Strange voltage spikes in MOSFET gate driver

Thread Starter

2155X

Joined Jun 5, 2018
7
Hello,

I have designed a TL494 half bridge power supply which simulates fine in LTspice but has some problems when soldered on a board and I can not figure out how to fix them.

Schematic pictures are attached.

1. If I connect GDT1 and GDT2 as seen in the schematic pictures, the output waveforms from the push-pull transistors have large voltage spikes. Pictures of the waveforms:

IMG_20210413_220237.jpg IMG_20210413_220249.jpg

2. If I power the controller part without GDT at around 8V, the circuit draws around 15mA, waveform looks normal. As soon as I raise the voltage to something like 10V, the circuit starts drawing 400mA and waveform voltage from Q1 Q2 push-pull is slightly lower than Q2 Q3. Not only that, but Q1 starts heating up much more compared to Q3. I have tried replacing Q1 and the controller - did not help.

3. After powering up the half bridge with a DC supply (not 320V for testing) the observed voltage shape at MOSFET gates does not look good and they heat up quite a bit. I think this is related to the spikes at the push-pull transistors. Picture of the waveforms:
IMG_20210414_172406.jpg

If needed, I can also take pictures of the soldered circuit.

Thank you for your help.
 

Attachments

Marley

Joined Apr 4, 2016
502
Not happy with your circuit. The BD140 transistors on the primary H-bridge are only driven with the 1k base resistors and will not conduct much current.
Instead of having only 1 transformer with 2 secondary windings why not use 2 separate transformers. Each with 1 primary and 1 secondary. The TL494 has push-pull outputs. Drive a transformer from each output. Drive them "single-ended". Possibly with a single extra NPN transistor on each output if more current is needed. The fly-back should switch off the driven MOSFET with the correct gate circuit.
Important to have some guaranteed minimum "dead time" on the high voltage H-bridge for safety. Can the TL494 do that?
 

Thread Starter

2155X

Joined Jun 5, 2018
7
Not happy with your circuit. The BD140 transistors on the primary H-bridge are only driven with the 1k base resistors and will not conduct much current.
Instead of having only 1 transformer with 2 secondary windings why not use 2 separate transformers. Each with 1 primary and 1 secondary. The TL494 has push-pull outputs. Drive a transformer from each output. Drive them "single-ended". Possibly with a single extra NPN transistor on each output if more current is needed. The fly-back should switch off the driven MOSFET with the correct gate circuit.
Important to have some guaranteed minimum "dead time" on the high voltage H-bridge for safety. Can the TL494 do that?
Thank you for the reply.

This is driving a half bridge circuit. The TL494 is running in push-pull mode with a maximum duty cycle of around 47% or so, that provides enough dead time. I tried to change R18 and R19 to a lower value like 270 ohms and that seems to have helped with the spikes but they are still not to my liking. Space is limited and I do not have another core for another transformer to drive them separately.
Is there anything else I could do to get a cleaner signal?
 

Thread Starter

2155X

Joined Jun 5, 2018
7
I have spent some time modifying the circuit, trying different resistor values. Many hours and a few smoking MOSFETs later, looks like the following runs without heating up, smoking or blowing up when running at 325V DC:

R18, R19 changed to 235 ohm
R11, R12 changed to 680 ohm
R13, R14 changed to 7.5 ohm
R20, R21 changed to 100 ohm

I have only tested it with a light load of around 12W (don't have any suitable load on hand right now) but it does reach 35V and is adjustable down to around 2 volts without heating up the MOSFETs at all, no heatsink.

Some of the changes I made might be over the top, if you have any suggestions to make it more efficient I would love to hear them :)
 

kaindub

Joined Oct 28, 2019
125
You need to understand why the changes you made worked.
What happens when you start further testing and it starts to smoke again? Without understanding the purpose and effect of each component, you're just going to try substitution again.
 

Thread Starter

2155X

Joined Jun 5, 2018
7
You need to understand why the changes you made worked.
What happens when you start further testing and it starts to smoke again? Without understanding the purpose and effect of each component, you're just going to try substitution again.
With changes to R18, R19 I allow the transistors to have more base current, R11 and R12 prevents the MOSFETs accidentally turning on via noise, R13 R14 are damping resistors to reduce ringing and control the MOSFET rise time. R20 and R21 allow the transistor to conduct more and help with fall times of the MOSFET.

Make D2,3 15 ohms for starters ...
Are you telling me to change the diodes into resistors?


Thank you for your replies.

I believe I mostly understand my circuit. I think the problem is me having little experience working with high frequency switching and not knowing why those voltage spikes were happening. I think if someone told me why they were happening I could try investigating and solving the issue, that way I know in the future how to prevent it from happening again.

For now, I have decided to replace the transistor mosfet driver section with an IR2110.
 

LowQCab

Joined Nov 6, 2012
4,022
I would suggest that you not use a "Pulse-Transformer" Drive Setup,
they can be very tricky to design.
Instead, replace the Pulse-Transformer with 2 small AC Power Supply Transformers.
( or a single Transformer with 2 identical Secondary-Windings )
( I'm assuming that you have AC Power Available )
The AC Output Voltage should be ~9 to ~10VAC, then run this into a Bridge-Rectifier,
and finally add a ~470uf Storage-Capacitor, this will result in 2- ~14 to ~15VDC Power Supplies.
Then use a SI-826-X FET-Driver for each FET.
This will give you enough Gate-Drive-Current to eliminate your
"out-of-control" Gate-Drive-Voltages.
Analysis still needs to be done to insure minimal Gate-Ringing,
but it will only involve selecting one appropriately sized Resistor.
Your existing Output Circuit will easily Drive the Input of the SI-826-X FET-Driver.
.
.
.
 

Attachments

Top