CRT HV power supply

[ealex]

Hello

I've had a 8LO29I CRT tube for a while
The datasheet is here: http://frank.pocnet.net/sheets/084/8/8LO29I.pdf

The main points of the datasheet are :
electrostatic deflexion
HV supply of ~ 1200 - 1500V ( max )
Focus voltage 280 - 516 V
Grid 67.5 - 22.5V below cathode

There is a post on the forum from someone that is using a simmilar tube and made a small TV - i'll add the link here when i find it.

As in that post, I plan to run the tube with all bias above ground. The cathode will sit at ~ +100V (maybe a zener diode to keep it ~ stabilized ), the grid will be weakly biased for the desired brightness level and pulled to ground for spot turn-off - i want only on / off , no analog modulation.


I've already tested the tube and the deflection drivers ( differential pairs with some constant current sinks for biasing ) and it works as expected.

I plan to get rid of the big transformer as it's recovered from old radio and is big, old and rusty, etc.

I've looked at some old scopes - the CRT biasing / HV section and found some interesting ideas.

The circuit will require +/- 12V and +/- 5V for the DAC and analog section, +400V for the deflection drivers and ~1500V for the anode. Also, some 6.3V for the filament and 3.3V for the mcu - those i can get from the low voltage input.

the 12V rails will require ~ 10-20mA, the 5V 50mA at most and the 400V around 5mA
i have no idea how much current the HV supply will have to source, but i estimate that it will be around 2mA, most of witch will be spent in the resistor divider chain.

I've thought of 2 ways of generating the power supplies:
1. A standard fly-back converter, using a uc3845 and an E42 core - for easy winding. The supply will generate the 400 and low voltage rails, the HV one will be generated with a voltage multiplier from the raw 400V winding.

2. An oscillating fly-back, as used in the old scopes - no idea how to design it, found some literature but nothing too specific.

There is no need of galvanic separation - the ground will be common.

Should i go for the UC3845 circuit ? Or the oscillating version is easier to make ?
Is there any other way of generating all those voltages ?

[Richard Head]

ealex

Your biggest headache with HV switchmodes is going to be the inter-winding capacitance of the secondary. The inter-winding capacitance causes an oscillatory reflected current to appear on the current ramp of a flyback converter. To reduce the effect you should allow more flyback voltage across your MOSFET and re-design the transformer to allow for fewer turns on both windings. Also, split the secondary into multiple layers with insulation between each layer. If you spread the turns on each layer apart a little so the wires don't lie right up against the adjacent turn it helps.
The TV manufacturers started to use a "diode split" arrangement in the LOPT which split the secondary windings into several series connected ones with HV diodes imbedded in the winding. This reduced the inter-winding capacity but is a real PITA to make.
High power HV supplies are best implimented as resonant converters I think.

[ealex]

i've made some quick calculation for the flyback based on E42 core.
i'm using the formula / examples from SWITCHMODE POWER SUPPLY HANDBOOK, Keith Billings,Taylor Morey

Fsw is 50kHz, Pout = 10W, Pin=12.5W
Vin between 12 to 24v

Max transistor stress voltage - 40V => Np/Nsm = 3.98E-02
TonMax = 9.48E-06s

If i increase TonMax i get the conduction time over 50% - i don't know if it's good, i want to stay in discontinuous mode.

Primary inductance: Lp =2.59E-05H
Primary peak current: Ip = 4.39E+00A, and RMS: Irms_prim = 1.75E+00A

I have a epcos E42/21/15 core: http://www.farnell.com/datasheets/1595809.pdf
Ae=178mm2.
For a 360um gap, Al = 6.30E-07H/sp2

Np = 7 turns, Lp=3.09E-05H
Ns(400V) = 176 turns, Ls(400) = 1.95E-02H
Ns(18v) = 9 turns, Ls(18) = 5.10E-05H and real output voltage of ~ 19V -> linear regulators for 12 and 5V

I've attached a ltspice simulation for this thing
From the simulation it reaches equilibrium in ~ 100mS

[Richard Head]

ealex

An E42 core is massively oversized for 12W @50KHz. I suspect you could get away with something like an RM8 or RM10 @ 50khz.
In your design the flux density in the core is only 96mT. You can realistically go up to 280mT in current mode designs.
If small size is more important than efficiency then run at 100KHz. If efficiency is more important (ie battery powered) then run at a lower frequency such as 50KHz.
Selecting the core is an itterative exersize as there are quite a few options to consider. Normally you would want to use a core with a pre-ground airgap so you don't have to mess around with spacers. However the cores only come with two or three standard gap sizes so you have to calculate the flux density and number of turns for each type before deciding.
Also, definately stick to discontinuous inductor current or you end up with all sorts of problems.

[ealex]

I already have an E42 core available, including a coil former.
i have the tools to measure materials for the air gap, so making some shims is not a problem.
I might have a E32 core + coilformer - if i can find it, i'll try using it.

the entire project is a 1-off, so no space / power restrictions - i just want to make it work for now, and try to prepare some demos.

[TerraHertz]

Perhaps an LCD fluro backlight supply might be useful? Their output transformers are built with segmented secondary windings for low capacitance. Add a HV diode and cap, and feedback regulation, and you'd have a compact HV supply for a CRT?
Very easy to find them for free, since most street-toss LCD monitors are still fluro backlit.

[ealex]

i have one of those and i'm using it for a ccfl with a royer configuration.
i've read some of Jim Williams app notes on this topic, it seems ~ simple to add some feedback.
but they are so small that adding some extra coils for the other needed supplies will be impossible.
maybe if i use several of them - like one for 1.5kV, one for 400V and a simple one made on a ferite ring for the 12V rails

[BennVenn]

I have this same tube running as a scope clock on my desk! I made my own power supply by modifying a cheap ebay 150w 12v-240v inverter ($10) Stripped off the secondary, re-wound with a 400volt secondary (from memory - will check when i get home) which is trippled to get +800v and -400 giving a 1200v differential. Added a tap at 250v for the x/y drivers and a 12v+5v winding for the MCU+opamps.

A bit of switching noise on the 12v was causing issues with 'fuzz' on the text but some LC filtering just about cleared it all up.

Ill add pics soon

[GK]

For my prototype monitor build I just recycled one of my generic HV PSU boards. This design (not originally intended for powering CRT's) is a hybrid switching-linear regulator. A fixed frequency (20kHz) driven flyback feeds a Cockroft-Walton multiplier. The output voltage is regulated by a linear control loop with a series pass transistor (emitter follower) regulating the supply V to the switching convertor. It's not particularly complicated. The design of mine pictured below has an output voltage fully regulated and adjustable from 0V to 1500V. The incarnation that I used for my prototype 2AP1 CRT supply needed to be beefed up with a heftier series pass transistor as the original application drew less load current.

 

[ealex]

that looks nice.
i'll try building something once i have some free time - hopefully in the next 2-3 weeks ...