UD2.7 dual Purpose Use

[FonziDaytona]

Hi all,

Not sure if this is the right place to ask, but figured it might be seeing as the intended purpose is for DRSSTC :-)

I want to use my UD2.7 as a closed loop driver for a half bridge meant to drive a ferrite transformer. I know that I can input a signal from a frequency generator into the feedback input. Is there a simple, non invasive, way to get the driver to oscillate like this? Is there any chance of damage from using it like this?

I know there are better ways to do this, but like the idea of dual purpose uses for everything and a DRSSTC driver (minus the MMC) looks promising for using with my high power ferrite transformers…

-Matt

[Mike]

Two relatively straight forward answers I can think of.

1) You can drive a flyback with a half-bridge (or full bridge) using a GDT and current transformer, just like you would a tesla coil primary. You can even make it resonant with a series cap (or your MMC) if you're keen.

2) It's possible to make the UD2.7 self oscillate as Dave describes here: https://highvoltageforum.net/index.php?topic=1373.msg10197#msg10197. Even without feedback this would work, but you need to manually set the correct frequency and have no over current protection so I don't really recommend it unless it's used in combination with 1).

[davekni]

1) You can drive a flyback with a half-bridge (or full bridge) using a GDT and current transformer, just like you would a tesla coil primary. You can even make it resonant with a series cap (or your MMC) if you're keen.

I believe this works only if resonant.

2) It's possible to make the UD2.7 self oscillate as Dave describes here: https://highvoltageforum.net/index.php?topic=1373.msg10197#msg10197. Even without feedback this would work, but you need to manually set the correct frequency and have no over current protection so I don't really recommend it unless it's used in combination with 1).

Self-oscillating mod to UD2.7 would be useful if wanting to avoid a frequency generator in non-resonant operation.  For resonant operation, self-oscillation can be useful to ensure startup.  Self-oscillation is not necessary if standard UD2.7 startup conditions are met (first half-cycle generates enough feedback CT signal to start next half-cycle).

Connecting a frequency generator to UD2.7 feedback input should work fine for non-resonant half-bridge drive.  Enable input fiber needs to have continuous light, or a jumper added to force enable (pretend fiber receiver is seeing light).  UD2.7 feedback input is low impedance (51 ohms).  Signal generator needs to have reasonable drive capability, or phase lead jumper SV1 needs to be open when using signal generator.  (Make sure jumper is returned before TC use.)

OCD can be used for any of above options.  UD2.7 has a separate OCD input.  Connecting a current transformer from half-bridge output to OCD input provides protection whether using feedback CT or not.  Resetting an OCD event will require disabling (removing light from fiber or jumper that forces enable).

I want to use my UD2.7 as a closed loop driver for a half bridge meant to drive a ferrite transformer.

When you say "closed loop", do you mean using feed CT as in (1) from Mike?  If so, ferrite transformer needs a series resonant capacitor, which could be the two divider capacitors in typical half-bridge circuits.

[Mike]

I believe this works only if resonant.

Agreed, my statement was misleading.

Self-oscillating mod to UD2.7 would be useful if wanting to avoid a frequency generator in non-resonant operation.  For resonant operation, self-oscillation can be useful to ensure startup.  Self-oscillation is not necessary if standard UD2.7 startup conditions are met (first half-cycle generates enough feedback CT signal to start next half-cycle).

Agreed again, I was specifically thinking about the non resonant mode without a feedback transformer or function generator. You're right about the OCD as well, my assumption was that with no FB CT you also wouldn't have an OCD CT, but as you point out, these are independent.

[FonziDaytona]

Thanks for the replies, guys!

By closed loop, I meant “no feedback”. I probably have that wrong…

I’m thinking that I can still use the Interruptor input to control current by adjusting the on-time of the pulses, no?

Glad that this looks like a very workable idea. I didn’t want to have to build something else when I can just use this for both my fusor power supply and my upcoming DRSSTC build. I’ll look into the self oscillation mode so I don’t have to keep my function generator tethered to it forever…Probably also look into resonant mode down the road to boost performance a little. Right now I’m planning to keep frequency around 30khz with my CW multiplier diodes in mind…

Meant to add that I’ve already connected this to my function generator and Interruptor to test since I bought this used. I was able to see the Interruptor square wave single on my scope so it would seem that I should be about ready to connect my half bridge (cm300dy) and see what happens.


-Matt

[davekni]

I’m thinking that I can still use the Interruptor input to control current by adjusting the on-time of the pulses, no?

Yes, interrupter duty cycle should be good for power control.

Right now I’m planning to keep frequency around 30khz with my CW multiplier diodes in mind…

Good idea.  Also intra-winding capacitance of transformer secondary should be more benign at lower frequency.

[FonziDaytona]

Time for an update:

Lashed it all up on a board and did some testing. I’m having a little trouble.

The output on the flyback that I’m using for testing is very spindly (low current) and noisy. I’m running over 25khz, so there shouldn’t be a lot of noise. I’ve tried using both my MIDI controller into the interrupt input as well as a bright flashlight. Both produce long, spindly, very noisy arcs from the flyback. It’s as if the frequency is adjusting down to the audible frequencies rather than the input frequencies.

Here’s a shot of the GDT outputs before applying power to the half bridge and an overall shot of the setup.

[davekni]

Scope traces of actual operation would help.  Do you have a differential probe?  If not, a turn or two of wire tightly around the center of the flyback primary winding with twisted lead wires will show a reasonable approximation of scaled primary voltage.  If the AC line power input is connected with neutral to voltage-doubler center tap (and hot to diodes), then scoping half-bridge output with a normal probe can work too.  Either leave scope ground probe open or connect through a Y-rated capacitor to neutral for improved signal quality.

What is the white capacitor with one lead against heatsink?

Many half-bridge designs use two film capacitors to split Vbus to one primary winding lead.  With your direct connection to bulk capacitor center-tap, voltage-doubler ripple will induce ripple across primary coil.  That line-frequency ripple may be saturating flyback core.  (Is the core gapped?  If not, saturation is highly likely with this connection.)

Are you sure 25kHz is well below flyback secondary resonant frequency?  Do small frequency changes help at all?  Possible that half-bridge frequency or some harmonic of half-bridge frequency is matching a flyback resonant frequency.

[FonziDaytona]

Hi Dave,

I’m using an isolation transformer between a variac and the half bridge rather than “hot” and “neutral”. I’ll wrap a turn around the primary of the flyback and scope that to see if that helps. First, I’ll use some poly capacitors across the main storage capacitors voltage doubler to see if that works.

Small adjustments in frequency don’t help. When I connect this to my audio amp (using function generator for frequency input) the flyback acts as expected throughout the range of frequencies that I’ve tried.

Here are a few more details in case something pops out:

Power to the ud2.7 is a Vicor flatpac 24vdc and 4A. Seems sufficient.
Half bridge uses a cm300dy-12nf IGBT. Maybe this is too large for the smallish GDT and the GDT is saturated?
Interruptor maxes at 150hz and 2.5mS. Maybe this just isn’t enough to get much current flowing? It was originally designed for Tesla coils and MIDI playback, not lots of power.

Maybe shining a light directly into the ud2.7 interrupt input just isn’t to imitate a constant high signal?

-Matt

[FonziDaytona]

Here is the waveform for a single turn over the flyback primary. I added a 330nF capacitor across each of the voltage doubler capacitors. Seems to have had a negligible effect, but 330nF might not be enough?

[davekni]

First, I’ll use some poly capacitors across the main storage capacitors voltage doubler to see if that works.

My suggestion was to use separate poly capacitors from Vbus- and from Vbus+ to one primary coil lead, not to parallel with bulk caps.  Paralleling doesn't fix possible ripple voltage issue.  Separate caps isolate primary coil from low-frequency ripple.

Interruptor maxes at 150hz and 2.5mS. Maybe this just isn’t enough to get much current flowing? It was originally designed for Tesla coils and MIDI playback, not lots of power.

Of course, when interrupting, 150Hz sound would be expected.

Maybe shining a light directly into the ud2.7 interrupt input just isn’t to imitate a constant high signal?

I'd guess light would be fine, presuming it is bright enough.  You could make a separate test with unpowered bridge, just scoping optical receiver output or GDT input or output while using the light.  See if you get steady operation.

Here is the waveform for a single turn over the flyback primary. I added a 33nF capacitor across each of the voltage doubler capacitors. Seems to have had a negligible effect, but 33nF might not be enough?

This capture shows no core saturation issue, at least not at the moment shown in the capture.  If all the captures look like this, I think your half-bridge is working properly.  Perhaps there is some issue with secondary or voltage multiplier?
I presume the spikes are the random sparks from your secondary that couple into scope connection.  Surprisingly frequent for sparks, however.  Any chance of a bad connection somewhere in secondary circuit?  Posting more details of your secondary circuit and spark gap may help.  Are you testing with your voltage multiplier connected, or just sparks directly from secondary winding?
Also, how many turns in primary winding?  Scope capture appears to be at low bus voltage.  What is the voltage?  Is this voltage your normal operating value where you get the spindly low-current noisy sparks?

[FonziDaytona]

Okay so, do you mean like this? (Attached).


Here is a shot of the GDT with a flashlight instead of fiber from the Interruptor. Seems wonky, not as expected.


Previous scope shot of the single turn primary is with the flyback producing corona in air. I suspect those spikes are from that. I haven’t connected my CW multiplier. BUS voltage (5vac into the doubler) is really low as the flyback secondary output is already getting unruly (close to 1in arcs).

I’ve got 13turns on primary. I probably need to increase this.

[davekni]

kay so, do you mean like this? (Attached).

No, that would fry FETs.  Here is a typical half-bridge circuit from Mads:
https://kaizerpowerelectronics.dk/wp-content/gallery/2009_07_22_-_Kaizer_SSTC_III/KaizerSSTCIIIbridge.gif
It happens to be fed from a full-wave line voltage rectifier.  Just replace that with your voltage doubler including your pair of bulk caps instead of C12 in above schematic.  C8 and C9 are the new caps to add, with no connection to bulk cap center-point.  330nF may be fine for your C8 and C9.  They need to be large enough so that resonant frequency with primary inductor is well below operating frequency.  (Unless you want to build a resonant converter.)
Of course, if primary current capability is high enough and voltage ripple low enough, you may be OK with existing connection to bulk cap center.

Here is a shot of the GDT with a flashlight instead of fiber from the Interruptor. Seems wonky, not as expected.

Agree.  Interrupter isn't remaining on continuously with flashlight input.  Any chance this is an LED flashlight with pulse-width brightness adjustment?  Or an internally AC-coupled optical receiver within UD2.7?  Adding a jumper to UD2.7 to force optical receiver input true is probably the easiest and most reliable fix.

Previous scope shot of the single turn primary is with the flyback producing corona in air. I suspect those spikes are from that. I haven’t connected my CW multiplier. BUS voltage (5vac into the doubler) is really low as the flyback secondary output is already getting unruly (close to 1in arcs).

I’ve got 13turns on primary. I probably need to increase this.

Yes, if you are getting reasonable output voltage with 5Vac line input, more turns seems appropriate.  Do you have a design target for this?  Do you know secondary turn count, DC resistance, voltage capability, ferrite core material and geometry (or measured saturation flux), inductance/turn^2, etc.?

[FonziDaytona]

Ahh, the led light is PWM. That would explain it….

-Matt

[FonziDaytona]

Updates:

I found a light that isn’t PWM to use. This definitely improved the performance. The arc is relatively silent, purple rather than loud and crackly. Frequency matches the input from the function generator.

I fixed the capacitor issue using the 330nF capacitors. This decreased performance significantly. Frequency still matched the input and the arc was silent, just less energetic. I’ve ordered some 2.5uF poly capacitors to try out.

Any way to easily modify this board to jumper the Interruptor input? The connections are so tiny that I’m afraid I’ll ruin something. It might be easier to just modify the MIDI Interruptor for a much higher frequency and on time?

Looking over the schematic, it seems that I might be able to attach a removable jumper between pin 6-7 of the ST fiber receiver (RX1). This should ground the output that is then inverted, no?





-Matt

[davekni]

Looking over the schematic, it seems that I might be able to attach a removable jumper between pin 6-7 of the ST fiber receiver (RX1). This should ground the output that is then inverted, no?

Yes, that should work fine.  Exactly the sort of patch I was suggesting.

It might be easier to just modify the MIDI Interruptor for a much higher frequency and on time?

That might be easy too.  I'm not familiar with your particular interrupter.

[FonziDaytona]

I managed to install a jumper to force the input Interruptor “on”. This works, but I have found that OCD seems to be triggered on startup. I have to pull the OCD LED and reinstall before I can operate. There isn’t any OCD input. Do I need to have something there to prevent this?

-Matt

[davekni]

There isn’t any OCD input. Do I need to have something there to prevent this?

Measure the voltage at TP1.  If R20 is adjusted to a very small voltage, then even tiny bits of noise will trip OCD.  If TP1 voltage is not close to zero, then there presumably is some other fault in the circuit, such as momentary supply voltage drop tripping UVLO.

[FonziDaytona]

Looks like it’s set high (7.6v).

In order to reset, I have to plug in the Signal LED and then unplug it. This causes the OCD led to go out and I am able to operate.

-Matt

[davekni]

Looks like it’s set high (7.6v).

So not that reason.  Leaves UVLO or some misbehavior of UD2.7 board.

In order to reset, I have to plug in the Signal LED and then unplug it. This causes the OCD led to go out and I am able to operate.

I presume "Signal LED" refers to the LED indicating enable (LED-2 pin connected to R9).

Suddenly makes sense.  Yes, UVLO is set on power-up until DC power input reaches threshold voltage, which sets OCD FF.  In normal UD2.7 operation, the next rising edge of enable input resets OCD FF.  Connecting LED makes a momentary glitch on enable signal due to LED capacitance or a bit of ESD from your fingers.  That glitch is enough to clock OCD FF back high.
Even if UVLO is disabled, there is no guarantee which state OCD FF will be at power-up.

If it isn't too much bother to have a push-button for startup, you could use a normally-closed momentary switch instead of a jumper for enable.  Then pushing the switch will make a momentary disable and re-enable event, clocking OCD FF.