Regulated SMPS strange mosfet waveforms :-(

[pergo]

hi all,

first of all, i wanna say congratulation for this amazing forum.

anyway,
i'm an italian electronic technician, and i like built car-audio electronic equipment like SMPS for pre, tubes, little amp.
I made several 12V SMPS with success, from +-24V to +-150V, always unregulated.

I tried to "mod" the project for regulation, with (maybe) success. The rail voltage are correct and regulated, but i have a strange waveforms on switching mosfet.
No noise at output, no ringing, no heat, low ripple (but more than unregulated..correct?)...no problem.

I see the same waveforms in several car-amplifier, like italian Audison, and american Phoenix Gold M series (the second).

I always use SG3525, zener+opto reference, +compensation at pin 9 of oscillator.

I think that waveforms are false triggering of o-scope, but i buy a Kicker XS100, with similar regulated SMPS, that not have strange waveforms on mosfets.

This photo come from an audison LRx1.400 (900Wrms @1ohm) provide +-53V regulated, but egual to my projects mosfet waveforms.
I always add a bjt buffer from oscillator out, to gate, with pull-down resistor from fast discharge.
I don't know if that Audison have bjt buffer, but open loop (5v at pin9..eheh), the switching is fine and i've +-63V @12Vin.

The amplifier is original, but i upgrade mosfet from 80n06 to irf3205. the waveforms doesn't change.

This is drain @10Vin:
http://img690.imageshack.us/img690/3995/10vdrain.jpg

Gate @10Vin
http://img200.imageshack.us/img200/7583/10vgate.jpg

similar to open loop.

but, drain @12Vin:
http://img709.imageshack.us/img709/6708/12vdrain.jpg

gate @12Vin:
http://img192.imageshack.us/img192/7397/12vgate.jpg

sinewave????? + square wave pwm

drain @14V:
http://img193.imageshack.us/img193/1791/14vdrain.jpg

gate @14V:
http://img109.imageshack.us/img109/4905/14vgate.jpg



for drain measure, 5V/div, 5us/div and AC coupled.
for gate meadure, same but DC coupled.


what a hell is that waveforms??
false triggering??

i've no problem at the output, no heat from mosfet. everything ok but a wanna understand "why"

thank you very much



ps: sorry for my english

 

[Sch3mat1c]

Circuit?

 

[pergo]

my projects are based on this:
Switchmode Power Supply For Car Audio

but with some changes.
- NON INV IN: Vref/2 by voltage divider from Vref 5V
- INV IN: 10K goes to optocoupler and to ground with 1K resistor.
- Opto is 4N35 with collector to Vref.
- COMP feedback resistor to INV IN is 33k.
- Compensation Cap from COMP is 1000pF to ground.
- OUT A and OUT B goes to BJT buffer totem pole MPSA06/56.

This changes are from Diamond Audio D7401 car-amp i've cloned.(by service manual)

Mosfet are STP50N06 or IRFZ44 in my projects, and STP80N06 in the original Audison with max 22ohm gate resistor (10ohm for STP80N06).

Audison exact circuit is unknown. Is difficult to "debug" it because is all SMD.

But Phoenix Gold M series have the same "strange waveforms" but they use the same regulation of my project with opto (like Diamond Audio i've cloned).

I tried to change COMP capacitor with other value but change a little bit (sometimes get worse and make noise).
I tried to rise deadtime but nothing..same waveforms


I don't understand why there is that sinewave, and why happens in several car amps i have (some never repaired or touched)

thank you

 

[pergo]

here is the circuit (used in my SMPS):
http://img121.imageshack.us/img121/8530/im1y.jpg

i use LC filter for +-Vrail with 2200uF per rail, and unknown L (stolen from ATX power supply)


Audison use 4000uF per rail, unknown L made by 1" toroid
Audison LRx 1.400 inside - Amp Guts

Phoenix Gold M25 with the same strange waveform (but only C output filter):
Phoenix Gold M25 inside - Amp Guts

 

[event horizon]

I don't understand why there is that sinewave, and why happens in several car amps i have (some never repaired or touched)

Looks like a typical type of waveform to me considering you are looking at the voltage accross the transformer. Perhaps you have not taken into account what the snubbers are doing? Don't forget that they'll be discharging through the transformer when the mosfets switch off

 

[pergo]

Looks like a typical type of waveform to me considering you are looking at the voltage accross the transformer. Perhaps you have not taken into account what the snubbers are doing? Don't forget that they'll be discharging through the transformer when the mosfets switch off

thanks for the reply

you're right, but is it normal? (i measure voltage across mosfet drain)
can snubber store so much energy that produce that voltage amplitude?

solution?? (RCD snubber?)

thank you


ps: kicker XS100, regulated 1000Wrms @1ohm, have snubber but don't make this joke

 

[event horizon]

thanks for the replyyou're right, but is it normal? (i measure voltage across mosfet drain)
can snubber store so much energy that produce that voltage amplitude?

If you are measuring the voltage at the mosfet drain (from ground) you are effectively measuring the voltage accross the transformer.

Yep, perfectly normal When i used to build, test & fix SMPS at a couple of firms i worked for that was the kind of waveform you'd see all the time once powered up (if working correctly).

You'll likely only need to modify the snubbers if when the PSU is fully loaded you get high voltage spikes that could threaten the mosfets breakdown voltage.

A full or half bridge & push pull SMPS as you have there will all display the same type of waveform when running correctly.

 

[pergo]

fiuuuuuuuuuuuuuuuuu

god bless you!
thank you thank you thank you!

i was going crazy for this unfairy joke

Everything work perfect at idle and full power, but i didn't understand why..now i understand!

Snubber are fine in my projects. Under heavy load (1kW) i have less than 1V overshoot.
I've rebuilt several car-amp SMPS but all unregulated
...and when regulated.......sineeee

again, thank you thank you thank you!

 

[Eva]

What you are seeing is transformer magnetizing inductance and output inductance resonating with parasitic capacitances.

Drain waveforms look ok (as long as they are symmetrical and consistent through consecutive cycles).

Gate waveforms are ok too (as long as parasitic turn on does not happen). The sine like part is produced by drain-gate capacitance feding drain waveform to the gate and a wide class B dead band from gate buffers and SG3525, probably +/-1V.

 

[pergo]

What you are seeing is transformer magnetizing inductance and output inductance resonating with parasitic capacitances.

Drain waveforms look ok (as long as they are symmetrical and consistent through consecutive cycles).

Gate waveforms are ok too (as long as parasitic turn on does not happen). The sine like part is produced by drain-gate capacitance feding drain waveform to the gate and a wide class B dead band from gate buffers and SG3525, probably +/-1V.

thank you Eva!

a question:
push-pull SMPS that not have this "joke", how are they made?
i mean, Kicker XS100 (and others), square wave is perfect (without sine).
Better trafo?
Better windings?

 

[rajudp]

try changing R170 and R10 ( 150 ohm ) to 10K Resister, check the transformer windings ( calculations )

 

[pergo]

try changing R170 and R10 ( 150 ohm ) to 10K Resister, check the transformer windings ( calculations )

but make slow gate discharge with 10k, or not?

 

[Eva]

Could you show the schematic of the output section with proper component values? Did you use an output inductor? The resonance is not a joke, it's not a bad thing at all, although I never liked project 89, it's good for learning because it's simple, but they can improve/optimize it in many obvious ways but the years go by and they never do. The true joke is inductor-less output regulation.

The resonant sine section should disappear in regulated circuits with output inductor operating in continuous conduction mode. Discontinuous conduction or no inductor results in sine section.

The purpose of the 150 ohm resistor is to keep the gates discharged when the SG3525 is powered off and is not supposed to be able to do it. However, 150 ohm is a very low value that results in wasted power. 10k is ok as suggested, although I never use that resistor because my multimeter tells that SG3525 can prevent the gates from being charged by leakage current even when it's powered off.

Consider replacing the gate buffer transistors by IR4427 driver IC, it will produce slightly better gate waveforms, the output swings to ground and it does not require the 10k resistor.

 

[pergo]

thank you Eva,

i made car SMPS from 300W to 1kW (unregulated, but i tried to make regulation), and i always use mur1620ctg(r) + LC output filter.
In some tube-car SMPS i use CLC filter

Well, for LC filter i use:
- C from 2200uF to 6600uF x rail (from 300 to 1000W)
- Inductor is token by old ATX supplies (0 cost) or broken car amps (obviously with the same power i need if it's possible).

i don't know the value, even the range of inductance

As you can se in the Audison, for example, Audison LRx 1.400 inside - Amp Guts

with the same """problem""", it uses 4000uF x rail + unknown inductance. The L is made by a 1" toroid, with 20 turn of 1mm diameter wire. Don't know material, i think iron powder.

but, Phoenix doesn't have L, only C, and there is the same waveform (+ little noise). Phoenix Gold M25 inside - Amp Guts
Unfortunately i sell the Phoenix, so i can't post an image from o-scope.

For the Kicker, it has a biiig LC filter. Inductance is 1,4" iron powder toroid with 2x1mm diameter wire with a lot of winding (cannot count, but sure over 50):
Kicker XS100 inside - Amp Guts
it doesn't produce that waveforms and it uses opto+zener feedback.


now i try to change resistor to 10k and measure drain voltage.
If i can, i try the IR4427 driver

thank you

 

[pergo]

update:

this is a schematic from Audison MOSFET section (the PWM oscillator is the same of last schematic posted)

http://img222.imageshack.us/img222/8320/audison.jpg

you could see that the original Audison project has 22k resistor from gate to GND, to discharge it.
It doesn't use pull-down resistor after BJT buffer.
...and there is sinewave on drain


ps: Kicker has r170 e r10 of 10k (like your suggest) in parallel of zener 1n5247+1n4148 diode (anti-series). No sinewave.

 

[Sch3mat1c]

What is the output circuit?

Forward converters are supposed to have choke-input filters. If the choke's current is continuous, the transformer's waveform is suppressed, producing a proper square wave. If you have a capacitor-input filter, you will get extremely poor regulation and most likely explode your power transistors trying!

Tim

 

[pergo]

thank you for the reply Tim,

Audison has big choke filter in the input voltage (12V) with 4400uF capacitor.
Audison output circuit is unknown but is a normal AB mosfet amplifier.
Same circuit the Kicker XS100.

My projects always have a 100uH (minimum) choke + 2200uF capacitor for 12V input voltage.

i don't understand this:
"If you have a capacitor-input filter, you will get extremely poor regulation and most likely explode your power transistors trying"



every SMPS, regulated or unregulated i saw or made, have input capacitor.
Some have also as input choke (always in my smps).

thank you


ps: i'm talking about only push pull smps

 

[event horizon]

What Tim means is the choke followed by the capacitor filter following the main transformer Years ago Maplin made an SMPS kit for a pair of 50W amps to go in a car, yet they neglected to include an output inductor. Consequently the switching mosfets had to dissipate a shed load of power Simple remedy was an output inductor (choke) & an extra turn on both secondary windings of the transformer.

I think the reason Tim mentioned it is none of the circuit diagrams you have upload show an output inductor after the switching transformer.

 

[pergo]

ahhhhh ok!

i cut the image because every smps (my, Audison, Kicker, etc.) uses different component at the secondary.
For example, I use AKA KAK diode, others use 4 single diode to220, or series double diode, or the capacitor/inductance value (only 470uF 200V for my tube preamp, but 13200uF for a rebuilt Orion smps).
So, i decide to cut the image


a question.
If i increase the output LC filter (both L and C value), the amplitude of sinewave decrease?
I ask it, so i don't waste time winding new toroid and find big capacitors

 

[event horizon]

a question.
If i increase the output LC filter (both L and C value), the amplitude of sinewave decrease?
I ask it, so i don't waste time winding new toroid and find big capacitors

That i'm not exactly sure about. One thing i will say though is it'll seriously alter the R/C feedback components that you'll need to maintain a stable & fast closed loop on the PSU.

Generally if you decrease the value of the output inductor you can use higher value capacitors & still maintain a fairly fast transient response, but you'll again need to change the resistors & feedback capacitor/s. Increasing both will make the PSU slower in response to an increased current draw from the PSU.

There is an optimum point & this depends on a lot of things including the output inductor DC resistance, inductance as well as the capacitor value in microfarads as well as it's ESR & ESL...It gets complex