Author Topic: Fullbridge LOAD TRANSIENT RESPONSE (Read 4115 times)

MacIntoshCZ · « **on:** January 30, 2023, 07:38:21 am »

Hello, i think i need to add another GDT transformer. At the moment i am using one with 4 secondary windings (Schematic is using wrong transformer).
Yellow trace in pictures is Vgs, blue trace is Vds. measured on transistor Q2
Turning on is okay, howeever turning off is quite bad. There is some ringing probably due miller capacitence coupling to gate of mosfet.
Fullbridge control chip is TL494.
When GDT is off, both gate drivers is off. I though that transistor will turn off via gate resistor, it does but not nicely =D.
I think i need to make another GDT. One for controling Q1, Q4 and for Q2 Q3. Only difference will be, that driving voltage for gate will go directly to negative side. So there wont be a gap where voltage on GDT is zero. That should solve my problem. What do you think?

Thanks

MacIntoshCZ · « **Reply #1 on:** January 30, 2023, 08:08:03 pm »

Ok so i have added another GDT and rework old one.
But to be honest it actually worsen the waveform.
Again measured on Vgs, VDS of Q2

Now there are strange oscilations

moffy · « **Reply #2 on:** January 30, 2023, 08:47:10 pm »

I am replying to your first set of waveforms. You are not measuring current which would tell you more than your voltage waveforms. When the bridge transistors are all OFF together, the bridge is in a high impedance state and the load and parasitics will dictate what is happening with the voltage, Vds, you have of course the body diodes of the MOSFETs to limit extremes. The only true voltage measurement is a differential one across the load.

2N3055 · « **Reply #3 on:** January 30, 2023, 08:50:54 pm »

There is also fascinating topic of measuring with a scope... You need to say more on your measurement setup.. It can easily be that what you see on a scope is not actual waveform...

MacIntoshCZ · « **Reply #4 on:** January 31, 2023, 10:25:51 am »

Adding waveforms, measured directly on transformer.
I am trying my best to limit EMI/EMF. I have designed everything to be as close as possible to fight parasitic inductance. i am using curled wires for feedback, mosfet driving... By curled i mean two wires close together, sorry for bad english =D...

moffy · « **Reply #5 on:** January 31, 2023, 10:41:38 am »

You have a lovely LC oscillation across the transformer, have you tried damping it with an RC snubber?

MacIntoshCZ · « **Reply #6 on:** January 31, 2023, 10:58:32 am »

I was thinking about adding one 😅. Howeever i though that magnetizing current from primary will recuperate through mosfet body diodes back to bulk caps. So it wont be a necessary. Seems i was wrong 😑

moffy · « **Reply #7 on:** January 31, 2023, 11:35:35 am »

It looks like your first set of waveforms are better damped than the second set. Can you go back to the first configuration? Or are there other issues involved?

moffy · « **Reply #8 on:** January 31, 2023, 11:50:27 am »

I just noticed on your first post that you have 2.2nF caps in series with 6.8 ohm resistors across each FET. The resonance is around 600kHz so the caps have an impedance of about 120 reactive ohms, which means the 6.8 ohm resistor is doing very little at that frequency. The 2.2nF caps might be contributing to the ringing you might want to raise the value of the 6.8 ohm resistor.

MacIntoshCZ · « **Reply #9 on:** January 31, 2023, 12:11:42 pm »

I was doing some changes, there are 50ohms resistors and 1nF caps used in RCD.

I have only added another GDT which is turning mosfet on and off way better (actually only off better, turning on is same). There was positive overshoot on gate which turned again mosfet on, so that why i improved GDT to go to negative voltage ass well.
And reason why i did this change was that Q2 as only transistor was getting hot. Dont know why. But with added GDT now all transistors gets hot. Temp of Q2 is lowerm howeever there is rise of temp on others. Its not critical temperature, you can hold finger on it.

moffy · « **Reply #10 on:** January 31, 2023, 12:54:16 pm »

Does the output transformer have a load on it? That would tend to dampen oscillations. 1nF/50R is a bit better but you need to kill the 600kHz so I would be tempted to try 4.7nf/50R. If you can keep your finger on it it's not hot, but that depends on whether it is driving a load, no load should be cool. The body diodes tend to be slow and since they are conducting some of the heat might be coming from them.

MacIntoshCZ · « **Reply #11 on:** January 31, 2023, 01:35:26 pm »

PSU is designed for 200V 5A, so far i have tested only aprox 700W. 130V on 5+Amps. So yes, i am measuring with load on it

moffy · « **Reply #12 on:** January 31, 2023, 01:49:42 pm »

Quote from: MacIntoshCZ on January 31, 2023, 01:35:26 pm

PSU is designed for 200V 5A, so far i have tested only aprox 700W. 130V on 5+Amps. So yes, i am measuring with load on it

If you can touch the FETs without burning your finger, you are doing pretty well, dissipation wise.

MacIntoshCZ · « **Reply #13 on:** January 31, 2023, 05:05:06 pm »

Mosfets exploded. I looked inside and i think i have bought fake transistor. This should be irf460A. Also measured RDS and its about .44ohms. in datasheet is 0.27 ohms...

What a nice buy.

MacIntoshCZ · « **Reply #14 on:** January 31, 2023, 05:19:00 pm »

So after explosion, two transistors are gone, zener protection of gate of those two transistors are gone and i have also noticed one of 1nF cap is gone. But this cap was on other side where transistor survived...

moffy · « **Reply #15 on:** January 31, 2023, 11:14:17 pm »

Sounds like excessive dv/dt, overvoltage. Those snubber components look awfully small for a 1kW psu.

MacIntoshCZ · « **Reply #16 on:** February 01, 2023, 06:29:31 am »

Maybe i am using wrong material for caps. Those ones are polypropylen 1kV
But there are two 100R resistors in parallel. I have added one on other side.

moffy · « **Reply #17 on:** February 01, 2023, 06:50:31 am »

Polypropylene is a great material, very low dissipation, but they need to be physically large to handle the power needed. If the caps are blowing and the resistors are not getting hot, then the resistors are not doing their job i.e. the snubber is not working.

MacIntoshCZ · « **Reply #18 on:** February 02, 2023, 10:47:23 am »

With combination of 4n7 caps and 50R resistors, it looks damped well. Measured Vds of Q4

Thanks

So its better to have higher resistor value than reactance/impedance of capacitor.

Wolfram · « **Reply #19 on:** February 02, 2023, 11:01:28 am »

How are you generating your PWM and enable signals? Keep in mind that GDTs will enforce a zero volt-second balance due to their magnetizing inductance. Any time asymmetry in your upper and lower drive pulses will also lead to a voltage asymmetry, and there is also the potential to excite resonances between your magnetizing inductance and DC block cap.

T3sl4co1l · « **Reply #20 on:** February 02, 2023, 01:21:17 pm »

Show the control circuit and connection.

Also show GDT input (before and after the coupling cap).

Tim

MacIntoshCZ · « **Reply #21 on:** February 02, 2023, 06:22:14 pm »

I am using two GDTs, one for OutA and second for OutB.
Upper zener diode tempt to blow up and positive voltage on output of GDT is higher in absolutive values than negative one. Its probably as Wolframe say.

Yellow trace - Out A on input of gate drivers
Blue trace - output of UCC37321P (Just noticed that i am using UCC37321p in place of UCC37322p and vice versa). So cap before transformer is near UCC37321p.
Purple - Output/secondary of GDT, without load on it (mosfet not connected)

MacIntoshCZ · « **Reply #22 on:** February 02, 2023, 06:38:27 pm »

here can be better seen voltage assymetry
Y - input of gate driver
P- secondary of GDT

(I have replaced noninverting and inverting driver between each other, so now its as in schematic. And also replaced outputs of GDTs. Nothing happen, but now its as in schematic =D)

VCC is about 12,5V aprox

T3sl4co1l · « **Reply #23 on:** February 02, 2023, 06:49:53 pm »

What is PWM_stav?

Tim

MacIntoshCZ · « **Reply #24 on:** February 02, 2023, 07:14:48 pm »

When there is GND at enable pins of gate drivers, it shutdown. But i am not using it at the moment.


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: Fullbridge LOAD TRANSIENT RESPONSE (Read 4115 times)

Share me