HV transistor arrays for nixie anode switching?

[TERRA Operative]

I have an ooollldd Casio 161K nixie tube calculator that has a fault in the main circuit board somewhere and parts are a little hard to come by due to its late 60's/early 70's vintage (Fairchild LSI chips, P-channel MOS IC's etc), so I have decided to build a new drop-in replacement board to get it working in a completely reversible way.

I think I have most of it worked out, except the anode drivers for the nixies.
I am constrained in how the buttons and nixies are already wired, hence the I/O expanders everywhere..

The thing I am a bit stuck on is finding NPN and PNP transistor arrays at a reasonable-ish price that can be used to switch the nixie anodes.
In the current circuit, I am using the common MPSA42 and MPSA92 in the typical circuit you can find all over the internet, but seeing as this calculator has 16 nixie tubes wired so all the cathodes are commoned to each respective pin (all 1 digits together, all 2 digits together, etc) and the 16 anodes are used for selecting which nixie is in use, I need 32 transistors all up (34 including the negative number indicator neon)......

Is there any reasonably cheap, and somewhat easily available transistor arrays, or another part to do the job that I can use here?


(The schematic is attached below. It's incomplete, but if you see any obvious mistakes, let me know )

[schmitt trigger]

Subscribing to thread.

Right now I don’t have any suggestions, but hopefully in the future I will.

[T3sl4co1l]

I don't know any arrays offhand. Diodes/Zetex makes them in SOT-23s so it can't be too too bad on space.

Tim

[TERRA Operative]

Yeah, I'll have a look at the SOT-23 parts. Might be easier than through hole parts.

Board space isn't an issue at all, but hand soldering 32-odd transistors is a little daunting, especially if they can be replaced with a couple arrays instead.

[T3sl4co1l]

Arrays aren't good anyway, beyond doubles, and specialty arrays like ULN's which I don't think you'll find in PNP, or high voltage.

Sure, they make, like, quad '3904 arrays, but they cost almost a hundred times more than singles or doubles.  Boutique stuff, and it's not even monolithic.  Go figure.

Hm, you may actually want to look into electroluminescent (EL) drivers.  I forget if they go quite this high in voltage, and they're usually for AC use so may not support static waveforms, but may be worth a look.

Tim

[pqass]

I don't know of a HV transistor array as I've only used discrete MPSA42 and MPSA92 (+resistors) for my nixie designs.  But you could make use of SIP and DIP resistor arrays to consolidate the passives.

For example, from left (TTL in) to right (+180V switched out to nixie anode) with packages vertical:
10KSIP---10KDIP---MPSA42--100KSIP--82KDIP--MPSA92--33KDIP

Where SIP resistors have one side attached to a common bus (9 pins total) and DIP resistors are all isolated (16 pins).
The transistors are oriented E-B-C horizontally but need to be of alternating hights as TO-92 packages are too fat for 100mil pitch. The emitters interconnect (vertically) on the top side of the board to GND for the MPSA42 or +180V supply for the MPSA92.  That should make it a 1900mils wide by 800mils tall through-hole design.

Multiply above by two for a reasonably compact 16 channel anode driver.  The sign channel is on it's own.  Oh well.

I don't know what to do for a SMT design.

[duak]

Are you aware of the Microchip (was Supertex) serial to parallel output chips?  https://www.microchip.com/ParamChartSearch/Chart.aspx?branchID=9021  I was on a project that useds gobs of the 32 channel parts to switch 100 V, low current outputs to a transducer.

When using the Microchip drivers, try to consider the characteristics of the Nixie vis-a-vis leakage current, operating currents and voltages.  I think you'll be OK with a series resistor to limit current but a diode might be needed for the pull up outputs to prevent ghosting.

Back in the early 70s, I couldn't figure out how the 7441 BCD to decimal driver could switch a Nixie's 180 V circuit even though it was rated for only 70 V maximum.  It turned out that when the Nixie was off, it was far more of an open circuit than the ouput transistors in a 7441 were, and so, it had most, if not all of the supply voltage across it.  When turning on and pulling the cathodes low, the 7441 just had to be able to support enough voltage to cause the Nixie to strike and start conducting after which the VCEs of the transistors dropped down to their saturation levels.

[schmitt trigger]

I also took a look araund, and I've to concur that you'll be stuck with discrete transistors. There are an awful lot in your project, but take consolation that when finished it will be an unique project!

On the other hand, your project also highlights the wonders of I2C and similar serial busses. With only two lines, and a few bus decoders, one can control over 50 I/O individual lines, which previously would have required at least 8 data and 5 address and some latch lines from a microcontroller, plus some ancillary decoders and latches to properly access all these I/O lines.

[NivagSwerdna]

I quite like your design...

I'm a MMBTA42, MMBTA92 and HV5812 fan-boy myself and your design isn't far off that and in fact the zener clamped SN75468Ds look quite nice.

Send some photos!

[TERRA Operative]

Well, I had a poke at it again tonight, I think I'll be using a bunch of MMTBA42's and MMBTA92's, unless the good lord blesses us with suitable HV transistor arrays soon....

I also replaced the resistors in that area with Panasonic and Bourns resistor arrays to simplify things too. (the 10K ones, and also the 82K and 100K, respectively) The Bourns are rated to 200V, so it should all be good.

As for photos, I'll take some when things get built for sure (ans a youtube vid).
Once I have it laid out, I'll have some 3D renders too.


I've never played with I2C before, but it is somewhat handy to be able to hang so much off one bus... I just hope that refresh rate will be fast enough to not see flickering in the digits with so much to scan...
I have a few (probably been thought of by others a long time ago) ideas to speed things up, but firmware is almost completely outside my field of experience so it'll be fun to figure out...
I always seem to jump in the deep end with new stuff and this is no exception........  

[schmitt trigger]

Something that has helped me when I use multiple I2C devices, in the schematic to write besides the IC reference designator, its full 7 bit I2C address.

Of course, the address can be deduced by reading the device datasheet and adding the A0,A1,A2 pin combinations, but it is nevertheless a quick and handy reference.

EDIT: I don't see the pullup resistors for the SDA, SCL lines. Does the nano include them on-board?

[TERRA Operative]

Good idea, I'll do that with the addresses.
Still working on the details on the schematic, the resistors are on the to-do list.

Are you aware of the Microchip (was Supertex) serial to parallel output chips?  https://www.microchip.com/ParamChartSearch/Chart.aspx?branchID=9021  I was on a project that useds gobs of the 32 channel parts to switch 100 V, low current outputs to a transducer.

I'll take a look at this and see what I find, thanks for the link.

[TERRA Operative]

Been tweaking stuff and got the board mostly laid out now.
Turned out pretty ok with SMD transistors and resistor arrays I think.
I kept discreet resistors for the anode resistors as they will dissipate more power than many resistor arrays can handle.

I also added an RTC too. Can't have that many nixie tubes in one spot without the ability to use it as a clock!...... 


If anyone notices anything obviously wrong, let me know.

[greenpossum]

>discreet resistors

Now I know what IC expands to: Indiscreet Circuits. 😉

[TERRA Operative]

Haha, I'll go ahead and leave that typo in there.

Some of the IC's you find around the place could be rather lewd I'm sure.

[T3sl4co1l]

You may want to increase the clearance around those HV traces.  20 or 30 mils should do it (or whatever the SMT device minimum is).

Tim

[TERRA Operative]

Ah, thanks for the pointer. I have 0.5mm on the 180V trace and the 90V trace before it hits the power supply module and steps down to 5V, but I'll double check the rest of the HV path too. I didn't make any special clearance on the other side of the transistors or resistors etc yet. They would be at the normal 0.3mm as per the rest of the PCB.

[KubaSO]

The board looks beautiful!!

[quadtech]

A bit late ... optocouplers can be used for Nixie anode switching -

https://blog.grandtrunk.net/2019/12/nixieclock/
https://www.microfarad.de/nixie-clock/