Questions about voltage doubling

[Paul Andrews]

Hi,

I would like to add a voltage doubler and a -ve voltage output to my power supply. I modified my LTSpice simulation by following the schematic at https://threeneurons.wordpress.com/nixie-power-supply/ (Adding Multiple Voltages) to add the extra output stages, but I am not getting the expected results. One difference is that I am using a flyback transformer rather than an inductor. It seems that the un-rectified output from the transformer goes negative, and this additional voltage is being added to the output. Here is the output part of the schematic (sorry it is a bit chaotic):

Here are the waveforms for out, outdouble, outneg and the PS1 (the cyan trace) :

• Vout is around 148V.
• The cyan trace goes to -62V.
  
• Outneg is -205V. If I subtract the -62V, then I get -143V (i.e. close to -148V).
  
• Outdouble is 351V. If I add the -62V I get 289V (i.e. roughly double the 148V).

Is there something wrong with how I am trying to apply what threeneurons shows? Is this just how it is with flyback power supplies? Is there something amiss with my simulation?

I don't now where to start to think about this, so any help would be very appreciated.

Thanks - Paul

[gregebert]

Actually, you dont need a doubler on a flyback topology, because the output voltage will increase to whatever value is necessary to dump the energy stored in the inductor (transformer).. Just be careful not to exceed the voltage rating (insulation) on the transformer.

To a first-order, whether you use an inductor or a transformer in a flyback converter is irrelevant. A step-up transformer will get a higher output voltage with less kick-back at the switching device.

To implement a doubler (or higher multiplier) you will need a push-pull arrangement

[gregebert]

If you want to do a doubler, use a CCFL transformer. The primary is center-tapped, so that is connected to the supply voltage, Then the other legs on the primary are alternately switched to GND.

Expect to spend a lot of time fiddling with the design to get decent efficiency and acceptable peak current.

I'm using this approach for my NIMO clock.

[gregebert]

A bit more commentary on the flyback converter here....when you see the PS1 transformer pin go negative, it's probably doing so when the MOSFET turns on, correct ?

That's normal transformer behavior, but I will caution you that  if you harvest energy from it (during the ON cycle) there will be less energy available for the positive supply during the OFF cycle. It doesn't cause any harm, but it will cause the circuit to behave differently versus generic equations for a flyback converter. The other thing to know is that the negative voltage you are using is fundamentally limited by the supply voltage and the turns-ratio of the transformer, whereas the output of the flyback transformer on the positive cycle is limited by the feedback-control.

Things get real interesting when the transformer starts to oscillate with the filter capacitor. They get worse with higher turns-ratio as secondary inductance is proportional to number-of-turns-squared.

Be sure to play around with voltage, frequency, duty-cycle, rise & fall-times for the gate of the NMOS driver, and parasitics. I guarantee you will learn a lot from running a lot of simulations, and it will prepare you for reality when you bench-test it. These things always behave so much worse under load.

[Paul Andrews]

On Friday, January 5, 2018 at 8:01:29 PM UTC-5, gregebert wrote:

A bit more commentary on the flyback converter here....when you see the PS1 transformer pin go negative, it's probably doing so when the MOSFET turns on, correct ?

Yes

 

That's normal transformer behavior, but I will caution you that  if you harvest energy from it (during the ON cycle) there will be less energy available for the positive supply during the OFF cycle. It doesn't cause any harm, but it will cause the circuit to behave differently versus generic equations for a flyback converter. The other thing to know is that the negative voltage you are using is fundamentally limited by the supply voltage and the turns-ratio of the transformer, whereas the output of the flyback transformer on the positive cycle is limited by the feedback-control.

That makes sense, and is about what I am seeing in the simulation. This is a model of some actual hardware I have built, I was just intrigued about the whole voltage doubling thing and figured I would see what would happen in my simulation.

 

Things get real interesting when the transformer starts to oscillate with the filter capacitor. They get worse with higher turns-ratio as secondary inductance is proportional to number-of-turns-squared.

And by 'interesting' you mean? I have the usual ringing on the input side caused by parasitics on the primary/fet capacitance, and the secondary parasitics(?) and the filter capacitor (I assume). I dampen the primary ringing but stopped short of adding a diode to cap it. I played around in the simulation with damping the secondary ringing, but nothing much worked and I'm not sure what the effect would be on performance with or without addressing it.

Be sure to play around with voltage, frequency, duty-cycle, rise & fall-times for the gate of the NMOS driver, and parasitics. I guarantee you will learn a lot from running a lot of simulations, and it will prepare you for reality when you bench-test it. These things always behave so much worse under load.

I spent 3 or 4 months playing with the simulation, then about another month with a prototype and another couple of weeks with actual boards. I experimented with just about every component on there! I had to force myself to stop and move on to actually using it, but I have some slack time so I thought I would fiddle with it some more :-). In particular I've found some transformers that I think might be better (this one, and this one). Not sure when I'll get around to trying them out though...

[gregebert]

With my converter, which isn't built yet, the filter-cap was resonating with the inductance of the secondary of the transformer near my target switching speed. Increasing the filter-cap value to reduce the resonant frequency increased the peak current in the transformer, and reducing it increased the ripple-voltage as well as increased the ring-freq. Reducing the switching frequency to get below resonance adds ripple and requires a higher on-time. But that increases current in the primary beyond Isat. Increasing the switching frequency adds more switching loss, and also the laminated core is not meant for frequencies above 100Khz. So I'm boxed-in. And that's where things get interesting.....

The ringing also contributes to the transformer's primary current, and is pushing it beyond Isat.

Adding resistance in the secondary circuit to reduce the ringing (ie, reduce the Q) was not working very well and not only started wasting energy, but also added a longer time-constant to charge the filter cap. So that was a bad approach.

In the end, I added some current-limiting in the primary-winding driver. I'm not happy with it because efficiency is lower, but it does seem to work better. I've also been adjusting the duty-cycle and frequency of the DCDC converter. I cant go into the details of the control loops, but it's controlled by a small FPGA and has enough 'knobs' to allow software to tune it. I'm hoping that when it's built with real parts I can tweak the knobs enough to prevent the current-limiter from kicking-in. So much of the simulation is dependent upon the transformer modeling that it's not worth the effort to squeeze everything based on a model that might be inaccurate; it will need a lot of bench work.

[Michail Wilson]

Tony,

 

https://youtu.be/pvbAqrQQiIU?t=3m30s

Maybe posted before, but just ran across it.

 

Floating nixie clock – When are these going to start shipping?

https://www.kickstarter.com/projects/lasermad/time-flies-levitating-nixie-clock

 

Michail

 

 

 

[Paul Andrews]

They are already shipping. I am part way through building one: http://www.lasermad.com/shop/shop/ 

Virus-free. www.avg.com

--
You received this message because you are subscribed to a topic in the Google Groups "neonixie-l" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/neonixie-l/gE8ei40HMDs/unsubscribe.
To unsubscribe from this group and all its topics, send an email to neonixie-l+unsubscribe@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/00fd01d387fa%248fd1ee30%24af75ca90%24%40com.
For more options, visit https://groups.google.com/d/optout.

[Michail Wilson]

Waiting on mine.  J

 

Thanks, I didn’t know the shop exists.  Need a tic-tac-toe now.

 

(Unfortunately, I purchased the chess kit many years ago, but didn’t assemble and went through a house fire – no idea how much of the kit I have anymore.)

 

Michail Wilson

206-920-6312

 

From: neoni...@googlegroups.com [mailto:neoni...@googlegroups.com] On Behalf Of Paul Andrews
Sent: Sunday, January 07, 2018 1:32 PM
To: 'Terry S' via neonixie-l
Subject: Re: [neonixie-l] Lasermad

 

They are already shipping. I am part way through building one: http://www.lasermad.com/shop/shop/ 

 

Virus-free. www.avg.com

On Sun, Jan 7, 2018 at 4:00 PM, 'Michail Wilson' via neonixie-l <neoni...@googlegroups.com> wrote:

Tony,

 

https://youtu.be/pvbAqrQQiIU?t=3m30s

Maybe posted before, but just ran across it.

 

Floating nixie clock – When are these going to start shipping?

https://www.kickstarter.com/projects/lasermad/time-flies-levitating-nixie-clock

 

Michail

 

 

 

--

You received this message because you are subscribed to a topic in the Google Groups "neonixie-l" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/neonixie-l/gE8ei40HMDs/unsubscribe.

To unsubscribe from this group and all its topics, send an email to neonixie-l+...@googlegroups.com.

To post to this group, send email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/00fd01d387fa%248fd1ee30%24af75ca90%24%40com.
For more options, visit https://groups.google.com/d/optout.

 

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.

To post to this group, send email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/CABCPn3jo-FdKewh-GD4L%3D6FbccWYTijr229q1NgwVzKaQz_mtA%40mail.gmail.com.

[Cqr]

Me too :)

 All I need to do now is add the symbol tube and assemble the base case, once instructions are posted.

 

Kind Regards,

 

Robin Bussell 
Research & Development Engineer 

Email: Rob...@excelerate.info 
Tel: +44 (0)845 658 5747 
Mobile: +44 (0)776 016 4551 
Website: www.excelerate-group.com 
Website: www.exceleratemarine.com 

 

Subscribe to our mailing list

Head Office Cardiff UK, Subsidiaries and offices in Melbourne, Perth, Washington 
DC, Abu Dhabi and Cannes.

Company No: 4260093

IMPORTANT - The information transmitted in this email is intended only for the person or entity to which it is addressed
and may contain confidential and/or priviledged material. Any review, retransmission, dissemination or other use of, or
taking of any action in reliance upon, this information by any persons or entities other than the intended recipient is prohibited.
If you received this in error, please delete this email and inform in...@excelerate.info immediately.

 

 

You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.

To post to this group, send email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/CABCPn3jo-FdKewh-GD4L%3D6FbccWYTijr229q1NgwVzKaQz_mtA%40mail.gmail.com.

[Nicholas Stock]

[Cqr]

Man those were some fiddly little components one that pcb! 

It didn't help that I've a Weller magnastat iron and the resistors would stick to it!

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/8E7BA740-4EEC-4CD4-9394-135E2AF91272%40gmail.com.

For more options, visit https://groups.google.com/d/optout.

<Video.MOV>

I’m waiting for the base case instructions too....:-)

Sent from my iPhone

On Jan 7, 2018, at 14:11, Cqr <ro...@cqr-ltd.com> wrote:

Me too :)

 All I need to do now is add the symbol tube and assemble the base case, once instructions are posted.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/8360AA54-8BED-48DA-AFE8-7CFB7E68E621%40cqr-ltd.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/8E7BA740-4EEC-4CD4-9394-135E2AF91272%40gmail.com.

[Nicholas Stock]

You’re not kidding....those magnets are pretty damn powerful.

Smallest SMD work I’ve had to do, but gratifying when the board lit up. 

Sent from my iPhone

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/EE14DE53-5C0B-400C-BA3C-78D4983B7FFF%40cqr-ltd.com.

[Paul Andrews]

Indeed. Change one little thing and everything else changes. The PS chip I use is already current-limited (LM3478). In my case, it is a nice backup that I use principally to prevent the transformer from bursting into flames. Something that I didn't expect was that running the whole thing at a relatively low frequency improved the performance. I started with 600KHz, but around 300KHz turned out to be optimal with the hardware I ended up using (I'm using a Wurth flex transformer, which is billed at being OK up to 1MHz). I also got through four or five different MOSFETs - my first choice got a little toasty in my prototype, so I searched diligently for low RDSon chips with a fast off-time. I assume the quite high RDSon of my first choice was the cause of the temperature rise, and indeed a lower RDSon seemed to completely get rid of that.

Of course, a faster off-time makes the ringing worse.

I would be interested to see what the resonance looks like in a wave form - i.e. what kind of amplitudes it gets up to.

[gregebert]

There's actually a lot of odd things happening all at once; resonance is one of them.

First, the output voltage doesn't 'spike' at/near resonance. But it does oscillate a few volts peak-to-peak.

Second, the diodes start showing their parasitic effects, like reverse-recovery. That seems to be fueling the oscillations.

Third, the simulation does not always stay well-behaved (see picture)

Waveforms are non-linear, so it's not always obvious where problems creep-in. You can see the ugly triangle of the primary transformer current. The resonant-frequency and the switching frequency of the inverter will heterodyne with eachother; I can easily change the duration + location of the 'triangle' by changing the switching speed.

Unfortunately, the control loops I have in the design are too complex for me to put into LTSpice, and I dont have a true mixed-signal simulator (such as Synopsys XA) to use at home. So all I can do is piece-together what I can with simulations, build the board, and tune it.

In case you are wondering, the answer is yes; I do plan to add some dithering (randomness) in the control loops so that heterodyning is unlikely to occur.

[David Forbes]

Interesting thread.

I learned a lot about voltage doublers and flyback transformers when I
designed my scope clock power supply back in 2000. The goal was a
-1.5kV supply, so I used a 10x multiplier. I also used a positive
doubler on that winding to make +300V for the deflection. However, in
order to get good regulation of all the outputs (I used six windings!)
I found that I had to use doublers on all outputs. This means that the
supply isn't a flyback converter, nor a forward converter, but a
combination of both.

I like to think that I made a new switching power supply topology,
which I will merrily believe until someone shows me that I'm wrong.