DC-DC 12V-1kV

Go To Last Post
88 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have a little cute CRT tube named 3ЛО1И and I feel guilty of not giving it enough love. From what I know, it needs whereabouts of 500-800V on the accelerating anode, I don't think the current will exceed 100uA. I know how to make 800V from 220V with a regular isolating transformer and a diode multiplier. But I would really like to keep it small and wallwartsome, potentially nixieclockable, so I'm shooting for 12VDC->800VDC conversion. Maybe someone has done something of the sort recently, care to share your experience? I'm thinking of using an AVR to switch a trusty IRF630, like in a buck-boost converter, but with a switchmode transformer in place of the coil and feeding a 2x multiplier from the secondary. Worth trying, or there's a better way to do it?

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Push-pull or flyback converter with an appropriate transformer? But with a ratio of 1:66 this is going to be a big transformer with boatloads of windings on the secondary. Maybe 1:33 with a 2x multiplier.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Re: the amount of windings, we can talk pretty high frequencies here so maybe fewer windings. I know of no good way of estimating them other than by experimentation. Say, how do you calculate how many do you need on the primary on a toroid with such dimensions and such permeability at such frequency, like, why don't we just make it 1 winding on the primary and 33 on secondary? I guess at some frequency 1 winding will make enough Z for the power supply not to burn down? I remember the past full of sparks when I tried to make a switchmode power supply and my secondary was 1.5 windings. It worked, until it exploded, but I don't think the transformer was at fault.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think you first need to determine at what frequency you want to operate it, and the amount current ripple allowed and from that an inductance can be calculated. From inductance you can determine the amount of windings needed on the primary side.

You might find good information in the datasheet of the LT3439 and the AN118 application note.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have generated high voltages using a xenon trigger transformer. Very small. Not sure how well it will work at high repetition.

It all starts with a mental vision.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Usually the inverters for the bulbs are incapable of producing sustainged voltages. They need some HV to ionize the gas and then the voltage is pretty low.

Thanks for the pointers jayjay, reading.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

1:33? Why use such high ratios? If, for example, you create 160V in the primary, the secondary will only need to have five times the primary turns.

Driving a flyback is easy: We turn the power stage fully on until the inductor's current exceeds a certain, pre-calculated value. Right after that we turn the power stage off, making the core nuts until it returns all the magnetic energy it has stored in its air-gap. The core will try to do anything to return the stored energy back: It will try to find the easier way, creating an opposite current either in the primary coil destroying the power stage or in the secondary one lighting up our load. When the core is fully (Discontinuous Mode) or partially (Continuous Mode) discharged, we turn the power stage on again and we repeat the above indefinitely until the generated voltage in the secondary reaches the desired value; then we stop and wait for that voltage to drop below another threshold in order to repeat everything. This is the secondary voltage regulation. We can monitor the secondary current if we need a constant current source. For safety reasons we should monitor both! In a more elegant way, we can avoid the so-called "hiccup" of the switching power supply if we try to regulate the output by changing on the fly the driving pulse width, or the frequency, or both, instead of brutally turning the whole thing on and off. Then, we have a so-called "green" SMPS! (I know how much you love that term :P)

Now, windings of a few only turns give the inductor not much inductance, which requires higher switching frequencies in order to avoid saturating the primary coil and melt everything by skyrocketing the primary current; multi-turn windings have larger inductance and require lower frequencies to give the core the time to store the energy. Actually, we are not talking about a transformer because we do not collect the energy as we are pushing it to the primary winding; this is the Forward Topology. Instead, we collect the energy stored in the magnetic medium, which is the ferritic core. So, this is two coils strongly coupled magnetically together using the same ferritic core, where the primary creates energy that is stored in the visible air-gap of the ferrite core (or in the invisible countless tiny gaps created by the bonding material of the ferritic powder the ferrite cores are made of), and we discharge the coil's energy using the second coil (the "secondary"). This is the so-called Flyback Topology that energised the electron beam of the TVs, where it was firstly used commercially and, thus, became known; the electron beam that "flew back" to the next line of the field/frame during the blanking time after rendering the end of the previous line. :)

So, we should know the power the flyback will need handle in order to choose the ferritic core material, properties and size. Knowing the type of the ferritic material, the maximum and the ripple current (I_max-I_min) of the primary inductor and the time of the current transitions (in other words, the switching frequency), and we can calculate the primary coil inductance per turn. An LCR meter will be proved to be valuable, if it exists, to verify the inductance of the masterpiece we created in order to calculate the exact frequency needed to drive it. For one-off's you might even need to trim the core gap with a fine file!

Anyway, regarding the boring details about formulas, snubbers and MOSFET driving, you can find almost everything you need to know in the manufacturer's application notes and data sheet of their power switching products. If I remember correctly, TopSwitch offers guides for the transformer design.

-George

I hope for nothing; I fear nothing; I am free. (Nikos Kazantzakis)

Last Edited: Wed. Mar 7, 2012 - 03:47 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Flyback or push/pull. Don't drive the transistor too fast otherwise you'll get huge spikes on your switch due to the transformer's leakage inductance. I'm a big fan of a high ratio transformer so that you can keep your primary switch's voltage rating low.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Check out the little inverter modules used for florescent LCD backlight tubes. The high AC voltage is there, just add a rectifier!

Tom Pappano
Tulsa, Oklahoma

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Tom Pappano has about the best idea of the lot in terms of ease of implementation. One caution for this approach is that these LCD CCFL drivers have high output impedances. The output voltage is designed to foldback after the CCFL tube "strikes". This may or may not effect you at 100 uAmp draw, it depends on the specific module. These inverter modules generally have two output voltages specified: the "starting" voltage and the "running" voltage. The starting voltage is always higher and is essentially the no-load voltage - and probably the specification you should be looking at for your application. The running voltage is the voltage the output will pull back to as it is loaded to the specified output current (after the bulbs "strike"). Generally these CCFL backlights take around 5-10 mA running current per bulb, so your 100 uA load will fall into the "no-load" or "starting voltage" mode of operation.

If you still want to build your own DC-kiloV converter, I'd recommend the diode multiplier approach. There is a little trick to getting the traditional diode multipliers to work from a DC supply rather than an AC one. If you want to pursue this path, let me know & I'll give you some guidance.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for all the advice. I'm a little bit worried that hacking a ready-made CCFL inverter may take me longer than building my own from scratch. I have experience making buck-boosts for the nixies and they are only a step away from flyback converters so I'm rather looking in that direction. Probably even push-pull or half-bridge to fully utilize the core. Anyway, at the moment I'm only lazily researching the options, any input is appreciated.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Especially when the current consumption is low, a boost converter is IMO the way to go. Finally found time for the little scope-tube ?

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I just want it to glow :)
Do you think a boost converter will do? Somehow I have never seen a boost converter that goes over 200V, but I do not know why. Do you?

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The 1000V+ schottky diode for the boost converter will be the hard part to find.
How about a boost converter to 200V (nixie-wise) followed by a multiplier: a push-pull-stage driven by a squarewave generator feeding a diode-capacitor matrix. Of course those should be flux-capacitors :lol: But the diodes can be ordinary 1N4007's or so.

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I am surprised no one has suggested a Cockroft-Walton multiplier http://en.wikipedia.org/wiki/Cockcroft%E2%80%93Walton_generator. That is easy to get working.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you want wall-wart style, 120V AC to 800V DC is a lot easier to accomplish than 12V DC to 800V DC. If you're gonna plug it in the wall anyways why would you reduce voltage before amplifying it?

From 120V 60Hz AC, feed into a 2-stages Villard doubler with 68uF caps to get ~580V AC, and rectify to ~820V DC, 100mA. Wear gloves and shoes.

Also because each stage of those ladders only operate at twice the input voltage, and since regular 120V peaks at ~170V, all components before the rectifier only need to be rated for 400V.

Last Edited: Thu. Mar 8, 2012 - 03:04 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hmm funny how similar Villard cascades are to CW... Same? Was there a battle for a name that was won somewhere?

EDIT ok so from reading a bit more....

CW = Greinacher <> Villard Cascade, but are sometimes incorrectly refered to as.... Phew...

A CW cascade is a Villard cascade with an added peak detector on each stage...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Plons wrote:
The 1000V+ schottky diode

I don't think you need a Schottky diode , the 0.4V vs 0.7V drop difference is negligible at these voltages.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi, I did not read all the posts, but I have done a 5V to 800V DC/DC converter recently.

I used the LT3580 and schematics according to datasheet for -350V. I use dual LDT565630T-011 transformers.

I think 800V might be out of spec. Im running at 350V. But I have tested up to 800V, and it worked at least for a while. And I have a little higher current consumption than you.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@hugoboss: because I want to use store-bought 12V (or 9V, or something) wallwart.

@dak664: lol

@AgwanII: good to know that those appnotes are scalable and thanks for the transformer reference.

@Plons: that could be a very neat idea, given that a good part of this circuit would be tried and true for me. Let's see, my typical nixie converter can probably put out ~200V at 30-40mA. Sounds like enough juice to quadruplify it and have enough for the second anode and deflection system.

But it bothers me that I could do all that in one stage by using a flyback or pushpull. And unknown transformer. Say, to begin the experimentation, I have a nice ferrite ring which is about 22mm OD. I could start experimenting with some random number, e.g. 5 windings on the primary and 350 on secondary and see how it goes. I would start right away but I do not have a wire that would physically fit 350 windings in that ring. Not to mention the questionnability of excitement of threading a loop of wire 350 times through a tiny hole. So 2-stage could win :)

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jayjay1974 wrote:
Plons wrote:
The 1000V+ schottky diode

I don't think you need a Schottky diode , the 0.4V vs 0.7V drop difference is negligible at these voltages.


Agree. But the switching speed requires the use of a schottky. IMO. A fast recovery type is also an option.

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

In a boost converter straight to 1000, shouldn't the Vds of transistor also be >1000? I don't know, sounds weird to me.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

svofski wrote:
In a boost converter straight to 1000, shouldn't the Vds of transistor also be >1000? I don't know, sounds weird to me.
Yep, the transistor as well.

@dak664: that's what I meant with "teh matrix"
I should stop posting after midnight.

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

svofski wrote:
@hugoboss: because I want to use store-bought 12V (or 9V, or something) wallwart.

But.... Why? Because they look good?

You can still use multiple stage multipliers from 9V but loss is MUCH higher, and there are a LOT more components involved...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think going from 12v to 1000v directly with a boost
is not so easy, you have an extreme dutycycle.

I also have experienced difficulties in making HV
transformers since the parasitic capacitance of the secondary together with high inductance results
in a low resonance frequency, so you cant operate
at high frequency.

I would suggest a push-pull (two logic-level
MOS transistors primary driven by two non-overlapping PWM outputs from the AVR. This effectively doubles the input voltage. Then I would transform to 250V and use a voltage quadrupler.
(Been there, done that....)

Another remark: It may be that you need a 800V diode
for 400V rectification, depends on the configuration.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If I can go from 5V (USB) to 800V with my small solution, it shouldn't be that hard to go from 12V to 1000V?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The problem is not that you CAN go from 5V to 800V, it's mostly that you probably shouldn't, when there is a perfectly good 120V waiting in the outlet.

WHY WHY WHY would you want to take 120V, make 5V from it, then boost it back up to 800V, using 10x as many components??!?!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You could make it portable then. How did they do this on camcorders that use a tiny CRT for the viewfinder?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Probably as I have done. My solution takes 10 components or so, and takes up 2-3 square cm of PCB. Could be smaller if I wasn't interested in having the voltage settable and mesurable.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@hugoboss: because I don't like dealing with stuff in the wall, I prefer it when it's dealt with by certified professionals in the area.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

svofski wrote:
@hugoboss: because I don't like dealing with stuff in the wall, I prefer it when it's dealt with by certified professionals in the area.

Hmmm.. Well 800V DC isn't much safer TBH... 800V burns through skin, 120V AC just jolts you (unless you somehow decide to hold on to the wires)...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

hugoboss,

In the OP svofski wrote:
I know how to make 800V from 220V with a regular isolating transformer...

220V is available, not 120V, all more the reason to avoid direct conversion from the line for the minmal power needed.

Stan

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'm carrying on with the project slowly. So here's the pushpull transformer driver:
https://www.circuitlab.com/circu...
Yesterday I tried it in real life. It worked nicely, up to some point. The MOSFETs that I had handy didn't have too much stamina, only rated for 20V BRVds. At 3.5 to 5V of input DC I had something like 100-140Vp-p on the output with my random-wounded toroid transformer, loaded on a 15K resistor. At 6V of input voltage the magic smoke was finally released. I imagined that the transformer would kickback at ~ double of input voltage, but what I have observed must have been at least a quadruple. Curious. The simulation doesn't show anything suspicious.

I'm waiting for some more durable MOSFET and I ponder meanwhile. Maybe I'll burn a few more of those puny ones just to see what exactly kills them.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You sometimes can get some nice MOSFETs from old
motherboards. In a push-pull it is important
thet the primaty is very "symmetric", best
is to wind two wires bifilar (together) and the
connect them accordingly. Its also important to
drive the mosfets with enough voltage if they arent
logic level ones.

I am sure we had a thread here about that stuff
some years ago...

Good luck !

Here is the old thread, make 12 Volts from 5 was the application. There is some info about push-pulls
there.

https://www.avrfreaks.net/index.p...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The symmetry indeed appears to be an issue. I now have a pair of good old IRF630Ns pushpulling my rather nasty looking coil. Here it is so you could have a laugh:

The primary is 5+5, the secondary is hell knows how many. I picked a coil that I used in some of my previous experiments.

Anyway, running at 80kHz, the smoke does not get released immediately. One of the transistors always gets rather hot pretty quickly, while another tends to stay cool.

Such is the output:

And such is the the drain of the hot transistor:

And the cold one:

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Update: bifilar-rewound the primary. Running at 80kHz now, both transistors are lukewarm. Awesome! Thanks ossi for the bifilar idea.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

As I'm moving on to the next step and trying to make the current design repeatable. Any hints on reducing the transients without dramatic redesign efforts? For the reference, after making the primary symmetric, both transistors see the former "hot" picture. A lot of energy seems to be wasted there.

I studied Peret's clock circuit a little bit. There are a lot of secondaries in the design. Initially I planned on getting what I have now, feeding it to a voltage multiplier and then just use a series of resistors to divide the resulting voltages into everything that I need. I'm not really comfortable with transformer design, so I'm trying to minimize the windings parts of the project. The currents in a CRT appear to be ridiculously small, so this plan should work, but what do I know.

Will the regular 4007 type diodes work fine in a multiplier at ~80kHz?

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Will the regular 4007 type diodes work fine in a multiplier at ~80kHz?

I am afraid not, Svo. The 30us recoverytime is the spoiler.

Nard

Attachment(s): 

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yeah, I was looking at the same figure, Nard and it was worrying me. So I'm going to get some that have "Maximum Reverse Recovery Time: 200nS". I hope those will perform well. Oh dear!

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It is nice to see that you sorted things out! Now, Nard is right about the 1N40xx/1N54xx series; these diodes are too slow for such tasks. For Ultra-Fast solutions there are the UF40xx/UF54xx series at 1.0A/3.0A as well as the MURxxx family, all at 50ns/75ns/100ns. For lower or low voltage rectification with recovery times down to 1ns(!) I prefer the MBRxxx family Schottky diodes.

-George

I hope for nothing; I fear nothing; I am free. (Nikos Kazantzakis)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Transformer design is quite a black art, maybe even magic some would say. Boatloads of variables and one big compromise. All the different unit systems don't help either for the novice.

The spikes could be due to leakage inductance. Try a snubber network.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

After some digging in my paper-storage I found it:

In my scope-clock I operate a DG 7-74A tube using the
following SMPS design (EFD30 core):

primary: 6+6 turns push pull from 12V DC

secondary 1: 3 turns tube heater
(The heater must be isolated because cathode usually
is negative with respect to ground and voltage
difference between heater and cathode must be small.

secondary 2: 20 turns + voltage doubler(2 diode)
supply cathode and grid1 (negative) voltage.

secondary 3: 100 turns+2 diode doubler
generates -400 Volts for catode

secondary 4: 50 turns+doubler generates +200 Volts
for deflection amplifier

secondary 5: 100 turns + doubbler
stacks onto +200V gives +600V

And it works !

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Jayjay wrote:

Quote:
Transformer design is quite a black art, maybe even magic some would say
O yes, it is black magic :!:
And I love it.

@ossi: Hello Professor Albus Percival Wulfric Brian Dumbledore, Head of the Academy for Magic Black Art. Always a pleasure to see you and/or one of Thou Art :)
We (me and my fellow students) would love to see a picture of that SMPS++ design.

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The master of perf-board and magic smoke responds:
Attached see my scope-clock. The EFD30 transformer
is surrounded by HV fast rectifier diodes and HV elcaps.

The 4 small transistors are the deflection amps. The two TO250 transistors are the push-pulls. The
many resistors form two R2R DACs driven by an ATmega32.

Transformer design needs much magic smoke. You take it
out of many transformers and put it back into one
good design. Some call it try and error.....

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Good point about the cathode voltage, I didn't think about this.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

And the pots at the top are for setting the levels for the specific grids ? Is there need for stabilization ?

Nice work, ossi.

@svo: with a core as ossi used it is easier to make high count windings. Your toroid is great for lower voltage SMPS's. But be carefull with the mu. Scavenged from an inputfilter with 2 equal windings: those have a high mu but saturate fast.
The cores that I use, apart from the ones I buy, are scavenged from old powersupplies: PC and other. Great source :)

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@Nard: I don't notice any problems with my core, I think it's fine for this purpose. 150VAC is low voltage :)

Anyway, an update:
0) yesterday i measured ~300Vp-p with a lot of transients to which I did not pay too much attention;
1) I assembled a half-wave doubler. I expected to see +300VDC on its output, but I saw 600. Wow I thought.
2) Soon I noticed that not all is going well, the transistors were heating up quickly;
3) After some experimentation, I added 10nF caps between the drains of the MOSFETs and ground and increased deadzone between pushy and pully pulses (it was 0, now it's like 10%). This made the transients go a great deal. The pulses on the drain now look pretty flat and I'm seeing expected 300V on the output of half-wave doubler. No extra 300V for free, but also no heat anywhere: I like it.
4) added a 1 winding secondary for the heater. Tried it with a 6.3 bulb.. To be honest, I'd feel safer if the voltage was lower. I added a 0.033 cap in series.

@ossi: i browsed through the epic inductor thread and I saw many messages about your scope clock but the images did not load. Do you have an album somewhere? I guess I love scope clocks :)

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Mmm one another: I tinkered a lot in the circuitlab and came up with this layout:
https://www.circuitlab.com/circu...
So far I only tried the top part of it and it appears to work absolutely as expected (after killing the excess of transients). I'm not sure why, but it worries me a little bit that in this configuration neither of the secondary ends is tied directly to ground (currently, with only the top part implemented, the bottom end is grounded). Shouldn't be a problem, but I learned to expect unexpected things here. What should I expect?

Also, the ripple on the output seems negligible. Would you recommend adding an RC filter anyway, or it should do just fine without? As for cathode/brightness/focusing/acceleration, they seem to not really care and as for deflection, I'm planning to use the longtailed differential configuration, like everyone else does, and it too should be pretty insensitive to ripple, no?

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
I'm not sure why, but it worries me a little bit that in this configuration neither of the secondary ends is tied directly to ground (currently, with only the top part implemented, the bottom end is grounded). Shouldn't be a problem, but I learned to expect unexpected things here. What should I expect?
Nothing serious. Just be aware that the floating secundary will be at a high potential. So good isolation is necessary. And what potential depends on the doubler configuration.

For the ripple: I don't know. Never built a scope clock ;)

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Okay that's cool! I wouldn't say that when one of the ends is grounded there's nothing to isolate either.

A picture due from the last night

Those little square thingies on the right are 2220 0.1/1kV X7R caps. It's crazy how tiny and seemingly harmless they are for such huge charges.

BTW. What do we count for "a winding" on a torus? How many "ties" does a wire make around the core, or how many times the wire passes through the hole in the middle? It never bothered me before, but now in the days of 5 windings on the primary and 1 on secondary, one has to ask himself such questions.

Quote:
Never built a scope clock

I'm also quite intrigued by hydrogen dekatrons, they require around 600V. Enough to sparkle the interest.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Okay, so I have advanced a bit. I have all the high voltages, adjustable and everything in HV is like I expected (full span is -450..+300). However 6.3V is giving me trouble. As you can see in the picture above, I had lighted the tiny bulb. But when I tried doing the same with the CRT heater, one of the trusty IRF630Ns in the push-pull died almost instantly. I'm trying to analyze the problem before I fry another MOSFET.. The heater resistance that I see is 1.7 ohm, which is ridiculously low: at rated 6.3V that would mean that it eats 4A: I would never be able to afford that. But I guess in a normal situation it heats up very quickly and its resistance then goes up. The datasheet says the nominal current is about 0.5A. Somehow it didn't happen, or didn't happen soon enough.

Another thing that bothers me is the mode of failure of IRF630N. It's rated up to almost 10A, my 12V source is able to provide 2A maximum. What went wrong? With just the resistive ladder and a lightbulb as a load they were running completely cold.

Until i figure it out I guess I can try heating the cathode from a set of 4 AA's.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


I don't know, this thing is so unstable. The only thing it doesn't react to is me blinking the eyes, everything else is instantly displayed and in the most unpredictable manner. The deflection amps are not yet done. Is this normal that there's no way to get a focused dot in the centre of the screen? I either get it blurred, or rather far away from the middle.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Svo,

Are you really using 1N4148 diodes? I thought that they had a PIV rating of 100 volts.

Cheers,

Ross

Ross McKenzie ValuSoft Melbourne Australia

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@Ross, I'm using SM4937, they are rated to 600V and have 200nS recovery time. The caps are 0.1µF/1kV. Here's a closeup, sorry for the excess of flux :)

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Svo,
bringing up the 6.3 volts of the heater is a real pain.
inmy scope-clock it takes I think more than 30 seconds
to heat up. Sometime its good to limit the supply
current. Getting a sharp dot can also need some
experimentation as I remember. I remember it was
important to have the deflection grids at the right
potential with respect to cathode and anodes. Then
having too much AC on the voltages modulates the
focus and may blurr the picture. If you have a
magnetic field near to the tube, that will also blurr.

Good luck, in the end you will win !

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Heater resistance will change with temperature and eventually the current will drop.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here you can see a short video of my scope-clock

http://www.youtube.com/watch?v=h...

The clk gets its time from DCF77 time signal
transmitter. Below the center you see a "rolling"
display of the incoming pulses.

Current consumption at 12V is 300mA. During heatup
it needs more than 600mA.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@ossi what a lovely sight! It's really hard to believe that mine is going to be like that one day.

Do you have some sort of feedback in your power converter, or do you just let it pump happily? I have noticed that with time the voltages tend to climb up a bit. OTOH, I have also noticed that my tube is happy enough with much lower voltages too, so I'm feeding the circuit with 7.5V, just barely enough for a 7805 to work. It appears that some +/- 100V don't really make a lot of difference for such a tiny tube.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have no regulation/feedback. The load seems fairly
constant so I simply let it pump. Do you have the data
of your tube ? Old scope schematics are also interesting
to get ideas how to built up th supply/defelection
and what voltages to apply to which electrodes.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Enjoy with me a perfect wireframe of just one of the dark boxes which are about to come:

Those darker green patches are not highlights or reflections on the glass, they're some nasty electrons flying sideways.

Now with the deflection amps working, the transistors in the pushpull are getting rather hot. Uh oh, no me gusta.

The Dark Boxes are coming.

Last Edited: Mon. Apr 9, 2012 - 09:07 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Congrats,
the shadow-lights probably will disappeear
by reducing the beam intensity. What is the supply
voltage of your deflection amp and what is the
collector-resistor ?

In my deflection amp I use 220k collector resistor,
so the amp draws only a few mA.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The deflection amp has about 200V. The collector resistor is 2M2, which I'm afraid is too huge for driving a capacitive load. My biggest problem is now to achieve symmetry in the long-tail amps. It's always pulling much harder to the common-emitter side (the active drive side), and this results in a huuuuuge nonlinearity on screen, I guess of inverse square nature. I guess I need to reduce the bridge potentiometer (it's 50K now).

I don't know, when I'm fiddling with the Gate potentiometer, I get all kind of special effects, like zooming, sliding sideways, but nothing that I would classify "brightness reduction". I can remove the beam from the screen alright.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I've made a scope clock using the 3LO1 tube, a few years ago now. Here's a picture and the schematic. I used the ferrite transformer out of a small switching PSU, the core is about 18mm across, with the turns as marked on the schematic and mylar spacers across the legs of the E to achieve the right inductance. That's a roll of mumetal around the tube, by the way, as it's sensitive to magnetic fields from the power supply and external transformers.

I found, however, that these tubes don't live very long. After about six weeks the phosphor fades in the places the beam hits, until it can only be seen in a dark room. I can see this clearly by adjusting the X and Y so it scans different areas, that the burned out areas are nowhere near as bright as the new areas, though there's nothing visible, no burning, on the face. I replaced the tube and it was perfectly bright again for a month, then the same thing happened. I have a couple more spare tubes and now I only get it out and turn it on when I want to impress someone :) Otherwise it sits dark to save the tube.

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have seen that some scope clocks always move the
picture slowly a small amount in order
to avoid the wear out of the phosphor.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@peret: in fact I used your schematic as a reference (and your clock for inspiration). Thanks for making it public! If only I could make a circle round...

Indeed, even the documentation from these tubes says that the projected time is about 800 hours, which is about one month of non-stop work. The datasheet (etiketka) also says that you're kindly asked to return the sheet to the manufacturer after the device has completed its lifecycle, specifying the time it was used, the operating parameters and the cause of going out of service. Given that the majority of them they were made in the 60s-70s, I doubt that this kind of information is still expected.

Here are the scans:
http://istok2.com/data/250/

That's rather sad. By making the image float around, you can extend the time to a few months but that's about it. Do other tubes also tend to have life this short, or do they last slightly longer?

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

My tube DG 7-74A is from this equipment:

http://shortwaveradio.ch/radio-m...

This equipment was certainly meant for continous operation over some years. The base-line is heavily
burned in.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

svofski wrote:
@peret: in fact I used your schematic as a reference (and your clock for inspiration). Thanks for making it public! If only I could make a circle round...

Oh good! Thanks for the credit, it's good to know I occasionally do something inspirational :D I didn't have many problems with the HV supply for this, but the long tail pairs gave me many headaches. It's a difficult compromise to do it with a single pair of transistors. I think if I did it again I'd use a cascode circuit, with low voltage high gain transistors at the bottom and low gain high voltage transistors up top. The reason I didn't do that here is the lack of an intermediate voltage around 40V.

The phosphor coating on these tubes is extremely thin, almost transparent. It's much thicker on other tubes. Perhaps with it being so thin, it gets sputtered away by the impact of the electron beam. I have a 6LO1 tube also, the 50mm rectangular version. Its phosphor is also very thin. Unfortunately I've never been able to get even a green dot out of it, otherwise I might have used it for something similar.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here's the datasheed of 6ЛО1И:
http://istok2.com/data/533/
It's very similar to 3ЛО1И, but it needs slightly larger voltages in general:
A1: 45..135V
A2: 600..1200V
It also states worktime 1500 hrs. A curious thing about this tube is the gate, look at the picture. It has pin 3 called modulator and pins 12 and 6 called something that I can't really translate: landing modulator, mounting modulator, fitting modulator, planting modulator... I'm not sure what does it mean. Could it be that it's because of this second modulator you did not see anything?

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Are pin 6 and 12 connected inside the tube ?
Are there given voltages in the datasheet ?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It only says that modulator voltage is -125..0 relative to cathode. From the diagram I guess that they are 6 and 12 are connected inside, while 3 is separate.

Meanwhile, a set of ideal concentric circles :mrgreen:

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The bright circles are coming :lol:

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ossi wrote:
The bright circles are coming :lol:

HAHAhAHA

here's today's result:
http://www.youtube.com/watch?v=I...

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

svofski wrote:
pins 12 and 6 called something that I can't really translate: landing modulator, mounting modulator, fitting modulator, planting modulator... I'm not sure what does it mean.

"Grounding Modulator" perhaps? It's the screen grid. It's taken to a moderately high, steady positive voltage and the effect is to be a barrier (screen) to isolate the control grid, or modulator, electrically from the anode. The capacitive coupling of voltage fluctuations at the anode to the control grid is greatly reduced. I'm not sure that's of much importance in a CRT, where the anode voltage is effectively constant, so perhaps here it acts as an accelerator to boost the electron beam speed.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Grounding modulator sounds plausible, yes.

It's not really clearly visible from you picture, but am I seeing some focusing imprefection in your clock? I'm asking because I seem to be unable to converge the beam well. Part of this probably has to do with the ripple, when zooming/panning around I can see some running texture in the lines, but there is definitely some intrinsic problem with the focusing too -- if I get it almost okay on one side, it blurs more on the other side.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

My clock is in the same focus all over the screen. The minute ticks and the second hand are one pixel wide and that is meant to be a double tick at the top, with one pixel in between. So while the beam may not be all that narrow, it is thinner than 2 pixels. The deflection sensitivity is 256 pixels from side to side, by the way. The voltage on your deflection plates will affect the focus - you should aim to have them rest close to A2 voltage when the beam is centered, and swing above and below it. That isn't always possible I know, but the close the voltage the better it will look. Also the deflection sensitivity is not exactly the same for the X and Y plates, because one set is closer to A2 than the other and the electrons are moving faster, but if you have a variable resistor in the tail you can adjust the gains to make the sensitivity equal.

The tube is really quite sensitive to magnetic fields, and if you have any kind of mains transformer within a couple of meters it will pick it up. If you don't have mumetal you can probably make a shield out of thin steel from a tin can - it must be magnetic - it won't be perfect but better than nothing.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Just an update:

And here playing with shapes and transforms:
http://www.youtube.com/watch?v=_...

The focus has become much better after I gave it back +12VDC, thus increasing A2 voltage significantly. I can see how it reacts to a metallic screwdriver when I trim the pots, but it doesn't seem to care at all about sitting on top of the transformer, I guess it helps that the transformer is toroidal and it's rather far away from the deflection plates. There is also some nonlinearity across the screen, visible on the grid, but at this point I do not really care: it looks great.

Haven't yet tried to power the heater from the same supply. One of the things that bothers me currently is that the voltages of my converter tend to drift a little bit, especially immediately after the power-up and every time I have to either adjust the image to be centered, or wait 10-30 minutes until it stabilizes itself. For a clock that runs 24/7 it's not a problem, but I'm not sure if I really want to burn it out in just 1 month.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

That looks really good! Congratulations.
My scope-clock is also very sensitive to voltage
fluctuations. I think the deflection amps are
sensitive. But when I run the clock I do it for
longer times, so it's not a real problem for me.

Do you etch PCBs for your setups ? Your setup
looks really clean and well built, while my
perf-boards always look ugly.

What about some kyrillic letters
(not that I could read them...)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

And scary music under the Youtube-movie. Fantastic result, Svo !

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Very nice! Brings to mind the idea that the medium is the message.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks! Sorry for not replying in time, I was away from the computers and soldering irons and close to grills and wineries.

The cyrillic letters I really would want to display, perhaps later. This font I borrowed from somewhere for my plotter and of course it did not have cyrillic characters so I would have to do them myself first.

I etch the boards at home, for me it's one of the most fascinating parts of every project, the magic does not fade with years. With cheap laminators and laser printers these days making pretty and rather complex double-sided boards has become ridiculously easy. BTW, I had to be rather creative with this one because it's longer than the free version of Eagle allows. I started it in Eagle and finished in gEDA/PCB. Had to write a data converter for that. For the curious the brave it's here (it's not a direct converter, it reads Specctra dsn/ses pair):
http://code.google.com/p/svofski...

Speaking of free PCB tools, the impression that I've formed while doing this project is that KiCad == EMACS, PCB == vi (but unfortunately not vim). Both appear to have stuck in the seventies usability wise. With PCB one can at least have some predictable results while doing some simple routing and refinement tasks, in the worst case you always have the text file to edit in a real vi. I don't want to say anything bad about KiCad but I couldn't do anything with it. Unlike EMACS, exiting it is very easy: it just crashes at will (disclaimer: unofficial OSX build) and the board files, although plain text, are streams of uncomprehensible gibberish.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'm puzzled by something weird. I'm using a TLV5625 DAC from TI. It's a single-supply 8-bit internal reference, voltage output blah blah DAC. Programmed via SPI. So when the programmer is plugged (it's a reprogrammed pickit2) everything works. When I unplug it, nothing works.

Scoping the signals revealed that pickit2 kind of trims down the signals to 3.3V. But the DAC datasheet says that it's good for 2.7-5.5V operation, and it's powered by 5V. What gives? My SPI lines are connected directly, without pullups or inline resistors. I tried adding pullups, they do nothing. Inline.. I guess they could help but I would like to hear some educated opinions first since they would require doing some controlled damage to the board. Any ideas?

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The PICkit2 has pulldowns, not pullups.

Warning: Grumpy Old Chuff. Reading this post may severely damage your mental health.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I am confused: Does the TLV5625 really have an internal
Vref ? The headlines in the datasheet say so, but I
only find a description for a VREF input in the
following sections. :shock:

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes, it has internal VREF.

I "solved" the issue by adding 220Ω/50pF network on SCK line. I can't believe I added a RC filter to a clock line, but with TI all bets are off.

I also solved my double-supply problem. Adding 0.5Ω in line with the heater worked a charm: it takes maybe a little bit longer to warm up than with a separate supply, but that's unimportant. There's still Z-modulation to do, but I'm so lazy about this now — looking at a dekatron tube and some egg-shaped cores I happen to have around ;)

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
I "solved" the issue by adding 220Ω/50pF network on SCK line. I can't believe I added a RC filter to a clock line, but with TI all bets are off.
Transmission line theory. My first contribution to AVRfreaks: the PPPPD. Same issue. It still resides in my signature.

Even digital is kind of analog ;)

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yeah well.. But in PPPPPPPGJXJJ the lines are long. Here they are like 2cm long and have a groundplane. And it's the first SPI device in my practice that's so sensitive to transients.

The Dark Boxes are coming.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have a Anita Sumlock (dismanteled) here, contains
also a Dekatron, many thyratrons, and a lot of nixies,
and in fact, there is one transistor in it !

http://www.vintagecalculators.co...

Seems we share the love for high voltages and glowing
tubes...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

We definitely do. That Anita is sexy with that argon-filled Dekatron!

I liked that silkscreen marking of the transistor. "That expensive thingy in a top hat goes here".

The Dark Boxes are coming.