Help on 2400W Phase Shift PSU

[switchmodepower]

Hold on. You still have the cs xfmr in series with the primary? You have to change this or it will never work in current mode.

And btw. Nothing wrong with voltage mode.

[mag]

Yes, I have still the xfrm in series with the primary.
I will change its position before trying to run in current mode.

I will put it in series with the two upper mosfets drains.

But before doing this I need to add snubbers on the output rectifiers to damp
the ringing

thank you

[mag]

Hi all,
I have made some experiments with voltage mode and everything seems good. The current xfrm is still in series with the primary and I am using the DC block capacitor.

I have pulled 80A / 24V from it with a global efficiency (PFC+converter) of 93.5%. At 80A there are some PCB track that starts to heat a lot, I need to reinforce them before going to 100A.

One question: my SMPS will work also in constant current mode (with a separate error amplifier). If I connect to a resistive load of around 0.1ohm I can easily regulate the current from 10A to 80A with no problem. If I connect a real short circuit (short 25mm^2 wire directly on output) there are some instabilities. I am able to regulate the current in the short but sometimes the primary current limit kicks in. This happens also if I set the output current to around 10A.

Is this happening because I am tring to operate a voltage mode converter as a constant current source or can it be just a loop compensation problem.

After changing to voltage mode I have changed my volage loop to type 3 with around 6kHz BW and 70° phase margin. What about the current loop?
No idea on how to proper compensate the converter when it runs as a constant current source, do I need type 3 or type 2 can do the job?
Any suggestion will be appreciated.

Thank you

-marco

[savu]

Originally Posted by mag

Hi all,
I have made some experiments with voltage mode and everything seems good. The current xfrm is still in series with the primary and I am using the DC block capacitor.

I have pulled 80A / 24V from it with a global efficiency (PFC+converter) of 93.5%. At 80A there are some PCB track that starts to heat a lot, I need to reinforce them before going to 100A.

One question: my SMPS will work also in constant current mode (with a separate error amplifier). If I connect to a resistive load of around 0.1ohm I can easily regulate the current from 10A to 80A with no problem. If I connect a real short circuit (short 25mm^2 wire directly on output) there are some instabilities. I am able to regulate the current in the short but sometimes the primary current limit kicks in. This happens also if I set the output current to around 10A.

Is this happening because I am tring to operate a voltage mode converter as a constant current source or can it be just a loop compensation problem.

After changing to voltage mode I have changed my volage loop to type 3 with around 6kHz BW and 70° phase margin. What about the current loop?
No idea on how to proper compensate the converter when it runs as a constant current source, do I need type 3 or type 2 can do the job?
Any suggestion will be appreciated.

Thank you

-marco

hello Marco,

any chance of posting the PFC schematic?
i am about to start work at a 5kw PFC and anything already made at kw levels would be of great info for me.

regards,

[wtob]

hello Marco,

have you already solved your problems? I'm actually also designing a similar power supply and possible we can share some experiences.

regards

[ontoaba]

Originally Posted by wtob

hello Marco,

have you already solved your problems? I'm actually also designing a similar power supply and possible we can share some experiences.

regards

marco, may this work for current sensing instead of using many sensor at each mosfet leads.

yes please continue the post. and the pfc sch too
I'm now designing buck based PFC for 500W limited load with 150V stable output.

May I asking why using phase shift and why full bridge instead of half?
thanks

[Eva]

With average current mode and a plain short circuit load, find out why current is increasing over the limit, how and when. You will need a storage oscilloscope and careful triggering to capture the waveforms.

It may be transformer saturation or overshoot due to loop instability.

Note that a plain short circuit load will turn output capacitors into a short, thus turning output filter LC, a 2nd order system, into just L driving a short to ground, a 1st order system with lower gain. Unity gain frequency will change. But I think the problem is transformer saturation (due to current transformer saturation).

[ontoaba]

Originally Posted by Eva

With average current mode and a plain short circuit load, find out why current is increasing over the limit, how and when. You will need a storage oscilloscope and careful triggering to capture the waveforms.

It may be transformer saturation or overshoot due to loop instability

The output voltage may just too low, or may be transformer.
Low output voltage will cause the inductor (transformer in this case) flux not discharged (look this graph).
In current mode also the transformer designed different from voltage mode. In picture above, there four inductor in current mode power transfer, I placed small ferrite in center with paper as spacer. It will have better flux density than without spacer. Please take a time for looking current mode (flyback) transformer design.

Hi Eva, how your class-D?

[mag]

@savu.

sorry I can not share the PFC schematic but I can tell you more or less how I did it.

I have used a borderline PFC controller to control a buck converter. The mosfet is driven through an optoisolated drive. The buck converter input is
connected to the output of a 3-phase bridge rectifier with some L-C filters to reduce input ripple current and thus EMI content.
At the output of the buck converter I have the capacitors bank and I regulate the output at around 385V.
The PFC works always in DCM or in BCM so there are no recovery time issues for the diode. Working in BCM allows also to reach quasi ZVS operation.

The PFC is designed for an input voltage between 350 and 500Vac 3phase.
The PF is around 0.95 and the input current is not a sinewave but has far less harmonic content than a standard diode-capacitor input stage so that I can fullfill class-A requirements on line harmonics.

@onto aban

I can not use a shunt resistor for current sensing; I need isolation. I have my
controller on the secondary side and I must sense the current on the primary side, a current transfomer is the easiest way. The best way to do the job with resistors is to connect the lower side mosfet sources toghether and then connect them to GND through a sense resistor. In this way you have your current sense volatge always referred to GND.

This type of PFC works only for 3phase systems. If you need 150V 500W output from your PFC and your input is a single phase AC this PFC will never
work. This because the buck goes at 100% duty cycle when the input AC falls below 150V and it will not correct the PF.
The good think of 3phase rectifier is that the voltage never goes to 0V, the
rectifier ouptut is an haversine of 300Hz with the maximum at Vin*sqrt(2)
(565V in a 400VAC system) and the minimum at Vin*sqrt(2)*sin(60°)
(490V in a 400VAC system), the point where two phases crosses. In my case
regulating the output at 385V ensures that the buck is always working because Vout < Vin at every moment of the line cycle.

If you need 150V stable a PFC of every kind is not the way to go. The PFC
output has many volts of 2*fline ripple and this is normal; more ripple you allow the better the PF will be.

If your system is single phase and you need a 150V 500W stable output for me there are only two way to go:

1) standard boost PFC (possibly CCM) regulated at around 400V +
buck regulator to step down the 400V to 150V

2) Flyback PFC, it removes the limitation of having vout>vin of the classical boost PFC. You have a transformer so you can play with the turn ratio to have the output voltage you need. It give also galvanic isolation if you need it. In any case your output voltage will not be a pure DC at 150V but it will have a 100Hz ripple of some volts. If you can accept this I think that this is best way to go.

I have used the phase shift topology because it allows a simple drive of the synchrounous rectifier and using the current doubler I can reduce the current
in the main tranformer from 100A to 50A (more or less) an the primary switches can switch at zero voltage elimitating switching losses. The expense is two more mosfet on the primary side and a more complicated controller and gate driver.



@ Eva

You got exacly the problem. With a short on the output the system becomes of 1st order and there are no way to compensate it with an adeguate phase margin using a standard type 3.
If you tune the compensator for the plain short (1st order) it will be unstable
when the output voltage rise and the system is operating as a current source (2nd order) and vice versa.
I have developed an alternative compensation network adding other two zeros on the compensators to have less phase boost but over a wider bandwidth. With this compensator I have always around 45 degrees of phase margin sweeping the load resistance from 1mohm to 200mohm.
My problem is now fixed and the system is stable on every load even when using it as a current source.
Now I am still using voltage mode, with series cap and with the current transformer in series with the primary of the main transformer. Nothing seems to saturate and everything runs fine and cool except the inductance in series with the primary (wrong calculation of core losses--> I need a bigger core with more turns).
At this stage I don't think to move the current transformer to use current mode control, everything is fine in voltage mode so why change it?

@onto aban
This is a forward converter, I need to reset the transormer only for its magnetizing current, not the full current as in the flyback.

[ontoaba]

Originally Posted by mag

@savu.

The PFC is designed for an input voltage between 350 and 500Vac 3phase.

Industrial?

Originally Posted by mag

@savu.

I can not use a shunt resistor for current sensing; I need isolation.

Transformer still there, but high impedance (voltage mode) calculated from R voltage.

Originally Posted by mag

@savu.
If you need 150V 500W output from your PFC and your input is a single phase AC this PFC will never
work. This because the buck goes at 100% duty cycle when the input AC falls below 150V and it will not correct the PF.

Yes, that's right. Its already three converter. Two of them working at low power. The output is stable 150V but I am still have big problem with this.
Using flyback need big electrolyt caps, I just avoid this one.

Originally Posted by mag

@savu.

I have used the phase shift topology because it allows a simple drive of the synchrounous rectifier and using the current doubler I can reduce the current
in the main tranformer from 100A to 50A (more or less) an the primary switches can switch at zero voltage elimitating switching losses. The expense is two more mosfet on the primary side and a more complicated controller and gate driver.

Interesting, is the transformer size reduced?

Originally Posted by mag

@savu.

My problem is now fixed and the system is stable on every load even when using it as a current source.

You mean short circuit at the output? My amp did it, still operating at short output, so there no need recovery when the short opened. I simply place impedance limiter from the current sense to limit the input and let the output devices dissipate low power. Since it is class-H the amp working at lowest rail.
You are using switching that none will dissipate any, I have no idea.