supply.
I would change the inductor to a higher value, about 100uH. And RSENSE to 0.05R, which could deliver about 50mA.
C4 is fine, but 4,7u would be enough.

Here is a very fine and detailed power supply using the MAX1771  http://desmith.net/NMdS/Electronics/NixiePSU.html


For direct driving, you could use the HV5122 or a similar device. But you would need a 5V to 12V level shifter. If you need more info, i can provide.

[Asstroman]

Hi Jonathan,

Okay I'll check out that page and see if I can understand why changing those values would be better. As per gregeberts recommendation above I have already reworked my design and put in the HV5530. I have 12V coming in from a wallwart to the power supplies and right now I just have this powering the HV5530. With some additional regulation and filtering would this be okay or is it necessary to level shift it from the 5V?

Shep

[gregebert]

You definitely want to use a level-shifter for reliable operation. I just finished a 14-tube nixie clock with HV5530's driven from an FPGA; I use a MC14504. Some people on this forum have driven HV5530's without a level-shifter; to me saving a few dollars on parts is silly compared to spending hundreds on nixie tubes.

If you are going to use more than 1 HV5530 (necessary for a 6-digit clock), then be careful about cascading the HV5530's. The datasheet info is incomplete, so it's possible to have a timing-violation if you connect the serial-out to the serial-in of the next HV5530 and share the same clock signal. Better to use 2 separate serial-data-in signals. In my case, I had to cascade two HV5530's because my ribbon-cable didn't have enough signals (a long story), but I used 2 different clocks.

The MC14504 has 6 level-shifters, so you can drive all of the necessary control signals for 2 HV5530's (clk, data_in, le, bl, pol). It's very easy to make a mistake on the 'bl' and 'pol' control-signals, so if you decide to tie them off, make sure you use resistors so you can pull high or low on the PC board. I drive them from my FPGA.

-----------------------------------------

FYI - The ribbon cable in my clock is over 3 feet long, and in a noisy electrical environment, it was absolutely necessary to terminate the clock-lines to control reflections. Even though the cable's impedance is roughly 120 ohms, I got good results (from simulations and bench measurements) terminating at 330 ohms. Though terminating at 120 ohms is best from transmission-line theory, it puts a lot more loading on the driver. As long as the reflections are well-below the threshold-voltage of the MC14504, they are harmless.

[gregebert]

Forgot to mention this: Tube sockets

As you know, IN-18's are getting rather expensive so you need to be very gentle with them.

I decided to use socket-pins soldered onto the PC board (inexpensive & available on Ebay), rather than actual 'sockets', because my past experience with nixie-tube sockets found they require quite a bit of force to insert & remove. All of my IN-18's, except 1 oddball manufactured in 1977, have very soft pins, and I'd be afraid to push them into a socket. Even though my clock sits on the shelf and I hope I will never remove the tubes again, I had to insert & remove all of them a few times while building the case.

If you decide to use socket pins, be sure to number each of your IN-18's and assign them to a socket location on your PCB. Then push the socket pins onto the IN-18, and solder to the PCB. This will minimize the stress on the tube because the socket-pins are custom-fitted to each tube; once you do this, though, you dont want to swap tubes. You can easily remove the tubes; re-installing requires some care but very little force.

I created a .stl file for a 3D printable pin-shell, but havn't bothered to build it because the tube is held securely enough without it.

[Asstroman]

Hello again everyone. Thank you so much for your help with the last version of my schematic design. After taking all that input into account and doing some additional research I've redone my design and have posted it below. Any additional input is welcome if you care to give it. Thanks again.

Shep

[gregebert]

Looks like you have enough bypass caps; just be sure to place them as close as possible to the supply pins on the ICs. Good to see you have separate clk & data lines to the HV5530's; you now have total flexibility to avoid timing problems.

Be sure to provide enough spacing around high-voltage signals on your PCB. Here's a handy calculator: http://www.smps.us/pcbtracespacing.html

The only suggestion is to add an ESD bleeding resistor of 1Meg (approx; not critical)  between your circuit GND and the actual power-line ground if you have that connection in the case.  You already have a resistive path from your HV to GND which will discharge your HV cap.

When I have multiple PC boards that are cabled together, I always have ESD/bleed resistors between all power supplies and GND. Usually, the largest & cheapest SMT resistor I can buy, which is 10meg, suffices for HV supplies.

[Jonathan F.]

About this calculator for insulation... if i type in 300V, i get 2.5mm for external layers.

So.... how is it possible for Microchip to sell a 315V rated IC in a PQFP44 case? The pins on a QFP have 0.5mm pitch!

[gregebert]

If you are driving nixie tubes, there actually isn't 300V between pins at the driver IC. If your supply is 200V, and the typical voltage drop of an IN-18 is about 140V when illuminated, the actual voltage-difference between pins is about 60V. That calculates to 0.6mm of spacing, which is still a violation for a 0.5mm pad-pitch (remember, pad-separation is less than pad-pitch due to finite pad-width).

My guess is that someone seeking regulatory approval/listing would apply conformal coating, or similar.

Even if you can't meet spacing rules around nixie cathode connections, you should use extra effort to meet or exceed spacing rules between high- and low-voltage sections, and also between HV connections that can have high current, such as filter-caps and mains-connections.

[David Forbes]

I use the Toshiba TD62083AFN in my Nixie watch as a cathode driver. It sees 50V
between pins on the 0.65mm TSSOP. I take extra care to keep the anode traces
well separated from all the cathode and logic traces on the PC board. I use
.010" (.25mm) trace spacing between cathodes, and .020" (0.5mm) between anode
and other traces. I also arrange the connector pinouts to minimize voltage
between adjacent pins.

A wristwatch is the most difficult Nixie product to do a PC board layout for,
since there is just not much room on the driver PC board. I've shipped over a
thousand watches over 10 years, without a single PC board arc-over problem.

[GeckospotNixie]

Just out of curiosity what Microchip IC is rated at 315V?

[Jonathan F.]

My point wasnt about nixies at all, i just wondered how microchip can supply this ic, because it would never fit these regulations!

About the ic , HV507 is a 315V 64channel sink-source driver

[Asstroman]

Thanks everyone for your help I appreciate it more than you know. I ended up taking into consideration pretty much everything you suggested. I've ordered it and will post a finished product in a few weeks...hopefully.

Shep