The case of the singing nixie tube

[Paul Andrews]

I have been making a set of one-tube clocks using hardware of my own design, which I use to power multiple tubes from my collection - I have adapters I can use to swap different tubes in and out.

Recently I built a clock for a CD47 tube and discovered something quite interesting. My clocks include digit fading and dimming, and I noticed that the CD47 would sing quietly whenever it was dimmed or whenever it faded from one digit to another. At full brightness it was silent. Naturally I use PWM to implement dimming and fading by turning a digit off and on. Change the duty cycle, and you change the perceived brightness. I figured that it was this that was causing the CD47 to sing, so I set about experimenting. First I verified that the HV was behaving itself - the clock for this tube was using different components than my previous clocks. I actually had to modify some of the components in the HV stage to get rid of some nasty ripple, but this made no difference to the singing. Second, I modified the software so that I could control the frequency of the PWM using the GUI, and it was quite easy to show that the PWM was causing the singing. As I increased the PWM frequency, the frequency of the singing would go up too, with some frequencies being louder than others.

None of my other tubes exhibit this behavior (with a caveat I will get into in a moment). My theory on this was that the large cathodes on the CD47 could vibrate at lower frequencies than any other tube in my collection, which might explain why the default PWM frequency I was using would have no effect on any of my other tubes. The next largest tube I have is a CD27, and there is a significant difference between the size of the cathodes in these two tubes - 135mm for the CD47 and 55mm for the CD27. One experiment that this suggests is to lower the PWM frequency on the CD47 until it stops singing. Unfortunately, even if I lower it to the point that I can see visible flicker, it still sings. Another experiment would be to increase the PWM frequency on one of my other tubes. I did this for an IN-18, and was indeed able to make this tube sing too!

Of course, PWM 'frequency' is an interesting term. If the base frequency is 10KHz and the PWM is quantized in steps of 0-100 (so you can generate 100 different duty cycles), the actual frequency of the resulting square wave is 100Hz. I am using the term 'frequency' here to describe how many times/sec the waveform will go through one oscillation. This is actually the default frequency of my PWM.

Apart from just being interesting, this is all of some concern as I assume that this vibration of the cathodes, when they are dimmed or fading, will cause some mechanical stress, which I obviously want to avoid on a tube like the CD47. I am wondering if vibration could be reduced, or even eliminated, by smoothing out the square wave - i.e. by gradually allowing more current to flow as a cathode is pulled to ground and vice-versa as the cathode is 'released'. I would be grateful for any ideas on how to do this. I am using an HV5523 to control the cathodes.

Here is a picture of the CD47 one-tube-clock:

Here is a link to a video of it in operation: https://youtu.be/24Oo7i2ItHM

 

[jf...@my-deja.com]

Does it also produce RFI on the AM band?  My 6-digit B-7971 clock sings and produces RFI at the MUX rate

[Instrument Resources of America]

The 'singing Nixie' in and of itself won't produce RFI on the A.M. band. The RFI will come from the electronics, including the elements of the Nixie tube radiating RF from the signals being applied to them, and would probably be plentiful especially in close proximity to the radio, unless very well shielded and power line filtered. The faster the rise time of the driving signals will be a factor in the RFI.   Irv

On 2/23/2019 8:04 AM, 'jf...@my-deja.com' via neonixie-l wrote:

Does it also produce RFI on the AM band?  My 6-digit B-7971 clock sings and produces RFI at the MUX rate

--
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/fff56e15-953e-4a98-a98a-f3537628567e%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[jb-electronics]

Nice clock! I am not an expert on this by any means, but sometimes the vibrations come from the inductor itself (the coils are moving and creating the sound, much like in a speaker). Are you positive the sound emanates from the glass bulb? This is quite fascinating. Maybe you can build a Nixie-speaker! ;-) Jens

--
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/cc2a01fb-b772-465a-a1f0-0bb7a3ba815c%40googlegroups.com.

[Paul Andrews]

I’m pretty sure it is the tube. If I put my ear on it, it is louder.

 

I was just wondering about the speaker idea myself. Maybe if I can get an IN-12 to sing, I’ll try playing a simple tune 😊.

 

On a more serious note (no pun intended), I guess I don’t have to try and limit the current change on each cathode, I could just do it on the anode. No idea how to achieve this though.

--
You received this message because you are subscribed to a topic in the Google Groups "neonixie-l" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/neonixie-l/h8BMlkUhkZw/unsubscribe.
To unsubscribe from this group and all its topics, 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/bfde00fd-8a93-2e2d-c54c-72d12baa181d%40jb-electronics.de.

[Terry Kennedy]

On Saturday, February 23, 2019 at 7:22:59 AM UTC-5, Paul Andrews wrote:

None of my other tubes exhibit this behavior (with a caveat I will get into in a moment). My theory on this was that the large cathodes on the CD47 could vibrate at lower frequencies than any other tube in my collection, which might explain why the default PWM frequency I was using would have no effect on any of my other tubes. The next largest tube I have is a CD27, and there is a significant difference between the size of the cathodes in these two tubes - 135mm for the CD47 and 55mm for the CD27. One experiment that this suggests is to lower the PWM frequency on the CD47 until it stops singing. Unfortunately, even if I lower it to the point that I can see visible flicker, it still sings. Another experiment would be to increase the PWM frequency on one of my other tubes. I did this for an IN-18, and was indeed able to make this tube sing too!

Apart from just being interesting, this is all of some concern as I assume that this vibration of the cathodes, when they are dimmed or fading, will cause some mechanical stress, which I obviously want to avoid on a tube like the CD47. I am wondering if vibration could be reduced, or even eliminated, by smoothing out the square wave - i.e. by gradually allowing more current to flow as a cathode is pulled to ground and vice-versa as the cathode is 'released'. I would be grateful for any ideas on how to do this. I am using an HV5523 to control the cathodes.

The B7971's in the MOD-SIX tended to sing as well and are also large tubes, but not as big as the CD47. Interestingly, changing from the original power supply to the potted supply used in the newer versions of the MOD-SIX stops those tubes from singing as well (simple swap of the PSU board, same tubes / CPU board / setup options). I'm not sure why changing the power supply made a difference. My guess is that the potted supply has a large-value integral cap in there, and that the PWM cycling was causing the original power supply to sag in time with the PWM duty cycle.

[Paul Andrews]

The PWM definitely causes the power supply to sag in my clock too. The output voltage also briefly sags, but quickly pulls itself back. As I mentioned, removing, or lessening the ripple has no appreciable effect.

--

You received this message because you are subscribed to a topic in the Google Groups "neonixie-l" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/neonixie-l/h8BMlkUhkZw/unsubscribe.
To unsubscribe from this group and all its topics, 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/04e146cc-0adc-44b7-b6b6-97a85cd401ad%40googlegroups.com.

[Dekatron42]

Can you put a microphone up to the tube and look at the sound on a scope while you also scope the anode voltage via an adjustable low pass filter, or perhaps an adjustable a band pass filter, if so you might see where the frequency comes from?

/Martin

[Paul Andrews]

I used an app called SpectrumView which gives a nice plot of freq against db. It shows a mess of noise below 4Khz, I think I would need a really quiet room to get rid of that.

With no dimming/fading there are noticeable peaks at 4KHz, 8KHz and 12KHz, 12KHz is the upper limit of the frequency scale I was using). With dimming there are many more peaks between 4KHz and 12KHz, which I'm guessing are harmonics from the PWM square wave. This just reinforces my desire to smooth out the ramp up and down of that signal.

I have scoped the anode voltage. Unloaded ripple is negligible (better be, right?), but there is an initial momentary drop (on the order of .01ms if memory serves me right) of around 10V when the PWM turns the tube on. The voltage doesn't quite recover to the unloaded level and there is maybe a 1-2 volt ripple when it is loaded, the period of which is also on the order of .01ms

I have literally listened to the tube while I moved the PWM frequency around, and the tone moves in lock-step with that frequency. Seeing it on the app would be nice though, so I can double-check that the frequency matches.

BTW, this is all quite quiet - you basically have to be within around a foot of the tube to hear it at all. I will post screen grabs later from the spectrum analyzer app, so you can see the peaks and the sound level.

Also worth mentioning that these tubes make noise if you just tap them. Somehow they manage to survive transportation though (I'm generalizing from the one tube I have here!).

On Feb 23, 2019, at 4:42 PM, Dekatron42 <martin....@gmail.com> wrote:

Can you put a microphone up to the tube and look at the sound on a scope while you also scope the anode voltage via an adjustable low pass filter, or perhaps an adjustable a band pass filter, if so you might see where the frequency comes from?

/Martin

--

You received this message because you are subscribed to a topic in the Google Groups "neonixie-l" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/neonixie-l/h8BMlkUhkZw/unsubscribe.
To unsubscribe from this group and all its topics, 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/ee25c96c-b1f9-482d-a6aa-be67c0ec90da%40googlegroups.com.

[Mac Doktor]

[This is my first post here. I haven't had time to introduce myself yet. My apologies if I'm just saying something everyone already knows.]

> On Feb 24, 2019, at 9:33 AM, Paul Andrews <judg...@gmail.com> wrote:
>
> With no dimming/fading there are noticeable peaks at 4KHz, 8KHz and 12KHz, 12KHz is the upper limit of the frequency scale I was using).

Try sweeping the frequency instead of the duty cycle to see how the volume (and spectrum) of the sound changes.

As already suggested, hooking a microphone or piezo transducer up to a spectrum analyzer or even just an oscilloscope would be enlightening.


[Proper introduction to come...]


Terry Bowman, KA4HJH
"The Mac Doctor"

[Paul Andrews]

Hi

That is just what I have done. I mentioned that I use an app called SpectrumView and have continuously adjusted the PWM frequency, not the duty cycle.

> --
> You received this message because you are subscribed to a topic in the Google Groups "neonixie-l" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/topic/neonixie-l/h8BMlkUhkZw/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to neonixie-l+...@googlegroups.com.

> To post to this group, send an email to neoni...@googlegroups.com.
> To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/88AA2D17-37C4-4450-8F63-EC5C0F0E7E13%40gmail.com.

[Mac Doktor]

> On Feb 24, 2019, at 4:24 PM, Paul Andrews <judg...@gmail.com> wrote:
>
> Hi
>
> That is just what I have done. I mentioned that I use an app called SpectrumView and have continuously adjusted the PWM frequency, not the duty cycle.

Oops, my mistake. A case of Great Minds, I guess. 8D

[Robert L]

My ears leave something to be desired these days... Singing? What singing! OK. My wife and animals hear it. It's real.

I use a small microphone and PC based spectrum analyzer to explore sounds I can't hear. The mic is at the end of a bit of heat shrink tubing close to a preamplifier on a longer cable back to a USB sound port on the PC.

The heat shrink tubing allows me to use the mic as a probe to localize sources and manually probe near dangerous voltages without risk of a nasty bite. 

Photo of the probe and links to the various components follow... Note that the Panasonic mic linked below replaces the larger mic that comes with the amplifier.

 

PC based spectrum analyzer to “see” noise. Does the job nicely and takes up no extra desk space.

http://www.fatpigdog.com/SpectrumAnalyzer/index.html

 

Mic preamp:

https://www.digikey.com/product-detail/en/adafruit-industries-llc/1063/1528-1013-ND/4990762&?gclid=EAIaIQobChMI2_a6kfqu3wIVi7jACh3OowKOEAQYASABEgJjhPD_BwE

 

Panasonic WM-61A… this mic is near the front of the black rubber tube coming off the preamp board. Rubber and heat shrink tubing prevent contact with HV around tubes. THis mic replaces the mic that comes with the preamp above.

https://www.ebay.com/itm/Panasonic-WM-61A-Replacement-Microphones/352220785310?hash=item5201feaa9e:g:W2cAAOxybi9ReW5o:sc:USPSPriority!94065!US!-1:rk:34:pf:0

PC sound port… An existing microphone port is likely fine as well. I use these small USB ports because they are relatively quiet and other sound ports are in use.

https://www.amazon.com/gp/product/B00IRVQ0F8/ref=oh_aui_search_detailpage?ie=UTF8&psc=1

As an aside, I resolved B7971 singing on my design by shifting PWM to a higher frequency. Pay attention to a broad spectrum as sub-harmonics eventually became a problem as well. 

Enjoy...

Bob