Nixie supply safety and noise reduction questions

[Max999]

Hi everybody,

Dreaming about building some tube pedals I found the possibility to use a nixie supply instead of a transformer. What got me worried though is the hight voltage and no earth connection to chassis. What I want to do is provide safety for the user.

What would happen if the high voltage wire got loose and touched the signal grounded chassis? Would something inside the psu or wall wart just blow and act like a fuse?

I am not worried about safety when building it, I can visually see what is going on then and keep one hand behind my back etc.

Another question is about the switching frequency. A NE555 schould do about 31-45KhZ, which is above human hearing range, but it could go( if it somehow got past guitar speaker roll-off) to the tweeters of a PA system I think.

Can I smooth this out like regular ripple with RC filters? Are there any RC filter tricks that let me filter out such a high ripple frequency easier or should a 3.3K-47uf filter be fine at it?

The type of nixie supply I would like to use is this, found on this forum:

[anotherjim]

I wouldn't worry. HV it may be, but the energy it can put out is puny. A couple of triodes only need a few mA.

That said, as you add the B+ supply filters, you are increasing the potential energy by storing charge in the capacitor. So it is actually unwise to fit a very large filter cap. The 20 to 50uF sizes found in old tube amps are still adequate. Also, electro caps are less efficient at high supply frequency, so a more suitable type of capacitor can be used to cut HF supply noise. A 100nF X2 rated cap for example.

[thomasha]

I build some amps with SMPS and choosing the right layout is as important as filtering.

There are some noises that are picked up that are in the guitar range, sometimes they are caused by changes in the switching behaviour, like when the chips reduce or increase the switching freq. to keep the output voltage constant or skips some cicles. This occurs at lower freq. than the switching.

A different inductors geometry can change the emission region and make things better or worse. The inductor needs to have a low resistance, otherwise it will heat up. A fet with a low Rds_on is also desirable. Low ESR fast switching capacitors are a nice touch. A film capacitor in parallel with the reservoir cap. helps with the high frequency noises. There are also some huge 2.2uF film capacitors.

The 555 is not the best chip for this task, I like to use the MAX1771, but it is expensive. There are other chips that do the same job, like the MC34063. There was also a blackstar schematic somewhere where they use another chip.

[rankot]

I've been doing a lots of research on this, and finally have a design I believe will be really quiet. But I don't have time to test it right now, maybe in few weeks. The idea is to use additional inductor for filtering the output, and that one can be low current rated, because we don't need much current there.

I use TLC555 at 100kHz, because I tried with MC34036 and it has always been too noisy!

[GibsonGM]

I dunno....I've made the standard one with the 555 that has always gone around...most likely the one the OP posted is the same thing.  Mine is quiet as h3ll, with just the normal filtering you'd include in a tube preamp.  The big ol' GTFO thread from a while back had all the files.

Quiet enough for direct recording...  Hey Max, if you have a ground path to your chassis, and the " - " of the wallwart you use to power this thing is also connected, then you do have a ground.   Granted not a true 'earth ground', but it will serve to short the B+ if it ever made a hot chassis.   As stated, not a lot of current-delivering capability there anyway. 

[Max999]

I am finding conflicting answers online about how much current vs voltage is deadly.
I like your trick GibsonGM, I think for a smps this is the answer. The noise inolved with a smps is something I have no experience with, so am am also considering back to back transformers and a regular trasformer.

With back to back transformers ( using a 12v ac wallwart plugged in the wall), there is no way to put the ground on the chassis right?

[GibsonGM]

Yes there is a way to ground transformers, Max, tho I AM a big fan of the SMPS (Nixie) supply, so hold on a minute.   If you power the Nixie from a wall wart, such as a 15VDC or something...you achieve isolation from the mains...that is a huge plus right there.   You get your high B+, and if the wallwart gets ruined, you can replace it cheaply.  Note I said "DC" not "AC", for which you'd need to do some rectifying, and it is far simpler to stick with a DC power input!

You use a jack just like for stompbox power to bring that into your enclosure....all points that are negative/ground connect to it.  Most of us would build on a turret board, and use a piece of 14ga. house ground wire for a ground bus...that connects to the " - " as well.   Some point on your chassis connects to " -".  If the hot wire somehow shorts to ground, it will probably either toast the SMPS board or the wall wart, but won't give you a hot chassis.   It provides a return path for 'stray  power', which provides a fair level of safety.

If you did the 2 transformers deal, you'd bring your power into the enclosure (preferably with an IEC connector).  The earth wire would connect to the enclosure by a dedicated bolt with star washer and solder lug.   The other 2 wires go to first trafo....then your circuit ground connects to the chassis.   You still would need to rectify the AC into DC, and deal with ripple.  And if there is an error on the 1st transformer side, the stakes go up since that is line power.

>>>1mA at 30V COULD kill; that is why they say "it's volts that jolts, it's mils that kills".   And anything over 30V could be considered high voltage.  Your skin resistance is pretty high tho (unless wet...), which makes the SMPS a pretty safe solution.  Yes, it can put out 240V, but the ability to deliver current is very low.  Don't go TRYING to touch it, now!  That would be stupid.  It is just...LESS lethal, if that makes sense.   There is no way around that if you want to use tubes at proper voltages such as 12AX7's.  Remember, touching a doorknob and getting a shock in winter - that is like 30,000 volts at NO amps.  I like the SMPS because once the board is installed, there is ONE hot wire coming out.  Watch it, and don't go near the board....all set.   REMEMBER that all these power supplies require the use of proper electrical safety rules - if you don't know what they are, please ASK!!   

Here is 1 pic of a tube preamp in a box I've done, you can see the jack at upper left, and green ground running to the ground bus up front that you can't see:



This one shows one I was messing with. You can see the bus wire.  I don't recommend you hang the tubes in your lap this way

[amptramp]

I am not a big fan of using the 555 as a power supply chip.  The current spike as the output goes from high to low or low to high can be on the order of several hundred milliamps.  I don't know what is available now but Unitrode made the 3824 and 3826 inverter IC's and they were designed for this purpose.  They used to be common but I have not seen them advertised for a while.  For the level of power needed for a preamp, a flyback topology would be the most commonly used and they can easily go above 40 KHz for simple filtering.

I like back-to-back 60 Hz transformers (or 50 Hz if that is what you have) because they are easy to use and work well.  I have an IBM module that provides regulated 150 volts from a 6.3 VAC filament supply from the age of tube computers.

[GibsonGM]

I think the issue may be the scarcity of schematics for such "exotic" switching supplies, Ron.  The ol' 555-based Nixie has been around for a few decades.  Easy to source all the parts you need, it works pretty well...the negatives *usually* don't bite your rear, so are easy to let go.   I have a tube preamp I use for regular recording work, based on that thing - no noise, whatsoever.   No negative effects...

I have yet to see an easily-achievable schematic like that using a common, newer and better $1 chip, with an easy to find inductor.  If you can post one, maybe we can move past the 555-based stuff!   Maybe it's not readily available since so few people need 200+ volts today?

I WISH I had some of those old tube computer transformers!   Things from that era would be useful today.

[Rob Strand]

I think the issue may be the scarcity of schematics for such "exotic" switching supplies

There's too many chips and too many different requirements (input voltage, output voltage, current, frequency, boost, buck ...).

The next layer is how integrated you want.  Do you want the switching MOSFET (or whatever) and diode on chip?   Some devices now are putting the inductors in the module or on the chip.

Some very common SMPS controllers are TL494 and UC3842/KA3842.   But there's a whole family UC3842 to 3845.  Then there's other families SG3524 to 3526 (maybe 2524 to 2526? as well).  These are controllers only. You need to add MOSFET + diodes.

These were used in a heap of off-the-shelf switch-mode supplies in the past.  Later on chips that were powered directly off the mains came out, then after that came chips which were powered off the mains and had the MOSFET built in.   You can see the evolution in off-the-shelf switch mode power supplies, PC power supplies, and power supplies used in consumer equipment like CRT monitors, LCD monitors and TVs.

The other side of the more modern market is all the stuff you find in your laptop and PC motherboards.  Small, high current, high efficiency, large amounts of integration.

[rankot]

Rob, if we use CMOS 555 and at big enough frequency (100kHz, for example), is it still prone to current spikes or not? I believe I've read somewhere that increasing freq. and R1/R2 resistor values reduces spikes. So I moved to 910k for R1 and 47k for R2 with 15p for timing cap, hoping it will reduce spikes. I've tested my SMPS with those values and it works fine, steady 250V with 47k load.

[printer2]

>>>1mA at 30V COULD kill; that is why they say "it's volts that jolts, it's mils that kills".   And anything over 30V could be considered high voltage.  Your skin resistance is pretty high tho (unless wet...), which makes the SMPS a pretty safe solution. 

You would be hard pressed to feel 30V never mind it conduct 1 mA through you. And as far as SMPS being safe, I have a common module sold online that can put out 400V and 50 mA that I would be hesitant to go touching around the high voltage side thinking it was pretty safe.

[GibsonGM]

>>>1mA at 30V COULD kill; that is why they say "it's volts that jolts, it's mils that kills".   And anything over 30V could be considered high voltage.  Your skin resistance is pretty high tho (unless wet...), which makes the SMPS a pretty safe solution. 

You would be hard pressed to feel 30V never mind it conduct 1 mA through you. And as far as SMPS being safe, I have a common module sold online that can put out 400V and 50 mA that I would be hesitant to go touching around the high voltage side thinking it was pretty safe.

Fred: I never suggested anyone touch anything that is energized (I said as much, above), and a person should also be aware of basic high voltage safety before attempting power supply projects...the possibility that filter caps may contain voltages after powering down, for example.  You should not make contact with a SMPS output, nor that of a transformer.   The thought of it being "less dangerous" is the fact that the one we're discussing would be current-limited - later, maybe I will load it and see what the 555-based one can deliver...back to back transformers could certainly deliver far more, and introduce mains voltage to the project, tho I don't have comparative specs on how much current the output of the 2 can deliver.   Both deserve respect and require the use of common electrical safety rules, so we should probably continue discussing that right off in this kind of post.  I apologize for not bringing it up right up top.   

As for 30 volts:
"It should be clearly understood that resistance of body will vary.  That is, if the skin is dry and unbroken, body resistance will be quite high, on the order of 300,000 to 500,000 ohms.  However, if the skin becomes moist or broken, body resistance may drop to as low as 300 ohms.  Thus, a potential as low as 30 volts could cause a fatal current to flow.  Therefore, any circuit with a potential in excess of this value must be considered dangerous."   Basic Electronics Volume I, United States Navy, 1973
They go on to discuss that a 1mA 60Hz AC current will be felt hand to hand, with paralysis occurring at about 10mA...That's what I was taught, and I'll stick to it!

You are absolutely correct -the output of SMPS can be 100s of volts...consider it dangerous.   Always use common sense and obtain a working knowledge of electrical safety when working with power supplies.      

[printer2]

And I am saying in most cases electricity is not as dangerous as to scare people. If I were to be dripping blood I probably would not be sticking my hands in a 30V circuit. Unbroken skin would result in no harm. Have I received a good poke or two, yes, mainly from 120V from inadvertent contact. Forty-five years of learning/working with it stuff happens. Not to say I have a caviler attitude towards electricity, every electrician I have worked with can tell of a time they got shocked or shorted out a circuit with a screwdriver or cutters.

48V is considered safe enough for the layman without an electrical background to play with. You can certainly feel 48V and remember the episode enough to keep your wits about you when you are working live. But I will not be breaking out the rubber gloves for it unless I am dripping sweat. I think the Navy instruction is intended or people who work in environments that may include high heat (sweat), salt water, maybe even lacerations with open wounds. As most times the context is important. On a ship with intense radar a person can have two amps running across their body. Thankfully because of the skin effect (not human skin but a conductors) the high frequency power just travels on the exterior of the body. Sill a danger and impediment to the operation on a carrier (basically like being cooked in a microwave oven) and wearing clothing to shunt the current around the body is a must (which causes further hazards). But none of this applies in the confines of our homes. No disagreement on 1mA or 10mA with you. How much voltage you need to get this current for the length of time needed to be dangerous, maybe a little.

As an aside, those electronic fly swatters can produce a voltage and current high enough to be lethal.

[GibsonGM]

Absolutely, Fred.  I work on tube amps with the chassis open - it's a risk.  Or, making a connection to a breaker in the box - the mains is always live, and that crap will fry you all the way.  Good to make it an educated risk, if we can.  Little things, like grabbing a jumper to connect the meter lead back up, and you forgot the thing is on - do happen!  I haven't done that, fortunately, but someone must have.   Yeah, I've touched 48V too...the navy books are probably conservative to scare the new guys into being overly cautious.   Skin effect is an interesting phenomena - I wasn't aware that humans could also exhibit it, ha ha!

So Max, pay attention, ok bud?   Don't get scared off, but also learn what you need to, to keep yourself isolated from the project you're messing around with.   

I love those fly swatters!

[Rob Strand]

Rob, if we use CMOS 555 and at big enough frequency (100kHz, for example), is it still prone to current spikes or not? I believe I've read somewhere that increasing freq. and R1/R2 resistor values reduces spikes. So I moved to 910k for R1 and 47k for R2 with 15p for timing cap, hoping it will reduce spikes. I've tested my SMPS with those values and it works fine, steady 250V with 47k load.

For some circuits crazy low values (<1k) for R1 could cause issues in some circuits, however, R1's of 10k or so wold be fine and any higher wouldn't achieve anything.  The spikes people talk about aren't related to R1 they are cause by the output stage  (they are very narrow 50nS to 100nS spikes).  I'm not aware of R1 changing that.   You normally put a 100nF cap across the device to absorb the spikes. 

I try to use resistors 10k to 100k (1M ok for CMOS) and caps > 1n for 555's.

There's actually about 30pF of internal capacitance on the 555.  This upsets the timing calculations and can introduce timing errors.  I've seen funny behaviour on monostables when the external capacitance is below about 22pF.  I suspect the 30pF will upset the thermal stability of the oscillator although many external caps aren't that great either.

[Max999]

I do work on tube amps, and yes I must admit that I have slipped a tool once and got a shock from the high voltage supply.
It was unpleasant but less bad than I thought beforehand. The kind of shoes you wear also help a lot in reducing the shock.
Please do not take this as an advise to be uncareful with high voltage in tube amps. This is just a personal anecdote and your results may vary!

So what if I use the back to back approach but have the first transformer be an 12v AC output wall wart.
This means that there is no earth connection to the enclosure.

Can I just use the minus of the high voltage after rectification as a safety ground for the enclosure?

I made a quick drawing of this because a picture is worth a thousand words right?

[Slowpoke101]

Have a read of this thread regarding the fun and games of running transformers back to back.

https://www.diystompboxes.com/smfforum/index.php?topic=121216.0

[rankot]

For some circuits crazy low values (<1k) for R1 could cause issues in some circuits, however, R1's of 10k or so wold be fine and any higher wouldn't achieve anything.  The spikes people talk about aren't related to R1 they are cause by the output stage  (they are very narrow 50nS to 100nS spikes).  I'm not aware of R1 changing that.   You normally put a 100nF cap across the device to absorb the spikes. 

I'll try with those values, too, although I'm nailing my working frequency to ~100kHz, because it is optimal for 22uH inductors I have at hand. However, when I measure output with my pocket oscope, I see there are some very small spikes in output. It can't go below 20mV and 10us, so those spikes are actually very small: 12mV at 37kHz. Probably quite weak to be heard even if they're lower frequency.

[Rob Strand]

see there are some very small spikes in output. It can't go below 20mV and 10us, so those spikes are actually very small: 12mV at 37kHz. Probably quite weak to be heard even if they're lower frequency.

On the output of the power supply or the audio?

Any glitches on the power rail can be removed with a second filter (either R+C or L+C).  The output cap on a switch-mode has an ESR (equivalent series resistance) so there is a limit to its filtering ability.  The MOSFET and diode can cause glitches the 2.2n + 100pF cap across the MOSFET helps.  I don't know if anyone has tried to fine tune those values.

Another source of glitches is due to the switching currents and spikes on the ground.

If you have this setup, you might get problems because the tube 0V line has junk on it from the switch-mode.

0V in ---+--- 0V smps ---+--- 0V tubes
              |
             0V to other audio circuits

This configuration might help but now the junk is transferred to the +HV rail.   So you add a filter on the +HV rail and connect the ground of the second filter to the *tube 0V*.  You can't connect the smps output to the 0V of the tubes as the 0V smps is noisy.

0V in ---+--- 0V smps
              |
             0V to other audio circuits  + 0V to tubes

In all cases keep the "0V in to 0V smps" ground line short and thick.
It's like chess maneuvers with the ground and power trying to prevent the glitches getting into the audio.