Glassman HV Supply / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Glassman HV Supply

first previous 5 6 7 8 next

Move Thread

LAN_403

GeordieBoy

Thu Dec 09 2010, 10:59PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

Steve C is right. Flyback converters are pretty rough, inefficient and electrically noisey in the broad spectrum of SMPSU topologies. However, they are cheap and therefore quite common at the low power level for things like phone or shaver chargers and in small bits of IT & consumer kit. Not requiring a buck choke to smooth the output allows a compact and lightweight design, and you can easily make it output several semi-regulated voltages simultaneously just by adding more secondary windings

As a general guide:

Hard switched Discontinuous current mode flyback is normally used up to about 60W of power throughput. Above this the Continuous-mode flyback converter offers the advantage of lower peak currents, at the expense of dealing with a right-half-plane zero and forced reverse recovery of the output side rectifer. Even still the continuous current mode can be viable up to about 150W.

The main problem for the flyback converter is leakage inductance in the flyback transformer. Most practical transformers that meet approvals requirements have something like 1-5% leakage inductance refferred to the primary side. All of the current that ramps up in this leakage inductance part of the primary during the transistor on-time is trapped energy which is not transferred to the secondary side during the flyback phase. This is normally dissipated in an RC or zener clamp to limit the peak turn-off voltage developed across the switch. Above about 150W the losses in this snubber become unmanageable in a modern compact SMPSU, and it's time for a change of topology...

The single-switch forward converter, dual-switch forward (assymmertric half-bridge), and conventional half-bridge all give better efficiency, at the cost and weight expense of requiring a buck-choke to smooth the output current. What's more they are far more tolerant of leakage inductance in the transformer. In the forward derived converters leakage inductance in the transformer mearly delays the transfer of energy from primary to secondary so unless excessive leakage inductance usually does not pose a big problem.

The "flyback" circuit used in TV's is better described as a quasi-resonant converter because the primary of the flyback transformer actually parallel resonates with a snubber capacitor across the switch at some high frequency during the switch's off time.

As far as I know those Glassman units use either a Series Resonant toplogy or a Parallel Resonant topology. Both are very efficient modern soft-switching schemes and can be designed to be s/c tolerant.

The SLR converter can be controlled proportionaly using the "pulse thinning out" method. The rate at which power pulses are delivered to the resonant transformer is controlled whilst keeping the pulse on-time constant because this is determined by the characteristics of the resonant network. For full output current you deliver one positive pulse, then one negative, then one positive etc, each right after the previous one. For less power you simply wait some time between pulses, but still repeating the +-+-+- pattern to prevent transformer flux walk. The downside of this control method is that the switching frequency decreases under light load. This means that low load ripple is not as good as it would be for a fixed frequency converter.

-Richie,

Pinky's Brain

Fri Dec 10 2010, 02:11AM

Registered Member #2901 Joined: Thu Jun 03 2010, 01:25PM
Location:
Posts: 837

GeordieBoy wrote ...
Above about 150W the losses in this snubber become unmanageable in a modern compact SMPSU, and it's time for a change of topology...

The single-switch forward converter, dual-switch forward

Dual switch is equally applicable to flyback (the circuits are nigh identical except for the secondary side) and solves the primary side leakage problem quite elegantly.

GeordieBoy

Fri Dec 10 2010, 10:31AM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

The problem with dual-switch flyback is that the reflected voltage always needs to be less than the DC bus voltage. If this is not the case then the primary-side clamp diodes are driven into conduction when the switches turn off and the output voltage will be clamped returning energy to the DC bus capacitor instead of to the output side.

This prevents you from being able to design an efficient step-up flyback converter where you would typically want to achieve some portion of the voltage rise with flyback action and the remaining portion with pure turns ratio. If the maximum reflected voltage is only as much as the DC bus voltage then you might as well just use a forward converter in the first place as all of the voltage gain has to come from the turns ratio alone. And you benefit from lower peak currents, better efficiency and lower conducted & radiated EMI with the forward topology too.

-Richie,

Pinky's Brain

Fri Dec 10 2010, 02:56PM

Registered Member #2901 Joined: Thu Jun 03 2010, 01:25PM
Location:
Posts: 837

AFAICS a forward converter will switch off the current hard, just like the flyback, and has one more diode to snap off. The airgap in the transformer will throw out EMI, so will the one in the output inductor for the forward converter. I don't see why there would be much between them EMI wise if both use dual switches.

AFAICS the only significant efficiency loss compared a forward converter comes from the slightly higher primary currents (in a continuous mode flyback the extra current is restricted to primary leakage). I'd be surprised if that even came up to 1% though ... 90% efficiency for a hard switched 1.5 KW converter really doesn't seem that bad.

radhoo

Fri Dec 10 2010, 05:30PM

Registered Member #1938 Joined: Sun Jan 25 2009, 12:44PM
Location: Romania
Posts: 701

I appreciate your feedback, and the discussion on what is the best driver is quite interesting.

However, given my not-so-advanced electronics skills/knowledge, my questions were more related to a more practical approach. You gave me lots of ideas with pros and cons, but unfortunately I don't have the experience to pick and use one.

I need to drive the flyback in the schematics I draw, use the divider to keep the output constant under load, adjust the output using a pot, add some measurement meters: u/mA and kV, and some basic protection so I wouldn't break it for the simplest mistake.
If you can point me the right direction for that it would be great. I don't have the time to learn everything theoretically so the step=by=step approach would be optimal.

Updates:
I got one driver board!

Remember, the schematics for the multiplier look like this:

Do you think I have any chance of using this original driver?

I have a negative Glassman multiplier and its twin positive multiplier each for 60KV so I could use them together for a 120KV potential difference. But this implies I need my own driver. Any hints? .

GeordieBoy

Sat Dec 11 2010, 11:39AM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

> The airgap in the transformer will throw out EMI, so will the one in the output inductor for the forward converter. I don't see why there would be much between them EMI wise if both use dual switches.

Input and output current for flyback converter are both discontinuous. Heavy pulsing input and output currents --> Harder to quieten EMI. Forward converter has continuous output current with typically 10% ripple at full load. Smooth output current with little triangular ripple ---> Much lower EMI problems on output side.

Gapped flyback transformer core can be source of interference but is easily tamed with flux band. Output inductor of forward (buck) converter is usually an iron-powder toroid with distributed air-gap so leakage field is much less than gapped ferrite or stick inductors!

> AFAICS the only significant efficiency loss compared a forward converter comes from the slightly higher primary currents (in a continuous mode flyback the extra current is restricted to primary leakage). I'd be surprised if that even came up to 1% though ... 90% efficiency for a hard switched 1.5 KW converter really doesn't seem that bad.

Try doing some calculations for a 1.5kW half-bridge converter and theoretical 1.5kW flyback converter including conduction losses, switching losses and diode recovery energy with a typical real world transformer model. You will quickly see why no commercial SMPSU above about 150W would use the flyback topology.

-Richie,

Pinky's Brain

Sat Dec 11 2010, 01:26PM

Registered Member #2901 Joined: Thu Jun 03 2010, 01:25PM
Location:
Posts: 837

GeordieBoy wrote ...
Input and output current for flyback converter are both discontinuous.

At the transformer the secondary currents are also discontinuous for the forward converter. A diode takes over when it turns off, but the capacitor in the flyback does the same thing. Although in principle the flyback can actually have a continuous secondary current ... something the forward converter can not do (it can have a continuous primary current).

wrote ...
Try doing some calculations for a 1.5kW half-bridge converter

I'm sure there are in practice better designs ... I'm just trying to unite what Glassman said in their tech page with the transformer on the board. There is no output inductor, no center taps, resonant circuits aren't fixed frequency PWM ... and on top of that it's zero current switch on.

Wolfram

Sat Dec 11 2010, 02:06PM

Registered Member #33 Joined: Sat Feb 04 2006, 01:31PM
Location: Norway
Posts: 971

Pinky's Brain wrote ...

I'm sure there are in practice better designs ... I'm just trying to unite what Glassman said in their tech page with the transformer on the board. There is no output inductor, no center taps, resonant circuits aren't fixed frequency PWM.

Why can't it be the SLR topology, like Finn suggested two pages ago? That seems a lot more reasonable to me.

Anders M.

Pinky's Brain

Sat Dec 11 2010, 02:30PM

Registered Member #2901 Joined: Thu Jun 03 2010, 01:25PM
Location:
Posts: 837

Even if you try to keep the switching near a given frequency by modulating the supply rails a change in load is still going to tear it out of sync ... only a discontinuous converter can really hope to maintain fixed frequency and zero current turn on.

GeordieBoy

Sat Dec 11 2010, 03:28PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

> At the transformer the secondary currents are also discontinuous for the forward converter.

Yep, they are for all switched-mode converters. That's where the switching bit comes from.

> A diode takes over when it turns off, but the capacitor in the flyback does the same thing.

The output capacitor in a "high-powered" flyback converter gets absolutely hammered though. It has to supply the full DC output current of the supply during the time that the primary side switch is conducting, then be fully recharged during the brief flyback period. Peak currents seen by Cout are many times the actually DC output current! Read: Heating, short C life, and excessive output ripple voltage.

> Although in principle the flyback can actually have a continuous secondary current...

Can it!?!? A flyback transformer can have a continuous flux in the air gap, but the current always switches between windings to transfer energy to the output.

> ...something the forward converter can not do (it can have a continuous primary current).

Can it?!?! I thought you had to chop the primary current to make it go through the ferrite transformer?

-Richie,

first previous 5 6 7 8 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob