Push Pull Converter with Load - voltage drop Issue

Status
Not open for further replies.
I think it's time to rule out transformers VA rating. when i checked with a scope using 100 watts load, primary is only 8V. I'm using a 12V battery. There is a large drop in MOSFETs. Mosfets are heating up in 15sec. Gate Voltage is 8. Mosfets are p55nf06. There is of course sufficient dead time between the signals.
 

So you think there is 4V across the fets? unlikely at 8V g-s, 100W is near enough to 10A out of the battery if the terminal volts fall to 10V, easy to happen for a small battery, you would need at least an 100Ahr battery to maintain 8A at 12V for any length of time, check your batt terminal voltage under load, and any volt drop in the cables from the batt to your converter... if the mosfets are heating up as you say then there are a few possible causes, not enough quality gate drive (the off fet is not properly off when it should be), not very good ON drive, no heatsinking on the fets, large turn off volt spikes causing the fets to avalanche due to high leakage inductance in the Tx. Too small dead time causing flux stair casing and high current in the fets (as seen by batt terminal volts dropping...) what is the dead time? do you have a 'scope to show gate and drain waveforms?

Also check your o/p diodes are OK and not being over-voltaged, also if you exceed the volt rating of your o/p caps (or near to it) they can act like zeners drawing current until they eventually get too hot and explode... so 400V - 450V electrolytics recommended...
 
Hi. I'm using 12V 7Ah Battery and the battery terminal voltage drops to 11.4V with 100 Watts load. I don't have 100Ah now, But will test it and post the results ASAP.
I have heatsink on the MOSFET.
Using SG3525 for driving the MOSFET directly. I also tried using BD139 and BD140 Transistor Pair to drive the MOSFETs. The "gate Voltage" at No Load is 12V, But it drops to 8V at 100 Watts Load. SG3525 is running on the same 12V Battery, and I believe it is because of Terminal Drop as you suggested.
Using UF5408 diodes
caps 100uF 450V electrolytics
I'll Post the gate and drain waveforms soon.
 

Tested with 150 AH Battery. Results are not convincing yet

15 Watts Load - 368V
100Watts Load - 310V

With 100 Watts Load the MOSFETs are really hot in 30 sec.

Attaching the gate and Drain Waveforms





The Gate Voltage was never below 10V. Tested with IRFz44n Mosfets.
 

It looks like the frequency is <10Hz which is too low for your transformer.
The blue Mosfet drain is completely dead and the yellow Mosfet drain goes negative instead of to ground.
 

Hi! Ignore f<10Hz, Observe the Time Period is M 2.5us, which gives a freq of 50KHz.

I have scoped only one Signal. The other side is the same. Well, the Second Image is the signals observed on the transformer primary. Between Center tap (+12V) of the Transformer and MOSFET Drain.

Sorry that I didn't explain this in last post.
 

Tested with 150 AH Battery. Results are not convincing yet
15 Watts Load - 368V, 100Watts Load - 310V,
With 100 Watts Load the MOSFETs are really hot in 30 sec.
Click to expand...

What are your fets? what is the on resistance? are they heatsunk well? 100W = 9A say, 81 x 0.1 ohm = 8 watts when on = 4 watts average, your fets should be in the order of 0.05 ohm giving 2 watts average if there is little heatsinking.

also can you show the two drain signals together please?

Also what are the volts right at the board, where the fets are, when at 100W? is there cable drop still? this will accout for the drop in o/p volts with load...
 
Okay, FETs are IRFZ44n, the Rds(on) = 0.017ohm (safe) and I have added heat sink to the Mosfets.

also can you show the two drain signals together please?
Click to expand...

I will actually be shorting both the drains together, The second waveform in the post #26 is the input to the transformer, ie centre of the push pull transformer with ground on drain.

There is no cable drop. When I tested with 150AH battery, I was able to overcome the drop issues, Other than the drop associated with the MOSFET heating.

The datasheet of MOSFET says
Continuous Drain Current, VGS @ 10V is 49A, I'm not even closer, its just 8.5 (max) from the battery.

I just used a totem-pole circuit with BD139 and BD140 and now having 2 Parrallel MOSFETs. I feel as if the heating is shared between the MOSFETs. Gate voltage is not less than 10V, Just made sure I'm not working in linear region.
 

Can you show the rise and fall times of the gate drive signals? Do you have a current probe that you can look at the drain currents?
 

emaniac said:
Okay, FETs are IRFZ44n, the Rds(on) = 0.017ohm
Click to expand...
Nope. It is 0.028 ohms maximum, only when it is cool and is higher when hot. You cannot buy "typical" Mosfets.

The datasheet of MOSFET says
Continuous Drain Current, VGS @ 10V is 49A
Click to expand...
Nope. It is 50A only if you can cool it so its case is not even slightly warm.
 

Your gate turn off does not appear to be hard enough...! pictures of drain-source more useful...! they can tell you a lot.

Heatsinks will be useful in keeping the case temp < 50 deg C...! look forward to the new info...!
 
Can you show the rise and fall times of the gate drive signals?
Click to expand...

Image in post #26 is the gate signals, is there anything else that you are expecting. Sorry i don't have a current probe.

You cannot buy "typical" Mosfets.
Click to expand...
I see that. And, Just to know - how can i identify the fake mosfets? Would like to get to know about that.

It is 50A only if you can cool it so its case is not even slightly warm.
Click to expand...

i do have heatsink on the mosfets. This heating issue can't be with the MOSFETs current handling capacity right? i don't honestly know if this mosfets can even run cool

will add the drain source signals ASAP
 


Number 1 problem is poor specs on each part, especially magnetics, as it appears you are saturating the ferrite core at too low a frequency

2nd is measurement errors of AC and turns ratio.

Turns ratio is unknown but assuming 24Vpp? input and 350Vac output ?p?avg?rms? unloaded, it maybe N turns ratio of 500/25=20

Now use Ohms Law and transformer impedance ratio.

An ideal iron core transformer not saturated, would give an impedance ratio of 400

What is your load R + L // C?

160V²/8W suggests it might be 3.2K but probably isn't.

So assume it is and primary load becomes 3200/400 = 8Ω



Meanwhile MOSFET resistance must be <5% of rated output worst case load resistance/400.


meanwhile primary results of 12V/0.9 A or 1.33Ω .. Not 8

.X400 suggest your 160V test load was 530 Ω and not 3.2K so your Vac rms or pp reading was bad!

- specify all exact magnetic properties
- Do the math and start thinking 30-150kHz PFC PWM pseudo sine wave into ferrite to have any chance of having this work.
 

pictures of drain-source more useful...!
Click to expand...

True! Here's the Drain-Source signal.



The Vds = 60V (from the scope image) at turn on. The Mosfets Vds is supposed to be 55V(max)

How could I avoid this spiking at turn on?
 

First, get rid of the yellow line.! and repost with 5v/div, during the fet off time, you can see the volts are 20-21V, you are missing 3-4 volts here, equating to 1.5-2V on the 12V, a 16% drop, due to i^2R losses in cabling connections etc, this largely explains the 368V to 310 volt drop in your o/p...

The spiking is at turn OFF, luckily for you the fets can handle this like a zener for short spikes, as long as you have some heatsinking. Else put 47 volt 5W zeners across the fets to make you feel better.

The spiking you see at turn ON is common mode artifact being picked up by your scope, obviously at turn on the Vds should go DOWN not UP...

So increase the copper in your wiring and you may well see your o/p become more constant with load - wait - didn't someone suggest this a long time ago? also re-check you battery terminal voltage under load, 12V ? 11.5? 11.0?

- - - Updated - - -

Also your Fsw is not 72kHz, read the divisions on the scope closer to 62,500 Hz.
 
I didn't design the Transformer, I got it designed by a company here.
I'll add the details of the Transformer


Frequency is not too low. It is 50KHz.

I'm not rectifying the signals as mentioned in POST #1 now, I removed the center tap and now using a full Bridge with 100uF - 450V capacitor. Haven't used any Inductor in filtering.

The image in POST #36 is with 7AH battery, Sorry but I can't get the scope and 150AH battery together, If have to POST some scope results I can test the converter with the 7AH battery only. Only thing I can ensure is even with a drop in Battery Terminals to 11.5V, the gate voltage doesn't drop below 10V.

Now the problem is not with the drop in voltage at the output. Once I get a bigger battery the Voltage will be 350V if and only if MOSFETs are'nt heating. Even with 150AH battery the MOSFETs were hot! May be I can reduce the load to 15watts, so i can make sure not more than 1.5amps from the 7AH battery is drawn.

Thanks for all the replies and suggestion, I'm learning a lot.
 

There will be some mosfet heating from absorbing the switching spikes - bigger & better heatsinking required...

Check your cabling for running off the bigger battery, its easy to lose volts at higher powers....

check the PIV of the diodes when operating flat out too, easy to overvolt these if you do not have snubbers....

- - - Updated - - -

p.s. even in post#26 you're only making it to 20 volt when it should be 2 x 12 = 24 giving the voltage droop on o/p, cable losses...!
 
Hi,

Please use the scope to measure g-s voltage of the low side fet. But connect scope gnd directely at the source leg and the
tip directely at the gate. Is the signal ok?

I think of a wiring or layout problem are the connections between driver and fet source pins short, low impedance and free of additional current?

Could you please show a photo of your circuit? I'd like to see all the fet to driver conbections and the bypass capacitors..

Klaus
 

Status
Not open for further replies.

Similar threads

Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…