# Interfacing to tube logic circuitry

Started by November 2, 2011
```Specifically Nixie tubes. I have this counter that has a connector
that provides me with all the active-low transistor drivers for each
digit. What are the gotchas for using these lines with a uC input? I'm
guessing that it's all very high impedance when the transistor is off,
but what happens when digits switch? Isn't there a plasma in there and
can there be high voltage spikes on the output during transitions?

Before you say "measure it", this is just conjectural as the counter
is not here at the moment.
```
```a7yvm109gf5d1@netzero.com wrote:
> Specifically Nixie tubes. I have this counter that has a connector
> that provides me with all the active-low transistor drivers for each
> digit. What are the gotchas for using these lines with a uC input? I'm
> guessing that it's all very high impedance when the transistor is off,
> but what happens when digits switch? Isn't there a plasma in there and
> can there be high voltage spikes on the output during transitions?
>
> Before you say "measure it", this is just conjectural as the counter
> is not here at the moment.

The usual IC would be the 74141:

http://www.g3ynh.info/digrdout/74141.html

The voltage on the anodes of the tubes is in the 150-200V range,
typically. But the 74141 can't do that so it has zeners built in. The
outputs only go up to around 60V, that is enough to cause the Nixie
element to extinguish.

When it is lit and the 74141 pulls low it depends on how big the Nixies
are and thus how many mA each element draws. There can be 1-2V residual.

In a nutshell, no, this is not suitable to drive a uC. You'd need at
least a resistive divider and tolerate hitting the substrate diodes once
in a while. Any connection to these nodes would have to be really hi-Z
or you might get a "sheen" on the Nixies.

--
Regards, Joerg

http://www.analogconsultants.com/
```
```<a7yvm109gf5d1@netzero.com> schreef in bericht
> Specifically Nixie tubes. I have this counter that has a connector
> that provides me with all the active-low transistor drivers for each
> digit. What are the gotchas for using these lines with a uC input? I'm
> guessing that it's all very high impedance when the transistor is off,
> but what happens when digits switch? Isn't there a plasma in there and
> can there be high voltage spikes on the output during transitions?
>
> Before you say "measure it", this is just conjectural as the counter
> is not here at the moment.

If discrete drivers are used, the high level output can go beyond 150V but
the low level may be below 1V. Nevertheless you cannot rely on a voltage
divider only but you'll have to take measures to protect the micros inputs.
The high voltage to power the nixies is not always regulated and may vary
due to load or mains variations.

Suppose the nixies are driven by the usual SN7441/SN74141 decoders, the
general answer is still NO. The high-level output is about 60V, still much
too high for a micro. The low level output can be as high as 2.5V too high
for a reliable low level micro input. Of course you can build an interface
circuit to overcome these problems but at a cost.

petrus bitbyter

```
```If you have the driver already, then it's probably cascoded -- hence
active low.  That's how I'd do it, cascode an MPSA45 or whatever, running
from a ~2.5V reference.  Perfect match to TTL, and regular 5V CMOS will
drive it with insane speed.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

<a7yvm109gf5d1@netzero.com> wrote in message
> Specifically Nixie tubes. I have this counter that has a connector
> that provides me with all the active-low transistor drivers for each
> digit. What are the gotchas for using these lines with a uC input? I'm
> guessing that it's all very high impedance when the transistor is off,
> but what happens when digits switch? Isn't there a plasma in there and
> can there be high voltage spikes on the output during transitions?
>
> Before you say "measure it", this is just conjectural as the counter
> is not here at the moment.

```
```On a sunny day (Thu, 3 Nov 2011 03:00:21 +0100) it happened "petrus bitbyter"
<petrus.bitbyter@hotmail.com> wrote in

>
><a7yvm109gf5d1@netzero.com> schreef in bericht
>> Specifically Nixie tubes. I have this counter that has a connector
>> that provides me with all the active-low transistor drivers for each
>> digit. What are the gotchas for using these lines with a uC input? I'm
>> guessing that it's all very high impedance when the transistor is off,
>> but what happens when digits switch? Isn't there a plasma in there and
>> can there be high voltage spikes on the output during transitions?
>>
>> Before you say "measure it", this is just conjectural as the counter
>> is not here at the moment.
>
>
>If discrete drivers are used, the high level output can go beyond 150V but
>the low level may be below 1V. Nevertheless you cannot rely on a voltage
>divider only but you'll have to take measures to protect the micros inputs.
>The high voltage to power the nixies is not always regulated and may vary
>due to load or mains variations.
>
>Suppose the nixies are driven by the usual SN7441/SN74141 decoders, the
>general answer is still NO. The high-level output is about 60V, still much
>too high for a micro. The low level output can be as high as 2.5V too high
>for a reliable low level micro input. Of course you can build an interface
>circuit to overcome these problems but at a cost.
>
>petrus bitbyter

How about an opto in series with the Nixie segments?
Optos work OK at low currents.
Things wont change fast, so you can use a huge resistor in the output:

opto
TTL driver -- |>|---- nixie segment --- +100
---
/ \   1M
---    ---==--- +5
|   e     |
///        |
micro

I think you canot use a voltage divider, as then the segmants of the nixie
would glow because of the current through that voltage divider.
```
```On Nov 2, 7:16=A0pm, a7yvm109gf...@netzero.com wrote:
> Specifically Nixie tubes. I have this counter that has a connector
> that provides me with all the active-low transistor drivers for each
> digit. What are the gotchas for using these lines with a uC input? I'm
> guessing that it's all very high impedance when the transistor is off,
> but what happens when digits switch? Isn't there a plasma in there and
> can there be high voltage spikes on the output during transitions?
>
> Before you say "measure it", this is just conjectural as the counter
> is not here at the moment.

OK, maybe I wasn't 100% clear. The circuit is the counter inside a
Tektronix 6R1. This is the digital display unit for a early 1960s
sampling oscilloscope. Big. Heavy. Clunky. Bandwidth in the GHz range.

The 6R1 has 4 decimal Nixie tubes, powered by 300V through 82k.
Each tube is driven by 10 open collector discrete transistors,
themselves driven by discrete counters. Not an IC in sight!

There is a large military circular connector that gives me access to
the open collectors.
I believe this was for use with a printer, so maybe it made all
decimals light up through the coils of whatever was inside the
electromechanical printer?

If I use the same type of electronics as back then, I believe I can
use a diode as a switch. If I select a low reverse leakage diode with
also a high enough Vbr, that should isolate the 6R1 from my circuit
and detect the low logic levels.

Would a humble 1N4007 do the trick? Or would it leak too much and
light up the digits?

I'd connect the cathode to the 6R1 side, and the anode to my uC with a
biasing circuit. VCE sat of about .2v in the counter, the diode
conducts and I sense a "0", well, a 0.9 - 1.0V "low".
Transistor goes open, the diode is reverse biased and disconnects the
uC from the high voltage, with leakage current in the uA range. "high"

Make sense? Maybe toss in some protection circuitry on the uC side?
```
```a7yvm109gf5d1@netzero.com wrote:
> On Nov 2, 7:16 pm, a7yvm109gf...@netzero.com wrote:
>> Specifically Nixie tubes. I have this counter that has a connector
>> that provides me with all the active-low transistor drivers for each
>> digit. What are the gotchas for using these lines with a uC input? I'm
>> guessing that it's all very high impedance when the transistor is off,
>> but what happens when digits switch? Isn't there a plasma in there and
>> can there be high voltage spikes on the output during transitions?
>>
>> Before you say "measure it", this is just conjectural as the counter
>> is not here at the moment.
>
> OK, maybe I wasn't 100% clear. The circuit is the counter inside a
> Tektronix 6R1. This is the digital display unit for a early 1960s
> sampling oscilloscope. Big. Heavy. Clunky. Bandwidth in the GHz range.
>
> The 6R1 has 4 decimal Nixie tubes, powered by 300V through 82k.
> Each tube is driven by 10 open collector discrete transistors,
> themselves driven by discrete counters. Not an IC in sight!
>

Yikes! That's true geriatronics.

> There is a large military circular connector that gives me access to
> the open collectors.
> I believe this was for use with a printer, so maybe it made all
> decimals light up through the coils of whatever was inside the
> electromechanical printer?
>
> If I use the same type of electronics as back then, I believe I can
> use a diode as a switch. If I select a low reverse leakage diode with
> also a high enough Vbr, that should isolate the 6R1 from my circuit
> and detect the low logic levels.
>
> Would a humble 1N4007 do the trick? Or would it leak too much and
> light up the digits?
>

trr is frightfully long on those. Better candidate:

http://www.semicon.panasonic.co.jp/ds4/DA2JF81_CED.pdf

> I'd connect the cathode to the 6R1 side, and the anode to my uC with a
> biasing circuit. VCE sat of about .2v in the counter, the diode
> conducts and I sense a "0", well, a 0.9 - 1.0V "low".
> Transistor goes open, the diode is reverse biased and disconnects the
> uC from the high voltage, with leakage current in the uA range. "high"
>

Check where the guaranteed threshold for "low" is for your uC. Sometimes
that's under 1V, a bit close for comfort.

> Make sense? Maybe toss in some protection circuitry on the uC side?

That gets involved. You'd need diodes to VCC and GND, plus another
series resistor from there to the uC. All that times 40 I guess. Plus
possibly a shunt on the VCC rail if it's a very low power uC (or a load
and some big caps). I'd see if it can't be handled by the substrate diodes.

--
Regards, Joerg

http://www.analogconsultants.com/
```
```On Nov 3, 8:18=A0pm, Joerg <inva...@invalid.invalid> wrote:
>
> > The 6R1 has 4 decimal Nixie tubes, powered by 300V through 82k.
> > Each tube is driven by 10 open collector discrete transistors,
> > themselves driven by discrete counters. Not an IC in sight!
>
> Yikes! That's true geriatronics.

FUNatronics!

>
> > Would a humble 1N4007 do the trick? Or would it leak too much and
> > light up the digits?
>
> trr is frightfully long on those. Better candidate:
>
> http://www.semicon.panasonic.co.jp/ds4/DA2JF81_CED.pdf

Good idea.

I need 50 of them, 40 digits, 4 decimal points, 5 symbols (V,S,n,u,m)
and a "print" command that means data valid.
I think the "print" isn't hooked up to a tube though.
Then I need 4 10:4 priority encoders to cut down on inputs to the uC.
```
```<a7yvm109gf5d1@netzero.com> schreef in bericht
On Nov 3, 8:18 pm, Joerg <inva...@invalid.invalid> wrote:
>
> > The 6R1 has 4 decimal Nixie tubes, powered by 300V through 82k.
> > Each tube is driven by 10 open collector discrete transistors,
> > themselves driven by discrete counters. Not an IC in sight!
>
> Yikes! That's true geriatronics.

FUNatronics!

>
> > Would a humble 1N4007 do the trick? Or would it leak too much and
> > light up the digits?
>
> trr is frightfully long on those. Better candidate:
>
> http://www.semicon.panasonic.co.jp/ds4/DA2JF81_CED.pdf

Good idea.

I need 50 of them, 40 digits, 4 decimal points, 5 symbols (V,S,n,u,m)
and a "print" command that means data valid.
I think the "print" isn't hooked up to a tube though.
Then I need 4 10:4 priority encoders to cut down on inputs to the uC.

I come to a pretty extensive circuit to cover the (all?) posibillities.
- The DA2JF81 as discussed.
- The Schottky to prevent the voltage to rise above Vcc
- R1 to make sure the voltage will not stay too low when it should be high.
(I consider leakage currents too unreliable)
- The voltage divider to make sure the voltage to become low enough when it
sould be low. (Only for Vcc<=5V. Not required with 12V CMOS.)
- The series resistor to the encoder and the capacitor may be omitted.

|300V
.-.
| |
| |82k
'-'
|             ___
|         +--|___|-+-----Vcc
.-.        |   R1   |
(   )       |        |
'-'        |        |
|         |        |
| DA2JF81 |Schottky|
+--|<---+-+--->|---+
|       |
|/       .-.
----|        | |
|>       | |
|      '-'
|       |   ___      to
|       +--|___|--+--encoder
|       |         |
|      .-.        |
|      | |       ---
|      | |       ---
|      '-'        |
|       |         |
-----+-------+---------+------GND
created by Andy&#2013266100;s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

If you use a 147 type priority encoder you do not need the 0-outputs of the
nixies. When the nine other outputs are inactive, the output of the encoder
automatically decodes to zero. This way you also prevent the need for extra
measures to detect the suppression of leading zeros.

petrus bitbyter

```
```On Nov 5, 11:11=A0am, "petrus bitbyter" <petrus.bitby...@hotmail.com>
wrote:
> <a7yvm109gf...@netzero.com> schreef in berichtnews:489a4e5e-8231-49f6-bae=
> On Nov 3, 8:18 pm, Joerg <inva...@invalid.invalid> wrote:
>
>
>
> > > The 6R1 has 4 decimal Nixie tubes, powered by 300V through 82k.
> > > Each tube is driven by 10 open collector discrete transistors,
> > > themselves driven by discrete counters. Not an IC in sight!
>
> > Yikes! That's true geriatronics.
>
> FUNatronics!
>
>
>
> > > Would a humble 1N4007 do the trick? Or would it leak too much and
> > > light up the digits?
>
> > trr is frightfully long on those. Better candidate:
>
> >http://www.semicon.panasonic.co.jp/ds4/DA2JF81_CED.pdf
>
> Good idea.
>
> I need 50 of them, 40 digits, 4 decimal points, 5 symbols (V,S,n,u,m)
> and a "print" command that means data valid.
> I think the "print" isn't hooked up to a tube though.
> Then I need 4 10:4 priority encoders to cut down on inputs to the uC.
>
> I come to a pretty extensive circuit to cover the (all?) posibillities.
> - The DA2JF81 as discussed.
> - The Schottky to prevent the voltage to rise above Vcc
> - R1 to make sure the voltage will not stay too low when it should be hig=
h.
> (I consider leakage currents too unreliable)
> - The voltage divider to make sure the voltage to become low enough when =
it
> sould be low. (Only for Vcc<=3D5V. Not required with 12V CMOS.)
> - The series resistor to the encoder and the capacitor may be omitted.
>
> =A0 =A0 =A0 =A0|300V
> =A0 =A0 =A0 .-.
> =A0 =A0 =A0 | |
> =A0 =A0 =A0 | |82k
> =A0 =A0 =A0 '-'
> =A0 =A0 =A0 =A0| =A0 =A0 =A0 =A0 =A0 =A0 ___
> =A0 =A0 =A0 =A0| =A0 =A0 =A0 =A0 +--|___|-+-----Vcc
> =A0 =A0 =A0 .-. =A0 =A0 =A0 =A0| =A0 R1 =A0 |
> =A0 =A0 =A0( =A0 ) =A0 =A0 =A0 | =A0 =A0 =A0 =A0|
> =A0 =A0 =A0 '-' =A0 =A0 =A0 =A0| =A0 =A0 =A0 =A0|
> =A0 =A0 =A0 =A0| =A0 =A0 =A0 =A0 | =A0 =A0 =A0 =A0|
> =A0 =A0 =A0 =A0| DA2JF81 |Schottky|
> =A0 =A0 =A0 =A0+--|<---+-+--->|---+
> =A0 =A0 =A0 =A0| =A0 =A0 =A0 |
> =A0 =A0 =A0|/ =A0 =A0 =A0 .-.
> =A0----| =A0 =A0 =A0 =A0| |
> =A0 =A0 =A0|> =A0 =A0 =A0 | |
> =A0 =A0 =A0 =A0| =A0 =A0 =A0'-'
> =A0 =A0 =A0 =A0| =A0 =A0 =A0 | =A0 ___ =A0 =A0 =A0to
> =A0 =A0 =A0 =A0| =A0 =A0 =A0 +--|___|--+--encoder
> =A0 =A0 =A0 =A0| =A0 =A0 =A0 | =A0 =A0 =A0 =A0 |
> =A0 =A0 =A0 =A0| =A0 =A0 =A0.-. =A0 =A0 =A0 =A0|
> =A0 =A0 =A0 =A0| =A0 =A0 =A0| | =A0 =A0 =A0 ---
> =A0 =A0 =A0 =A0| =A0 =A0 =A0| | =A0 =A0 =A0 ---
> =A0 =A0 =A0 =A0| =A0 =A0 =A0'-' =A0 =A0 =A0 =A0|
> =A0 =A0 =A0 =A0| =A0 =A0 =A0 | =A0 =A0 =A0 =A0 |
> =A0 -----+-------+---------+------GND
> created by Andy=B4s ASCII-Circuit v1.24.140803 Betawww.tech-chat.de
>
> If you use a 147 type priority encoder you do not need the 0-outputs of t=
he
> nixies. When the nine other outputs are inactive, the output of the encod=
er
> automatically decodes to zero. This way you also prevent the need for ext=
ra
> measures to detect the suppression of leading zeros.
>
> petrus bitbyter

Super. I came up with similar stuff. In one case, I simply lifted the
ground of the 10:4 encoder by .7V with a diode.
Another approach is to create a D/A with resistors after the diodes
and use the uC DAC to read the result.
I'm also wondering what can be done with transistors instead of
diodes, high voltage parts.

I'm still chewing over various designs. Since I need so many inputs, I
want to minimize parts count.
I'm getting ready for my winter project.
```