How to power 18 IN-9?

221 views
Skip to first unread message

Mikołaj Walkowiak

unread,
Mar 21, 2018, 3:52:45 PM3/21/18
to neonixie-l
Hello everyone! I'm doing project for college and I've decided to make an audio visualiser using IN-9's. The problem is I don't have any idea how to power them. I'd like to have 18 tubes for different sound frequenties and I'll be using stm34 to calculate the value each tube should show..

Now, I know I need 140v and 0 - ~~12 mA for 1 tube but that's all I know honestly. I think I'll use something like the 0-offset driver found in HERE controled by PWM of my uC but I honestly have no idea how to power them. Is 140V(or 150V? everywhere except there it says 140V) enough even for 18 tubes? If yes is it possible to make this kind of supply myself as I can't find anything reasonable price-wise with 140V. Do I need to consider anything else?

Paul Andrews

unread,
Mar 21, 2018, 4:21:48 PM3/21/18
to neonixie-l
I hadn't seen that paper before. Nice find.

There are several power supplies that will give you 150V, seatch for adjustable nixie power supplies. But as the paper says, this is a current-driven device, so as long as you limit the current through each tube, you can go higher. If you go higher, you will need to increase the value of the current-limiting resistor to keep the current the same.

The limit for 18 tubes is the power. Your power supply needs to be able to provide 150*.012*18 (i.e. 33) watts without collapsing.

Or you could use multiple power supplies, each powering a group of tubes.

Paul Andrews

unread,
Mar 21, 2018, 4:28:54 PM3/21/18
to neonixie-l
My power calculation is wrong! Not all that voltage drops across the resistor, so I think it is more like 10 watts - I new I should have taken more time!

Mikołaj Walkowiak

unread,
Mar 21, 2018, 4:40:18 PM3/21/18
to neonixie-l
HI! Thanks for the quick reply! I also thought I'd need to find over 30W so that's good news. Do you think something like THAT is enough for all 18 tubes?

Mikołaj Walkowiak

unread,
Mar 21, 2018, 4:42:07 PM3/21/18
to neonixie-l
Oh wait that's from china ;/ maybe there is a way to convert 12V 30W power supply to a 140V?  

Nick

unread,
Mar 21, 2018, 7:37:14 PM3/21/18
to neoni...@googlegroups.com
Think of it as 18 tubes all driven at 12mA, giving a maximum total requirement of about 240mA @ around 150VDC, i.e around 40W. You need a bit of headroom, i.e. 50W and up.

You are therefore in transformer territory.

Typically, you'd overrate the transformer, so you'd be looking like one having a 120V @ 500mA secondary which when run through a bridge rectifier and smoothed with about 2,200uF will give around 150VDC out. A small toroid, e.g. a Hammond 1182G117, might fit the bill.

Edit: the paper you found was written by a neonixie-l member, Jeff Malins, in 2004 after a long discussion on the original Yahoo group...

Nick

John Murphy

unread,
Mar 22, 2018, 8:23:24 AM3/22/18
to neonixie-l
I am currently working on a very similar project.  Here is what I've learned:

1. The IN-9 tubes don't all act properly using a 150V DC switching power supply (I bought several different models from eBay).  The glow of the tube becomes disconnected from the bottom and "floats" in the middle - this is not how a bargraph should behave.  I have been successful using a full wave rectifier circuit from 120V mains through a 1:1 isolation power supply (US).  If you have 240V mains, you would need a 2:1 step-down transformer.  I used the KBP307 rectifier instead of discrete diodes.  Do NOT try to use the rectified mains without a transformer.  In addition to this being dangerous, if the DC 0V has a path to earth ground, the rectifier will be destroyed, and also likely take out other components like the uC.  I am using a Triad VPS230-110 as it's fairly inexpensive.  It has a maximum output current of 220 mA, but for a music visualizer, you won't have full scale output on all the tubes at the same time - you could enforce this using code in the microcontroller if you're worried about it.

2. Most microcontrollers have a limited number of pins that you can output PWM on.  If you want to drive 16 tubes, you need 16 PWM (or analog out) pins.  I am using two serial octal DACs (LTC1665) to drive the base of the MJE340 transistors using the circuit in the paper you referenced.  DAC values are set with a synchronous serial stream. The clock and data lines are common to both DACs and each one gets its own chip select for a total of 4 digital pins needed to drive 16 tubes.

3. If you buy New Old Stock (NOS) tubes, some of them won't initially glow to full height due to being in storage for so long.  They need to be "burned in" - something about the physics of the cathode tube that I don't really understand.  I have found that driving them at 15mA for about an hour has been enough.  Over that time the glow gradually creeps from about 1/2 scale to full scale.  After that, they all have worked to full scale.

Have fun!





On Wednesday, March 21, 2018 at 3:52:45 PM UTC-4, Mikołaj Walkowiak wrote:
Full wave rectifier circuit.png

Mikołaj Walkowiak

unread,
Mar 22, 2018, 8:46:36 AM3/22/18
to neonixie-l
Thank you SO MUCH! You've solved most of my problems I have right now! Thank you once more ang good luck to you too!

Alic

unread,
Mar 22, 2018, 11:09:32 AM3/22/18
to neonixie-l
I read that glow disconnecting from the bottom has to do with the speed at which the signal changes. So the « steepness » of the signal. To prevent that, Pete Virica has implemented a filter in his bargraph kit (page 10) :
http://www.pvelectronics.co.uk/kits/DuoDrive/duodrive_v1.pdf

Mikołaj Walkowiak

unread,
Mar 22, 2018, 11:38:17 AM3/22/18
to neonixie-l
Thank you for that link! I was well aware of that but it's nice to know there is someone who've solved that for me!

Mark Moulding

unread,
Mar 22, 2018, 1:28:24 PM3/22/18
to neonixie-l
You beat me to writing a similar reply to the original post, but I wanted to expand on your use of the lines-powered transformer.  One way to completely avoid dealing with the direct AC line at all (although not high voltages, of course) is to use back-to-back low voltage transformers.  An off-the-shelf 12-volt AC "wall wart" can then be plugged into a jack on your project that connects to the secondary of a 12.6V filament transformer.  If you ever decide to sell one of these, this is a *huge* benefit, because it completely short-circuits (so to speak) the entire UL/CSA approval process.  This also gives you the option of adjusting the final output voltage, by simply using another voltage wall wart (9 or 15 VAC, for example).  Note that the inexpensive wall warts *always* run high in voltage at less than their rated load.

One thing to consider would be the addition of a filter capacitor to this circuit.  As others have said, the glow of the IN-9 tends to detach from the bottom if it sees too much of a transient, and this includes the power supply.  Conversely, other people seem to have had good luck driving the tubes with pulsating unfiltered DC...

If you do add a filter cap, you might want to insert a low-value (22 ohm, perhaps) in series with the one of the bridge lines to prevent a large current surge when first powering up.  Also, make sure the capacitor is adequately rated for the job.  You wouldn't need too much capacitance - 50 uF would be sufficient, but it must be at least a 250V rated unit.  This brings up another point: with the circuit shown and the addition of a filter capacitor, the output voltage available will be considerably more than the RMS value of the input voltage - theoretically 1.414 times the input, but in practice under load closer to 1.3.  This means that if the incoming line is a bit high - say 120 instead of the nominal 117, the voltage at your capacitor will be somewhere between 155 and 170 volts - still OK for this application, but unwise to forget about.
~~
Mark Moulding

gregebert

unread,
Mar 22, 2018, 3:13:23 PM3/22/18
to neonixie-l
Or, you can just deal with the AC mains directly. I have 6 clocks of 3 different designs that run hot on the AC line to varying degrees, but in all cases the nixie anode supply is not isolated. Most have been running for 6 years now, with no issues. All told, there's almost 30 years of runtime (250,000 hours) with zero circuit issues.

If you are going to go this route, it is absolutely essential that you provide redundant fuses, have good surge-protection, proper PCB trace-width and separation, and noise-filtering. And yes, be sure to use an isolation transformer during testing/debug. Being extra careful will go a long way; it's been more than 30 years since I've been shocked, had something smoke/overheat, or explode.

There's nothing inherently "bad" about the AC mains; the real issue is that they are typically protected by a 20amp circuit breaker, so it takes a lot of power to trip the breaker. When all of that power is running thru your gizmo, it will cause a lot of damage. In other words, NEVER rely on the circuit breaker in your house to provide safety for your project.

For starters, use a UL-listed (or equivalent) power-entry module. They have a socket for an IEC power cord, a fuse, and a power switch, all in one nice compact unit. So no matter what happens with your gizmo, a properly-sized fuse in the power entry module will blow before damage occurs. I suggest the smallest slow-blow fuse that is rated for your typical operating current. That way, any power-on transients to charge power-supply filter caps will not blow your fuse. If you are extra diligent, run a simulation of the peak AC line current during power-on, and use the fuse-manufacturers datasheet curve to see if it will withstand the transient. If you're lucky, a fast-blow fuse might even work.

Next, when your AC power comes onto your PC board, have *both* legs (hot and neutral) run thru their own fuses; PCB-mounted fuses are inexpensive. Now you technically have triple-redundant fusing; if any of the 3 fuses (power-entry, hot-lead, neutral lead) blows, you will not have a complete circuit for the AC line. After the fuses, add a varistor for surge protection; a nasty surge will blow a fuse though the rest of your gizmo should survive. Then place a 0.01uF 1kV cap across the varistor to absorb fast transients.

If you run SPICE simulations on your gizmo, you can compute the RMS current, and use that to select your fuses. Otherwise, use a 'Kill-a-watt' device to measure the actual current consumption.


--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
In case you wondered, it's pretty easy to find out what time it is in my house....total of 61 nixies are on-the-job, and I've only had 2 fail (both were Burroughs 6091)

Paul Parry

unread,
Mar 23, 2018, 4:01:54 AM3/23/18
to neonixie-l
Hi Mikołaj

I've recently made an 18 tube visualiser but used the IN-13 tubes.. they are slightly taller and they also have an extra 3rd wire which creates little dot at the bottom of the tube - similar to a decimal point in a regular Nixie, but it seems to effectively anchor the glow to the bottom of the tube. I've had on a couple of occasions the glow float up to the middle of the tube, but as soon as it then touches the bottom it attaches again.




John Murphy

unread,
Mar 23, 2018, 8:07:07 AM3/23/18
to neonixie-l
For a clock this might be an acceptable solution, but for a music visualizer, it depends on the source of the audio.  If there is a headphone jack or the like for audio input, a grounded source will fry the rectifier circuit!

Robert L

unread,
Mar 24, 2018, 12:56:06 AM3/24/18
to neonixie-l
Sorry Greg,

<Hot button warning...I need to put this out there before I go to sleep tonight...>

I have to speak up... As you are well aware, there are real hazards working with the AC mains... lethal hazards... hazards only to be approached with proper training, experience and caution. The hazards go way beyond fried electronics up to and including the risk of nasty and potentially lethal shock. Mains can and do kill... 

I know you know this, that you respect the hazards, and that you trust your level of experience and training. I respect your designs... I respect your skills! I respect your training!

In contrast, this thread was started by a student with what seems to be little practical experience and likely no training on working with mains or other high energy circuits. My strong belief / conviction is that they have no  place working with a non-isolated mains design prior to acquiring appropriate training. I personally will not do a non-isolated mains powered device. The risk/reward equation doesn't begin to work for me.

What follows is only intended to give an idea of what's involved... The list is not complete and may have inaccuracies, but it gives an idea of some of *my* concerns! Old maps used to say "Here there be dragons" in areas where little was known... for me, that saying applies to mains connected  non-isolated designs. Go there if you must, but go with caution and the knowledge to properly understand the risks and how to mitigate them.

Many of the hazards can be avoided or significantly mitigated by using an appropriately safety-certified isolated power source. Given a choice, I choose to let someone else do the mains connected part of a design and have that design tested and certified for compliance on their dime... and fully documented as part of what I check both on my hobby projects as well as when I compile a documentation package that will accompany a device sent for compliance testing.

There are safety standards for compliance when connected to higher voltages (not safety extra low voltage "SELV" which covers most of what I do - in part SELV will be under 60 V and ISOLATED by current standards...)... you mention some.  Here are some of those standards in no particular order and absolutely not a complete list: we have multiple means of patient and operator protection (MOOP, MOPP: 60601-1 3rd... if you happen to be in medical... other standards apply for other devices...) - multiple failures have to happen for a person to connect to the mains. There are component ratings for various applications of mains connected devices ("X" and "Y" capacitors as an example... these are not just 1 kV rated caps...), creepage and clearance, leakage limits... The list goes on for quite a bit. Then what happens when the device is knocked off a shelf and someone goes to pick up the pieces? What happens when the babysitter plugs her headphones into the jack? There are reasons why some odd and unusual connectors show up on equipment - they are highly unlikely to be accidentally connected to an improper device. 

There are loads of accessible articles  out there that give a starting place on mains electrical safety... this two part article was the first hit off a Google search (search "working with mains voltage safety") and seems like a nice easy intro...

Low voltage also deserves respect... As an example, I remove my ring and watch before reaching into a chassis or circuit regardless if there's hazardous energy inside. This is a habit I choose to always enforce so that it's there when it matters.... Low voltages may not be a hazard for electrocution, but a short across a ring can result in a nasty burn if there's sufficient energy available. There was the time I was reaching into an old Heathkit tube device with both hands... violating multiple safety practices. I'd acquired new habits over a bunch of years working with 15 VDC and below very low energy circuits... My pulse jumped... a lot... when I noticed what I was doing. I decided then that I would have only one set of habits... 

Anyway... had to say something... NOW I can sleep!

I'd welcome others to chime in... Please call out or fix anything I have wrong, etc... This is a worthy topic for discussion!

As always, best regards,
Bob





Tomasz Kowalczyk

unread,
Mar 24, 2018, 6:28:20 AM3/24/18
to neonixie-l
The power supply for such circuit will be tricky, as the load will vary a lot during operation. I think that a simple boost converter won't be able to achieve enough power and due to high duty cycle, stability will be a problem. A flyback design should do the trick, however making your own will take time to achieve results. I think that a transformer looted from ATX power supply might be useful, however I didn't test one yet - in my project I'm leaving the PSU till I'm sure my digital processing is done correctly.

Using mains voltage directly is not a viable method in music visualizers! Keep in mind that unlike clocks, this device will be connected to an external source of audio signal. If you use mains voltage directly, then you have to make the input isolated. There are two reasons:
1. Your input source will likely have ground connected to earth 
2. If you plug a minijack into the visualizer and then grab the other side with your hand, you're touching mains directly, and there is a good chance that it won't be the neutral wire


W dniu czwartek, 22 marca 2018 13:23:24 UTC+1 użytkownik John Murphy napisał:
I am currently working on a very similar project.  Here is what I've learned:

1. The IN-9 tubes don't all act properly using a 150V DC switching power supply (I bought several different models from eBay).  The glow of the tube becomes disconnected from the bottom and "floats" in the middle - this is not how a bargraph should behave.  I have been successful using a full wave rectifier circuit from 120V mains through a 1:1 isolation power supply (US).  If you have 240V mains, you would need a 2:1 step-down transformer.  I used the KBP307 rectifier instead of discrete diodes.  Do NOT try to use the rectified mains without a transformer.  In addition to this being dangerous, if the DC 0V has a path to earth ground, the rectifier will be destroyed, and also likely take out other components like the uC.  I am using a Triad VPS230-110 as it's fairly inexpensive.  It has a maximum output current of 220 mA, but for a music visualizer, you won't have full scale output on all the tubes at the same time - you could enforce this using code in the microcontroller if you're worried about it.

2. Most microcontrollers have a limited number of pins that you can output PWM on.  If you want to drive 16 tubes, you need 16 PWM (or analog out) pins.  I am using two serial octal DACs (LTC1665) to drive the base of the MJE340 transistors using the circuit in the paper you referenced.  DAC values are set with a synchronous serial stream. The clock and data lines are common to both DACs and each one gets its own chip select for a total of 4 digital pins needed to drive 16 tubes.

3. If you buy New Old Stock (NOS) tubes, some of them won't initially glow to full height due to being in storage for so long.  They need to be "burned in" - something about the physics of the cathode tube that I don't really understand.  I have found that driving them at 15mA for about an hour has been enough.  Over that time the glow gradually creeps from about 1/2 scale to full scale.  After that, they all have worked to full scale.

Have fun!

Are you sure the disconnection doesn't happen because of poor filtering or inadequate driving? A proper driver should be a constant current sink, which would mean that the tube should be able to see supply without ripple (the ripple should be present on cathode as well, making the voltage across the tube constant). Maybe the supply is emitting too much noise into other circuits and your driver misbehaves?


BTW. STM32 microcontrollers are a good choice, as F4 and higher series have many hardware PWM channels avaible. I'm using STM32F429ZI and as far as I remember, I can use 29 or 31 PWM channels, and there are some with up to 33 channels available. I'm planning to try the new H7 series, which have 30+ channels with PWM and there are still some timers left for other applications. 
Also, the FPU in Cortex-M4 really speeds up the calculations.

Phill S

unread,
Mar 30, 2018, 12:39:21 AM3/30/18
to neonixie-l
Bit late to this post.

I have had success with this ebay HV supply for IN9's.
They are kind of touchy though.

Blown up 3 just being a little bit careless with loading etc.


For driving lots of them you will defiantly need to go down the transformer route OR... lots of little supplies with a really fat laptop  or LED power supply.


Reply all
Reply to author
Forward
0 new messages