Announcement

Collapse
No announcement yet.

11" CRT high voltage supply

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • 11" CRT high voltage supply

    Hi all,
    turns out that I have two nice Asteroids PCBs to play with (one -02 and one -04). Now, I wanted to make a matching CRT display, so I picked up an old 11" TV (that was working, I tuned it to a DVD rf output and displayed some good movie).
    Here's the first big mistake: I didn't check the actual EHT supply before having ripped off all electronics. I have an XY driver board so my plan was rewiring the vertical yoke coils and make a new EHT supply using the original high voltage transformer of the old TV.
    The CRT is a Philips A28-14W and according to all the other brands datasheets (seems the Philips one has not been scanned by anyone), this tube runs at 11kV anode voltage.
    At first I've tried to run the original flyback circuit by removing the horizontal yoke connection and experimenting with different values of the flyback capacitor, but results were very poor, the driver transistor's collector and base waveforms were really terrible with lots of ringing, no matter how I would try to adjust drive frequency and duty cycle.
    Then I replaced the horizontal yoke with a similar inductance big coil (multi layer air coil, 1mm wire diameter) and that worked a lot better, the collector waveform has the usual double peak, no ringing at all and the peak flyback voltage can be tuned by tweaking the drive frequency and duty cycle. However the coil gets hot soon, so clearly that can't be a permanent solution.
    I am measuring both the G2 supply and the anode supply (with a 1Gohm HV probe).
    Now, before tearing the original TV chassis into parts, I had measured the G2 supply and that was 460V (with a displayed picture though).
    With the replacement coil, I can get easily the G2 supply to be 460V or even higher, however the anode supply can only reach 7.6kV

    So question for the experts: what could be the reason for nominal G2 supply but way less than "expected" anode supply (ok, I could have a bad transformer/rectifier, but the picture on the old TV looked normal to me).
    Second question: is there any proven circuit to generate the EHT supply using the original transformer?
    I've examined the Electrohome XY monitor schematics and they have a very clever regulated drive to the transformer, but that requires a separate small winding for the base feedback on the driver transistor and the usual TV transformers don't have that winding.
    The vectrex anode supply is instead (as far as I understand) using a normal TV flyback transformer, and my next attempt will be replicating that circuit, but of course I have a different flyback transformer and base-driver transformer, so I expect to need to tweak that circuit quite a lot.
    Another option I'm going to try is rigging up a current-mode PWM drive using an UC3843 based circuit. However, in this case I would get the voltage feedback from the G2 supply, so if that voltage doesn't track well the anode voltage (as it seem the case?) then I'm screwed again.

    So, what the experts suggest?
    Thanks!
    Frank

  • #2
    Ok, some recap on what I did (and failed), maybe one day someone else will be repeating such a conversion.

    First of all, the combined horizontal deflection and HV generation in the typical raster CRT based monitor/TV is designed, of course, around the needed horizontal scan timings.
    Removing the horizontal coil and trying to modify the existing circuit to behave has not given any useful result. It's probably because the energy stored in the horizontal deflection coils
    is really needed to be used by the HV transformer during the flyback to reach the correct voltage on the primary side.

    My second approach has been then to use a self-resonant push-pull drive, adding a new primary windings on some old BW CRT (11"-14" size) HV transformers.
    These are the old kind of TV transformer where the primary and G2/G4 windings are on one leg and the HV secondary (potted) is on the other leg.
    I've added initially 4+4 turns on top of the primary's leg windings. I've used 0.8mm diameter enameled wire. The winding is center tapped because of the push-pull drive.
    I've used a couple of IRF640 mosfets and 12V supply.
    I've measured the self-resonant frequency of all the HV transformers I was going to test, and always the self-resonant drive settled on a lower frequency with respect
    to the self resonant one of the given transformer.
    Results weren't bad, but also not usable:

    1) 29 KHz (self: 36 KHz), DC 8.5kV
    2) 25 KHz (self: 32 KHz), DC 6.6kV
    3) 42 KHz (self: 58 KHz), DC 5.8kV (this one sounds and smells like it's arcing internally)

    I then tried to reduce the primary windings to 3+3 on the first transformer and that killed it permanently. Probably a short into the HV secondary (best guess).
    This thaught me to always add a current limit to a drive circuit, which I did after this.
    With the current-limited drive and 3+3 turns on the transformer 2), I've got 8.5kV out, but the ferrite core gets too hot soon.
    I've then examined the cores and found one that has "N27" engraved on the ferrite, so that probably points to the TDK N27 material.
    This material can run at 0.5T saturation flux at 25 degrees celsius, so I estimated the core cross section and come out to a bit more than 6 turns per 12V at the frequencies
    these transformers are settling at. Infacts, on 4+4 turns, the cores get just a bit warm.

    So in the end, I'm still stuck far away from the needed 11kV DC, and I'll probably buy a ready made supply at this point.
    I've not really understood how they can obtain that output from these transformers given that I've run all these at almost the saturation flux and with decent waveforms obtaining
    roughly half of the rated output.
    One possible explanation would be that the HV diodes are too slow already at 25 KHz, but I kind of doubt that's the real case.
    Unfortunately I haven't found anyone that can discuss and explain what I'm missing.

    HTH
    Frank

    Comment


    • #3
      Hi Frank. I've been at this puzzle on and off for over a year for sure! I have no good news sorry. I found similar things to you. I found a 'dummy" yoke strategy problematic too as the coil I made got warm.. then warmer.. then hot. So not trying to dissuade anyone from trying this but I think it's beyond me. The frequency, values, coil sizes, resistances, impedances, all add up to a functioning stable system. Hey it good to see someone else had the idea too.

      I was using a portable b&w 12" TV tube. Dez has a PCB for sale on his site.

      www.amazingarcading.com.au
      http://www.amazingarcading.com.au/Am...ro_2016-01.jpg


      I've had hoped to marry the GO5-805 deflection PCB to it. Dez has the HV supply pcb for the Atari 19 & 25 inch Colour XY Display. Using this does has its considerations though. The voltages are much higher and need careful redesign. It may not actually be possible. It's a great pcb though for monitor. I just can't say for sure as I don't know. I'm thinking of using it to make a colour 19 inch. I'm still waiting for the LOPT. Dez has the other parts specific to that HV unit, namely, the drive transformer, the Magnetic correction 1 (MC1).The other parts you can probably source locally.

      You seam to have gone into it at a higher technical understanding than me so again, sorry I can't contribute. I did get get 11K out of my tv pcb with the dummy coil but, yes, it warms up. So close but no cigar. And yes I discovered current limiting to be of great benefit too doing this project. I've blown a lot of HOT's lol. And 3 old Colour TV flybacks, ( from hard rubbish), still it's sad when ya blow up one. Thinking about the life it could of had instead of at the hands of mad scientists.

      For me I'll be examining other TV circuits to see what I can bring to my design. Quite possibly I, and you maybe, didn't put in essential resistors and tuning coils. Who knows? Was a good idea. Seamed feasible when I started but not sure now.
      Last edited by taito; 23rd May 2020, 06:08 PM.

      Comment


      • #4
        Hi!
        Ok by now I'm becoming quite good with these kind of circuits.
        First, let's recap how the original horizontal yoke driver and flyback works, together with the HV transformer:

        1) the basic circuit is: B+ supply that goes into one primary connection of the HVT, another tap of this primary goes to the driver transistor collector, the transistor's emitter goes to ground. Damper diode (D) cathode is at the collector, anode to ground. The flyback capacitor (CF) is in parallel with the D. Yoke and S-correction capacitor (CS) is also in parallel
        with D, CS goes from yoke to ground. There're a lot of variations, there're the linearity coils, size coils etc, but the basic circuit is always this one.

        2) DC starting point: let's assume the transistor is off, then CF and CS are both charged to the B+ voltage (CS charged through the primary winding and the yoke).

        3) AC behaviour: for AC, the Yoke and the primary HVT winding are in parallel, as one side of the transformer and the yoke are joined together and the other sides are both at B+ (remember that CS is charged to B+)

        4) the transistor is turned ON, current starts flowing from both the primary winding and the yoke to ground through the driver, this is the "scan time", the raster goes from left to right, as this happens, both the transformer and the yoke are building a magnetic field. However, since the yoke inductance is much lower than the primary transformer inductance (like 1:5
        ratio as I measured), most of the current goes through the yoke (and most of the magnetic field is built on the yoke too).

        5) the transistor is turned OFF, the retrace or flyback begins. Now the transformer and yoke currents can't flow anymore through the transistor, so they MUST go into the flyback capacitor. This is now an LC series tuned circuit, that is free to oscillate and it's usually tuned to a quite higher frequency than the horizontal scan one. Like for example 80 KHz or more. In this phase, the energy from the Yoke and primary transformer gets transferred to the CF. It's 1/2 x L x I-squared. I is the final current reached in the yoke during the scan
        phase, if we equal that to the energy in a capacitor: 1/2 x C x V-squared, we can calculate how high will get the voltage in the flyback capacitor. In these two phases, there's still
        no HV generated as the current flow and the HVT phasing is contrary to the HV diode current flow.

        6) All the energy from the yoke and the primary winding has been transferred into the flyback capacitor, the collector voltage peaks and start going down, now the current flow is reversed and as in any LC resonant circuit, the energy begins flowing back from the capacitor to the L, the current now in the primary has the right phasing for the HV diode to conduct and the high voltage is generated with the primary winding seeing the much higher voltage reached on the flyback capacitor (much higher than B+).
        7) as the energy went back on the "L" side, the voltage on the capacitor wants to become negative, but at this point the damper dioded D start conduction and "dampens" the oscillation, the control circuit senses this condition through a feedback tap into the transformer usually. We can begin a new scan line.

        Now there're also some "secondary" aspects that I'm not describing, like third harmonic tuning on the flyback frequency... but that would add only confusion now.

        So, what happens when we simply remove the horizontal yoke from this?
        First, we lose almost all the energy storage during the scan phase, since the primary winding inductance is much higher than the yoke's one, the current through the primary will reach
        a much smaller value and the lower energy transferred to the flyback capacitor will make it reach a much smaller voltage than before.
        Second, we upsetted the original flyback resonant frequency, the flyback pulse is now much longer than before.
        Can we compensate for this two things?
        Maybe...
        First, we should measure as much precisely as we can the flyback pulse width, that's half the flyback resonance period (the other half is shorted out by the damper D), that is when the yoke is still connected. Then we should lower the flyback capacitor's value until we reach the same period without the yoke.
        At this point we still have not sufficient voltage building into the flyback capacitor since the energy stored into the primary alone during the scan period is still much lower than
        what it was stored in the yoke.
        The possible (I'm still trying some circuit configurations for this) cure for this problem is to lengthen the scan phase to allow more current to build into the much higher primary inductance.
        Seems easy?
        Well it isn't so easy. The problem now is ferrite core saturation. At some current level, the ferrite core of the transformer will be completely magnetized (saturates) and it won't build any more magnetic field, this causes a sudden drop of the inductance and a sudden rush of current that can burn our driver transistor.
        So we want to avoid saturation, this current level must be measured or calculated to stay a bit under the saturation current level.
        Then there's another problem: the primary winding wire size was calculated for the old primary current level, not the new one, so it will dissipate more heath and we are also wasting
        current in this way...
        Another possibility to get more current flowing (remember 1/2 x L x I-squared, so I makes really our energy level) is to increase the core gap of the transformer.
        These transformer are made by two C-shaped half cores, kept together with screws that go from one end to the other. Where the two half-cores meet, there're usually two very thin
        insulating sheets that make a tiny gap in the core, the gap decreases the inductance and increases the current level at which the core will saturate.
        So we could even try to make those gaps a tiny bit larger (just a tiny fraction of a mm would do probably) and increase our storable energy level before core saturation,
        but then primary wire size would be really the limiting factor (I have also rewound some primary winding, but it's a pain...).

        Ok, it's an ongoing experimentation really. If someone wants to join and help in developing a better method to modify and drive these old transformer, well, welcome!
        Remember that you're going to need a lot of time and quite some test equipment too.

        HTH
        Frank

        Comment


        • #5
          Hi Frank,

          Sounds like you're as keen to solve this as me. This is where I'm at in my thinking. We really just need to take working HV/horizontal deflection circuits and use them as they stand. The problem is the yoke. The yoke from the b&w TV is being used by the deflection PCB. So we need to make a proper dummy yoke. My original idea and theory was that if we can replicated the original Horizontal yoke then that would be it. Not an elegant solution of course and I had wanted to avoid this, having a large coil installed somewhere in the CRT cage, but never the less funtional. The signal to the H Driver Tranformer would come from an external source, a PCB with a XRay protection feedback circuit too like the one I built. So the question is how to create the H coil.

          A few things came up.

          Resistance on the original coil was 1.3 Ohms
          The impedance 400uH

          Some things I'm assuming.
          The more impedance the less heat. For example, I increased the impedance by inserting a ferite core into the coil while running revealing the HV dropped. Less heat but lower voltage. I measured 1.3 ohms of wire to begin with and formed a coil that was 400uH, but it got hot. But this was an air coil. So I think its the ferite material and the shape also possibly that will find the sweet spot. HV and low heat/energy dissipation. If I do that, we might find people trying to build there own XY monitors. Well there is actually a lot more to it than cracking the HV puzzle. Tubes need matching identifying parts sourced. It's not cheap. I gave up the smokes 20 years ago, so that's my reasoning, justification.

          Comment


          • #6
            Hi Again,
            the problem with a replacement yoke is twofold:
            1) impedance needs to be correct. I'm not saying inductance and/or resistance. It really needs to be the same yoke: it must be the correct inductance to store the correct field, it must have the correct resistance to dump the correct amount of energy into the flyback capacitor (and not dissipate it in heat), it must have the correct self-capacitance not to upset the flyback pulse resonance.
            2) it's big and makes a big magnetic field and we don't want it near our CRT.... Unless we can came up with a suitable "cored" replacement (read point 1 again if you think it's easy).

            So, I'm trying to work with a schematic that either doesn't need the exact yoke replacement or can work well enough without one.

            Also there's another fact: the Vectrex console, I'm sure everyone here knows it, uses a standard TV transformer and the engineers settled with an amazingly low 5.8kV anode supply for the 9" CRT.
            Now, two of the original Vectrex engineers left this planet and the lead engineer seems living in his car nowadays, so I can't easily ask for confirmations, but I have made some
            educated assumptions based on the Vectrex schematics.

            Facts:
            1) the Vectrex uses 5.8kV anode supply, most of the 9" BW CRTs that I could find the datasheet for, need a nominal 10kV (that's what I measured for example in my 9" PET monitor).
            2) It uses audio (slow!) amplifiers for XY deflection
            3) It seems to be using a regular TV transformer for HV and K/G1/G2/G4 supplies.
            4) I couldn't find a datasheet for the Samsung CRT they indicate in the schematic.

            Assumptions:
            The CRT wasn't really designed to work with that low anode supply, but it appears to work fine with it (and different K/G1/G2/G4 levels, until one gets the proper cutoff and the proper focusing).
            The HV transformer drive is a bit too complicated, it seems some parts have been thrown in to correct issues (like base drive, primary waveform, unwanted oscillations...)
            The reduced HV allows for a much easier XY deflection (this is a fact actually), so they could get away with slow audio amplifiers to drive the yoke, as they couldn't get
            the "datasheet" HV value anyway with that transformer.

            So I already have a working supply for my 11" CRT with 7kV anode (regulated, settable to up to 7.3kV with the current transformer/drive), 580V G2/G4 (settable indipendently on G2 and G4), -/+120V for K/G1 (still working on the K/G1 drive currently). Of course it will need a lot less than 120V cutoff probably but that's what I could get out of a tap on the transformer, so I'll probably regulate down with zeners or something when I confirm I can focus and cutoff the tube with the "new" voltages.
            This is using an easy self-resonant-push-pull drive with voltage and current feedback, but of course the transformer isn't getting the required flyback pulse so they normally settle at
            about 60-70% of the original HV.

            I'm still experimenting with a proper "flyback" pulse generator circuit, but you know, life gets always in the middle.

            Frank

            Comment


            • #7
              Click image for larger version  Name:	amplifone_hv.jpeg Views:	0 Size:	92.0 KB ID:	2134677
              Yeah, I think moving away from that design, with the dummy yoke, is probably the way to go.

              So if anyone was interested, I used this circuit as a basis for my design. (I've not yet got it working properly)

              It comes from the Atari technical manual 239. TM-239 and is the manual for the Atari 19" and 25" colour xy display manufactured by Amplifone.

              If anyone is following this post and you have got this far, and your brain hasn't fried, I'd say you're doing well. My brain fried on this project, along with quite a few HOT's a while ago.

              Comment


              • #8
                Originally posted by IZ8DWF View Post
                Hi all,
                turns out that I have two nice Asteroids PCBs to play with (one -02 and one -04). Now, I wanted to make a matching CRT display, so I picked up an old 11" TV (that was working, I tuned it to a DVD rf output and displayed some good movie).
                Here's the first big mistake: I didn't check the actual EHT supply before having ripped off all electronics. I have an XY driver board so my plan was rewiring the vertical yoke coils and make a new EHT supply using the original high voltage transformer of the old TV.
                The CRT is a Philips A28-14W and according to all the other brands datasheets (seems the Philips one has not been scanned by anyone), this tube runs at 11kV anode voltage.
                At first I've tried to run the original flyback circuit by removing the horizontal yoke connection and experimenting with different values of the flyback capacitor, but results were very poor, the driver transistor's collector and base waveforms were really terrible with lots of ringing, no matter how I would try to adjust drive frequency and duty cycle.
                Then I replaced the horizontal yoke with a similar inductance big coil (multi layer air coil, 1mm wire diameter) and that worked a lot better, the collector waveform has the usual double peak, no ringing at all and the peak flyback voltage can be tuned by tweaking the drive frequency and duty cycle. However the coil gets hot soon, so clearly that can't be a permanent solution.
                I am measuring both the G2 supply and the anode supply (with a 1Gohm HV probe).
                Now, before tearing the original TV chassis into parts, I had measured the G2 supply and that was 460V (with a displayed picture though).
                With the replacement coil, I can get easily the G2 supply to be 460V or even higher, however the anode supply can only reach 7.6kV

                So question for the experts: what could be the reason for nominal G2 supply but way less than "expected" anode supply (ok, I could have a bad transformer/rectifier, but the picture on the old TV looked normal to me).
                Second question: is there any proven circuit to generate the EHT supply using the original transformer?
                I've examined the Electrohome XY monitor schematics and they have a very clever regulated drive to the transformer, but that requires a separate small winding for the base feedback on the driver transistor and the usual TV transformers don't have that winding.
                The vectrex anode supply is instead (as far as I understand) using a normal TV flyback transformer, and my next attempt will be replicating that circuit, but of course I have a different flyback transformer and base-driver transformer, so I expect to need to tweak that circuit quite a lot.
                Another option I'm going to try is rigging up a current-mode PWM drive using an UC3843 based circuit. However, in this case I would get the voltage feedback from the G2 supply, so if that voltage doesn't track well the anode voltage (as it seem the case?) then I'm screwed again.

                So, what the experts suggest?
                Thanks!
                Frank
                Can you get the HV monitor boards from an asteroids machine? And an Arcade B&W screen?
                Regards

                Dez

                DSB Electrical
                DSB Electrical Facebook
                VAPS
                Vector Monitor Repro Parts
                Amazing Arcading Facebook Page

                Another AA Visitor. Stay a while, stay FOREVER!

                Comment

                Users Viewing Topic: 0 members and 1 (guests)
                Working...
                X