smps problem, simulation vs practice, need advice

i have a SMPS power supply, very similar to the one found at page 30 on this document:


my goals are to obtain:

flyback configuration

Here is my circuit



- I added primary and secondary ESR, based on the measurements i made from my inductance

when doing a spice3 transient simulation, i get this:



green is Vout, blue is Iprimary and red is Isecondary. The primary current goes up to 0.9A in this simulation, which is well within the maximum allowed by my MC34063 controller

So everything works fine in simulation; i use a current source as my load to have better control, but same result with spice resistance. I put initial condition to get to steady state faster otherwise simulation takes a lot of time.


My goal is to have a ratio of 1:10 on my flyback transformer to boost the output voltage, since the chip cannot support an output voltage of more than 40V directly
In practice, i use an EP7-3E6 ferroxcube, with 2 windings.
the main specifications are:

• Ae Effective Cross Scetion Area: 10.7mm²
• Core Material Grade: 3E6
• Core Size: EP7
• Effective Magnetic Path Length: 15.5mm
• For Use With: -
• Inductance Factor Al: 5.8µH
• SVHC: No SVHC (17-Dec-2015)

According to some theory, the Al factor of 5.8uH means the henries per turn squared: 5.8uH/N^2, which means at the primary, i have sqrt(145/5.8) = 5 turns.
at the secondary, i have sqrt(13000/5.8) ~= 50 turns

in practice, the 145uH and the 13mH were measured, after i did my windings manually, with 6 turns with 30 AWG at the primary, and 60 turns of 38 AWG at the secondary. So basically, it seems close enough to the 5/50 turns i shouldve needed according to the specs.

The problem is that when measuring the output voltage, with the required load for the 3mA current i am expecting, i get some much lower voltage, like ~50V. Also, i noticed that the chip MC34063 heats a lot, which makes me think that the current goes beyond the expected 0.9A in the primary, and also above the maximum of 1.5A.

Since everything else seems to be modeled correctly in my spice simulation, i think the problem is related to my flyback transformer. I am new to magnetics, so i would like to have the opinion of someone who knows the main problems you can get when attempting this sort of circuit.

Thanks!

pasau said:

Vin=9V
Vout=180V

Click to expand...

My goal is to have a ratio of 1:10 on my flyback transformer to boost the output voltage,

Click to expand...

180 / 9 = 20

Can you confirm your desired step-up ratio? 10 or 20?

Thank you Brad.

the step-up ratio for my flyback transformer is aimed to be around 1:10, but the SMPS itself has a 9V input and a 180V output.

The converter ratio depends on the PWM ratio of the controller and the feedback which controls this PWM.
The transformer is used to isolate the output of the controller from a high voltage it cannot support at its output, but is not directly related to the input and output of the whole system.

This is basically a buck-boost configuration, with a coupled inductor instead of a simple one.

What is your switching frequency, or "on" time ?

It looks very much like you may not have enough turns on the transformer and it may be going very deeply into saturation.

Your IC has a pin labelled DRC. Does this mean diode-resistor-capacitor snubber? If not then your primary may need a snubber on it, due to inductive kick producing severe voltage spikes. It is liable to destroy nearby components, including your IC.

I have a simulation running and it confirms your transformer is okay with a 1:10 ratio. A 38% duty cycle steps up voltage to 180V at 3mA. The action is similar to a boost or buck-boost converter as you stated.

The DRC pin is the collector of an emitter follower which drives the power BJT. It is available so that the designer can control how deeply into saturation the output BJT goes. The simulation in the OP has it connected such that the driver and the output BJT will act as a darlington together. Connecting DRC to a fixed positive bias (like 9V) would allow the output BJT to saturate, probably increasing efficiency.

I also suspect that reverse recovery in the output diode is contributing to heating. What diode did you use in the circuit?

Warpspeed said:

What is your switching frequency, or "on" time ?

It looks very much like you may not have enough turns on the transformer and it may be going very deeply into saturation.

Click to expand...

Rough calculations making some assumptions indicate that the primary turn's calculations may be way too low, by a factor of 10 at least.

E-design said:

Rough calculations making some assumptions indicate that the primary turn's calculations may be way too low, by a factor of 10 at least.

Click to expand...

I agree, the turns count needs to be increased a great many times, and its not just a slight error.

We need to know the operating frequency so we can guide you in the right direction.

Eyeballing the plot in the first post shows roughly 30khz

O/k, we now have lift off....

Vin = 9v
Ton = 16uS
Flux = 0.22T
Ae = 10.7mm squared

Minimum primary turns (9v x 16us) / (0.22T x 10.7mm) = 61 turns

That would be required at ANY required output power level.

How much the current ramps up during each 16uS depends on the inductance.
And that depends on the air gap.
The bigger the gap, the more output power we get.

Required output power 540mW, so lets aim for one Watt to be sure.
9v at 1w is 111mA.
Current needs to peak at four times that so that average will be 111mA
Say 444mA.

mH = 9v x 16uS / 444mA = 0.324 mH
We need 324uH from 61 turns
Al = 324 / 61 x 61 =.087uH per turn or 87nH per turn

Looking up the data sheet for EP7 core
https://www.ferroxcube.com/FerroxcubeCorporateReception/datasheet/ep7.pdf
There are no gapped figures for the high permeability 3E6 that you have.

But if it was something like 3C94,
A total gap of 0.27mm would give an Al of 63
And total gap of 0.15mm an Al of 100

We need Al of roughly 87 so maybe about 0.2mm total gap ???

Try using a piece of paper about 0.1mm thick between the core halves with 61 turns, and see if you can get about 324uH.
The results with 3E6 material will be different, but try that first and see how you go.

I don't think the core will be large enough for the total amount of (prim + sec) winding's.

Primary current is only about 100mA average, probably less, so 0.25mm wire should be sufficient for that.
Sixty primary turns should leave about 3/4 of the bobbin area still free.

Secondary turns are not that critical anyway, just put on as many turns as will fit of the thinnest wire available. Should work fine.

Warpspeed said:

O/k, we now have lift off....

Vin = 9v
Ton = 16uS
Flux = 0.22T
Ae = 10.7mm squared

Minimum primary turns (9v x 16us) / (0.22T x 10.7mm) = 61 turns

That would be required at ANY required output power level.

Click to expand...

This is a good start. From here we can look at the total energy the core can store, which is Bsat²/(2*µr*µ0)*Ve. If ur is 1500 and Bsat is 0.3 T and Ve is 165mm^2, then the max energy is about 4uJ. At 30KHz, the core can transfer up to 0.118W, under ideal conditions. This can be increased by adding an air gap to decrease the effective µr, and/or increasing the switching frequency. Or increasing the core size, of course. I'd recommend doing some of each.