Dummies guide to building a 120watt 140 volt power supply to drive IN-9 Nixie tubes.

[fireant007]

Yup .. the op amp recommended by you.
But a different transistor ..one that look like they can take a bit of punishment.
I will have to look in the lab and report back on just what I am using.

But yes .. so far so good ... loving this scope .. bit slow but boy .. the things you can do with it are mind boggling for an old fart like me.

edit: ... yup .. opamp is off the 5v line

 

[fireant007]

Hey Spec,
Even though I have a circuit working of sorts.
Your transformer circuit is puzzling me.

How are your windings oriented at the output ?
ie: are pins 7 and 6 both the "dot" or zero side of the secondaries ?
Not quite sure how this works exactly because I am no sure of the secondary phasing ?

I am now thinking that you meant to make it.
Pins 8 and 5 are zero volts. ? ie: the dot on the transformer ?

Making a sort of charge pump ?

 

[spec]

Hy fire,

Re circuit in post #21.

(1) The two 55V RMS secondary windings are connected in series boost. ie non dot to dot. To check that the two windings are correctly connected, connect your multimeter, set to ACV, on the two free ends of the connected secondaries and the reading should be 110V RMS.
(2) As far as the rectifier diodes are concerned, the top of the two secondary windings (pin8) is the 0V reference point.
(3) The upper rectifier diode (D10) and reservoir capacitor (C2) are simply a half wave positive rectifier generating: (55V* 1.414)- 1V =76.77V with respect to pin 8 of the secondary winding.
(4) By the same token, the other rectifier diode (D1) and reservoir capacitor (C8) are a half wave negative rectifier generating -(2 * 55V)- 1V = -154.54 with respect to pin 8 of the secondary winding
(5) This gives a total of, 76.77V + 154.54V = 231.37V across both reservoir capacitors. Call it 231V
(6) Because of the isolation of the transformer secondaries from the mains supply, the 231V supply is completely floating and any point can be made 0V on your tube driving circuit.
(7) Thus the negative line of the 231V is connected to your tube circuit OV. Thus the tube circuit has supply rails of OV and 231V (if not stated otherwise, given voltages are always positive).

The circuit of post #21 show the 231V line as 183V, which is not correct. Note that the 231V line will drop and the ripple voltage will increase from zero as you draw current from the 231V supply.

spec

 

[fireant007]

Thanks Spec,

Above post ignored

I had it correctly wired. In series.
I tried it in parallel and it didn't work.

Data sheet.
http://www.farnell.com/datasheets/1829278.pdf

Is incorrect.... sticker on the transformer says that, the dots are the other way around.
Shouldn't really matter if they have both been flipped.
I wired it to the sticker on the transformer.

Sticker secondary windings are (to your circuit numbers)
Black = 8 (dot)
Red = 7
Orange = 6 (dot)
Yellow = 5

Joined Red and Orange and 110VAC coming out when 240 going in (roughly).

Red and orange are connected to the top diode.
Black connected to the center of the two caps.
Yellow connected to the bottom diode

The thing is ... I get 76vdc on the top cap and 37vdc on the bottom one using my fluke 87 true RMS multimeter.

I know I am not that observant sometimes .. ehhem.... but I am sure that I have wired it to your circuit.

Least it lights and the op amp works.

edit: of course only above 140 when I turn up the vairac

edit again: ... looking here **broken link removed**
it seems that for a full wave bridge the diodes should be the same way round ?
And the centre tap should be zero ?

 

[spec]

Hy fire,

Post 143 now corrected. UPDATE no it wasn't but is now

As you are going with the transformer PSU approach, I have had a closer look at the transformer characteristics and also the circuit design.

As a result, I am half way through a better design and should be able to post today. The 100uF reservoir capacitors are too small and should be 3,300 uF for all 100 tubes but they can be scaled down for less tubes or if the tubes will not all be running at the same time. If the new PSU design turns out OK there will only be one reservoir capacitor instead of two. Basically you need to add 33uF minimum to the reservoir capacitors (or 16.5uF minimum to the reservoir capacitor for the new design) for every 12mA that you intend to draw.

spec

 

[spec]

fire,

I have changed post 143 again and it is now correct- I hope (can't understand my own circuits). I will still work on a new PSU though.

spec

 

[spec]

UPDATE 2016_04_12. This circuit is now toleranced and should work OK with the revised mains input voltages specified in post #202: 230V-5% to 240V+5% and 110V-5% to 115V+5%. (just for the record, this translates to the voltage on the 55V RMS windings being, 57.25V RMS +- 9% total worst case)

Hy fire,

Here is the latest incarnation of the mains transformer cold cathode tube PSU. I managed to simplify the design because the secondary voltages of the transformer are specified under full load conditions rather than the more normal off load conditions. While the tube supply line voltage is a bit on the low side (170V optimum) the latest design does eliminate one expensive and bulky capacitor.

NOTES
(1) The reservoir capacitor (C2) (3m3F= 3300uF) shown is for a full compliment of 100 tubes, all taking 12ma, which will give a ripple voltage of 5V peak to peak maximum worst case. The reservoir capacitor can be scaled down in proportion to the number of tubes actually used, but the minimum value of reservoir capacitor is 100uF. C2 can be made up from any number of individual capacitors in parallel. The capacitors do not have to be the same value or maximum voltage so long as the their voltage rating meets or exceeds the voltage stated on the schematic. The wire connecting the capacitors should be as short as possible and substantial.
(2) C2 ripple current rating should be 71mA RMS minimum.
(3) D1, D8, D9, & D10 form a bridge rectifier. A 10A minimum, 400V to 600V schottky bridge rectifier module can be used instead.
(4) For 115 V RMS supply operation connect the transformer pin 1 to pin 3, and pin 2 to pin 4 (parallel).
(5) RC1 to RC100 are essential to protect the transistors from any problems with the tubes (flash over) and to reduce the transistor dissipation.
(6) The interconnecting wire from the transformer secondarys up to the reservoir capacitor should be substantial (100s of amp peaks will flowing in the rectifier circuit). This will help keep the DC supply line voltage up.
(7) Follow the wiring connections for the tube supply line indicated on the schematic. ie the tube supply lines connect directly to the terminals of the reservoir capacitor. If more than one capacitor is used connect to the highest value capacitor terminals.
(8) The transistors must be in a position to allow a good flow of cool air all around them by convection. The transistor ambient air temperature should be as cool as possible.
(9) Solder the transistors to substantial PCB traces to help conduct heat away. The transistors should be mounted about 3mm proud of the PCB to allow good air circulation.
(10) The collector resistors should also be mounted in as cools as possible ambient air temperature.
(11) The collector resistors should be mounted about 4mm off the PCB to allow a free flow of cool air all around them by convection.
(12) The tubes should also be mounted to allow adequate air flow for convection cooling.

ERRATA
(1) Add one 100nf low loss capacitor across the tube supply line for every 5 tubes. Suitable capacitors are ceramic and polycarbonate metal film.
(2) Higher current, 10A minimum, 400V to 600V PIV schottky rectifiers should be used to keep the supply line voltage up by lowering the diode forward drop.
(3) Change RC1 to RC100 from 3K3 to 5K6.
(4) QV1 to QV100 are transistor types 2N6517.
(6) Supply line voltage should read, '152V MINIMUM PEAK, 175V MAXIMUM PEAK. RIPPLE VOLTAGE= 5V PEAK TO PEAK MAXIMUM.'

DATA SHEETS & SOURCES
(1) Transformer
https://www.farnell.com/datasheets/1945759.pdf
(2) Transistor
https://www.onsemi.com/pub_link/Collateral/2N6515-D.PDF

 

[fireant007]

Hey Spec,
I have to say ... it's a bit like Christmas every day round here ....
Wake up early every morning and lo and behold another present under the tree.
Building today will report results.
Looks nice and simple, even I understand it. ( I think )

Not trying to be critical or anything .. but is there any reason you don't like this design ?

 

[spec]

Hey Spec,
I have to say ... it's a bit like Christmas every day round here ....
Wake up early every morning and lo and behold another present under the tree.
Building today will report results.
Looks nice and simple, even I understand it. ( I think )

Not trying to be critical or anything .. but is there any reason you don't like this design ?

View attachment 98830

Hy fire,

Your Xmas tree analogy made me smile.

Glad you liked the bridge circuit. It may seem obvious now, but I didn't fully appreciate the characteristics of the tubes and hadn't realized that the transformer that Les recommended had a secondary voltage of 55V and not 50V RMS, or that the voltage is specified at full load. The supply line is a bit lower than optimum but it does comply with the data sheet for the tube striking voltage- just. A 110V mains input, as opposed to 115V will be critical and, if you do a worst case tolerance calculation, the supply line is a bit on the low side. But I think it will be OK on the day, as they say, especially for a one-off build. There is also the get-out of putting a few extra secondary turns on the transformer by hand to increase the DC supply line voltage a touch.

From what I see on the net, a 170V supply line is optimum for this type of tube, and you may find that the odd tube may not strike at 140V especially as the tubes are old and when they have never been used before. In those situations, give them a blast with a higher voltage supply line and once the tubes have run for a time they should then conform to the data sheet. Any reluctant tubes could also be heated and/or have a bright LED flashlight shone through the glass tube to get them to strike.

No need to worry about being critical. It is always appreciated when someone analysis my circuits and I welcome it when they find genuine errors and make specific suggestions for improvements. It is quite difficult just doing paper designs. Normally you would prove the circuit on the bench as you go along- that normally weeds out any problems/errors.

About your suggested rectifier circuit- there is nothing wrong with it and, in the days when rectifier diodes were expensive, it was very popular. It also has the advantage of having only one diode forward drop in the rectifying path. It does have the disadvantage though of not making maximum use of the transformer copper which results in a higher DC supply impedance for a given transformer size. The major problem with using it for the tube power supply, with the transformer available, is that it would only result in a supply line of (55V * 1.414)- 1V [rectifier forward drop] = 76.77V, which would not be sufficient to strike the tubes (140V required).

spec

 

[Les Jones]

If you have problems with any tubes striking if the mains voltage is low then here is a suggested solution. The Idea is to still have the main HT supply at about 140 volts but also have an auxiliary low current supply of about 280 volts that would just be used for striking the tubes. The Idea is to feed the anode of each tube from the 140 volt rail via a diode. there would also be a feed to the anode of the tube from the 280 volt rail via a high value resistor. (Say 2.2 meg) So before the tube strikes there would be 280 volts on the anode via the resistor. This should be enough to strike any tube. When the tube strikes the voltage across it will be less than 100 volts so it will then be fed from the main 140 volt supply via the diode.

I hope this makes sense.
Les.

 

[spec]

If you have problems with any tubes striking if the mains voltage is low then here is a suggested solution. The Idea is to still have the main HT supply at about 140 volts but also have an auxiliary low current supply of about 280 volts that would just be used for striking the tubes. The Idea is to feed the anode of each tube from the 140 volt rail via a diode. there would also be a feed to the anode of the tube from the 280 volt rail via a high value resistor. (Say 2.2 meg) So before the tube strikes there would be 280 volts on the anode via the resistor. This should be enough to strike any tube. When the tube strikes the voltage across it will be less than 100 volts so it will then be fed from the main 140 volt supply via the diode.
View attachment 98834

I hope this makes sense.
Les.

Hy Les,

Another brilliant idea- I like it.

As the max voltage that can be applied to the tube is 250V it will be necessary to juggle with the voltages a bit, but that would be no big deal.

spec

 

[fireant007]

Brilliant guys.
I am learning a lot here.

Thanks for the waking them up suggestion with high voltage Les. I will see if I need it.

Points taken about the other circuit spec and as it stands your circuit is working very well.
I have an oversized capacitor at the moment so need to see how far down I can take that.

Mine is about 100uf @400v ... so a bit big for the job.
Although having said that .. there is still nearly a volt of ripple on it.

My experience in the shed today has been very good.

All using 240vac.... will test with just the 110 winding tomorrow and see how far down I can go reliably.

Circuit works very well and have hooked up the arduino into the op amp with a pwm input.
To get analog I just use a low pass filter with a 0.1uf ceramic and a variable resistor (at the moment for testing) to get volts out.
Seems to be the way to do it and is way fast enough for these tubes.

Final values for the transistor biasing and anode/cathode resistor are still in the experimental stage.
I am not much for calculating bias voltages etc .. so just chuck resistors in and probe around to see how efficient it is in practice.
It s a bit of a juggling act.

I can get microscopic ... movement out of the tubes with just a little code so that is great.
Getting about 0 to 3 volts out of the op amp so that is plenty of resolution for the tubes.

I found that driving the PWM at about 55 ... not the full 255 is about right for full scale.
That gets the op amp to get the 3vdc to drive the tube to the top of the bar.

These things are defiantly NOT linear though .. because at 27 PWM or so is over half scale, not by much but noticeable.
Might have to make a lookup table up in code to get it linear.

Fast .. these things are not ... tried really getting them going up and down really fast and found that I need at least a 10ms delay in code to get it looking nice.
Any faster and the top does not light up as it should. The bottom sometimes breaks away too.

EDIT: this could be my low pass setup too .. will check.

Very happy to see code driving the thing up and down like a yo yo though, a real treat.

The resistor is back in as suggested and it has taken a lot of load off the transistor.
Using a 4k7 1 watt at the moment and it seems to be dropping about 50vdc across it at 12mA.

It does get quite hot and is dropping about 0.6 watts at the moment.
The transistor I was using was rated at 4 watts .. but as you have said spec that is a job for a resistor.

At the moment I am just optimising component values and trying to use a smaller transistor.

I have found after testing/abusing about a dozen of these tubes they all start off with their own personality.
Striking them at a high enough voltage seems to wake the reluctant ones up.

I have abused the hell out of the 12 I have been testing with so they probably won't last very long.

But ... any "new" tubes that fail to strike will be getting current limited to 12mA max ... and just struck and exercised to their full bar length.

Now ... I have a LOT of work to do.

Getting a some big verro board and expanding this out to 50 tubes first to see how that goes.

More to follow including final circuit.

Cheers fellas

 

[spec]

Brilliant guys.
I am learning a lot here.

Thanks for the waking them up suggestion with high voltage Les. I will see if I need it.

Points taken about the other circuit spec and as it stands your circuit is working very well.
I have an oversized capacitor at the moment so need to see how far down I can take that.

Mine is about 100uf @400v ... so a bit big for the job.
Although having said that .. there is still nearly a volt of ripple on it.

My experience in the shed today has been very good.

All using 240vac.... will test with just the 110 winding tomorrow and see how far down I can go reliably.

Circuit works very well and have hooked up the arduino into the op amp with a pwm input.
To get analog I just use a low pass filter with a 0.1uf ceramic and a variable resistor (at the moment for testing) to get volts out.
Seems to be the way to do it and is way fast enough for these tubes.

Final values for the transistor biasing and anode/cathode resistor are still in the experimental stage.
I am not much for calculating bias voltages etc .. so just chuck resistors in and probe around to see how efficient it is in practice.
It s a bit of a juggling act.

I can get microscopic ... movement out of the tubes with just a little code so that is great.
Getting about 0 to 3 volts out of the op amp so that is plenty of resolution for the tubes.

I found that driving the PWM at about 55 ... not the full 255 is about right for full scale.
That gets the op amp to get the 3vdc to drive the tube to the top of the bar.

These things are defiantly NOT linear though .. because at 27 PWM or so is over half scale, not by much but noticeable.
Might have to make a lookup table up in code to get it linear.

Fast .. these things are not ... tried really getting them going up and down really fast and found that I need at least a 10ms delay in code to get it looking nice.
Any faster and the top does not light up as it should. The bottom sometimes breaks away too.

Very happy to see code driving the thing up and down like a yo yo though, a real treat.

The resistor is back in as suggested and it has taken a lot of load off the transistor.
Using a 4k7 1 watt at the moment and it seems to be dropping about 50vdc across it at 12mA.

It does get quite hot and is dropping about 0.6 watts at the moment.
The transistor I was using was rated at 4 watts .. but as you have said spec that is a job for a resistor.

At the moment I am just optimising component values and trying to use a smaller transistor.

I have found after testing/abusing about a dozen of these tubes they all start off with their own personality.
Striking them at a high enough voltage seems to wake the reluctant ones up.

I have abused the hell out of the 12 I have been testing with so they probably won't last very long.

But ... any "new" tubes that fail to strike will be getting current limited to 12mA max ... and just struck and exercised to their full bar length.

Now ... I have a LOT of work to do.

Getting a some big verro board and expanding this out to 50 tubes first to see how that goes.

More to follow including final circuit.

Cheers fellas

All looks good fire.

Just a point about transistor dissipation. The data sheet may say that a particular transistor is rated at 1W for example, but you will probably notice that 1W is with a case temperature of 25 deg C. There is also the secondary break down characteristics to consider. This is also known as the safe operation area (SOA). If you post the transistor type number we will be able to look into that.

Also any biasing calculations can be done for you.

The micro control side sounds fun. PWM is a simple but effective approach for controlling the opamp inputs.

Looking forward to seeing some pictures of dancing tubes soon.

Cheers

spec

 

[fireant007]

Roger that.... YouTube monster ugly vero board dancing tubes coming up.

SOA .. you say ... the mystery's of the data sheet ... is there anything that can't be ignored !?

I can only show you one dancing at the moment.

Might as well take a video of that.
Long way to go.

I bought a whole bunch of the ones you recommended in the first circuit so am going to try to shoe horn them in.

https://www.farnell.com/datasheets/671645.pdf

The transistor I am using at the moment is up in the shed !... and my shed is 30M from my house and I am buggered.
It's Killkenny O'clock here !

I WILL let you know what I am using tomorrow.
No real point in using them though if I can get the above working.

 

[spec]

fire,

For the record, here is an updated issue of the version 1 tube PSU:

​

ERRATA
(1) D1 should be, 2A 400V
(2) D10 should be, 2A 200V
(3) Add one 100nf low loss capacitor across the tube supply line for every 5 tubes. Suitable capacitors are ceramic and polycarbonate metal film.
(4) All transformer windings shown on the schematic should have a dot uppermost.

NOTES
(1) C2 & C8 values are shown for the full complement of 100 tubes. The value of both capacitors can be scaled down in proportion to the actual number of tubes used. C2 & C8 minimum values are 100uF. C2 and C8 do not have to be the same value provided the minimum value requirements are met in accordance with the number of tubes fitted.
(2) The transistors must be in a position to allow a good flow of cool air all around them by convection. The transistor ambient air temperature should be as cool as possible.
(3) Solder the transistors to substantial PCB traces to help conduct heat away.
(4) The collector resistors should also be mounted in as cools as possible ambient air temperature.
(5) the collector resistors should be mounted to allow a free flow of cool air all around them by convection

 

[Les Jones]

If the - end of the top left capacitor was taken to the centre tap on the transformer instead of the end of the winding then only about 70 volts would be added to the main full wave rectified supply. So this would make the striking supply about 210 to 225 instead of about 280 to 300 volts. (Just a variation on spec's design.)

Les.

 

[spec]

If the - end of the top left capacitor was taken to the centre tap on the transformer instead of the end of the winding then only about 70 volts would be added to the main full wave rectified supply. So this would make the striking supply about 210 to 225 instead of about 280 to 300 volts. (Just a variation on spec's design.)

Les.

yeah that would be better Les.

spec

 

[fireant007]

Morning has broken,
Off to the shed.

Question: How come we are going back to this design ? I could not get it to work the first time and it seems inefficient ?
ie: having to drop such a large voltage (232vdc) in the form of heat ?

 

[spec]

Morning has broken,
Off to the shed.

Question: How come we are going back to this design ? I could not get it to work the first time and it seems inefficient ?
ie: having to drop such a large voltage (232vdc) in the form of heat ?

Hi fire,

I am not suggesting going back to the version 1 PSU. I just updated the schematic to be correct and also to make the transformer connections clear by including the transformer lead colors.

As you say, PSU version 5 is more efficient. It is also more compact and cheaper requiring only one reservoir capacitor.

Incidentally, if you are thinking about experimenting with version 1 PSU, I have found a transformer which gives around the ideal 170V tube supply line with modified PSU version 1 circuit. https://uk.rs-online.com/web/p/prod...ll79VgCOMOvrHJfJbDWGZlM9pj-2UL94x4RoCZajw_wcB

spec

 

[spec]

SOA .. you say ... the mystery's of the data sheet ... is there anything that can't be ignored !?

I bought a whole bunch of the ones you recommended in the first circuit so am going to try to shoe horn them in.

https://www.farnell.com/datasheets/671645.pdf [2N6517]

I have analyzed the 2N6517 with respect to power dissipation and, good news, its maximum power dissipation figure is in an ambient air temperature of 25 degC, rather than a case temperature of 25 degC: there is a vast difference between the two. This, and the low cost, is probably why I selected this transistor in the first place, but I can't remember (just for my reference in future, ThR Junct to Ambient is 200 DegC/W, and maximum junction temperature is 150 deg C).

There is no Safe Operating Area (SOA) information on the data sheet because, in this case, that is obviously taken care of by the maximum power rating. The 2N6517 has a maximum collector current of 500mA and will only be running at 12 mA maximum, so there shouldn't be a SOA problem anyway.

Whatever a component's maximum power dissipation figure is, a full thermal budget calculation needs to be done to establish the heat sinking required. I have done that for both PSU version 1 and PSU version 5 and have included the heat sinking requirements in the notes for both the transistor and collector resistor. Basically, both components must be mounted in air as cool as possible. Also the air must be ventilated to allow cooling by convection. In general, the cooler a component runs the more reliable it will be. With transistors it is the junction temperature that counts.

The collector resistors are essential to limit the power dissipated by the transistors, and you may need to consider fan cooling for your 100 dancing tube mega display creation.

spec