Roman Black SMPS with mosfet driver?

[TheBlob]

Hello to all and thanks for this forum.

I've just discovered the wonderful Roman Black SMPS circuit(s) and I've already built the 12v current limited version with some success.
However , after blowing some BC327 in the process, I came to the conclusion that that transistor simply cant sustain much more than 100mA of current without burning up.

And still in the official website (www.romanblack.com) Roman reports about tests done with much higher current, and with the very same transistor.
So I guess it *should* work without blowing up.

In the meantime I tried, as suggested by others, to use a power transistor with with HFE > 50 and fast (BD140, and also a darlington BDX33 ) and , while they get hot (I had to put a small heatsink on them) they work without blowing up.

But my question is: why do they get so hot? As I understand switching power supplies, the main switcher driver should not get so hot, as it happens with power transistor in linear power supplies.

So I thought I could replace the transistor with a power mosfet (P channel eg. IRF9Z34 ?).
How should I modify the circuit to use such a mosfet?

Thank you in advance for any feedback you may provide!

Cheers,

TheBlob

 

[Mr RB]

It's possible you wired the circuit wrong, or some parts value caused it to fail to oscillate properly. In that case the mains sw (BC327) will run as a linear regulator and will cook.

If it is running properly the BC327 will run quite cold even with 300mA or more output.

Do you have a 'scope to check the oscillation?

 

[TheBlob]

Hello and thank you for your reply!

Yes I have an old (but functional) scope, I'll check (on the collector of Q1, right?) and will let you know.

P.S: I'm using the circuit as a lead acid battery charger , with VOut fixed at 13,6V (13V zener + 2 1N4001 diodes), and with a 2Ohm Rs to limit current at about 300mA since I'm trying to gently charge some motorbike batts (9 to 11 AH) while I try to desulfate them.

I'll let you know as soon as I get home and get a chance to look at the signal with the scope

Cheers,

TheBlob

 

[4pyros]

Is it posible your desulfater is feeding back into your SMPS?

 

[Mr RB]

That's unlikely, the moment the load voltage is above the normal threshold voltage everything in the regulator turns off.

But for an application like charging 12v batteries 9Ah to 11Ah I think you should be using a much larger transistor for Q1 and put it on a (small) heatsink.

I've made a 2 amp version of the circuit using a darlignton PNP for Q1 and a few minor parts changes.

 

[TheBlob]

Is it posible your desulfater is feeding back into your SMPS?

Hello!
I can exclude this for a simple reason; the transistor overheats even with a simple load connected, such as a light bulb, and no battery and no desulfator connected.

The amount of heat produced seems to be proportional to the load: with a 100mA load it only gets slightly warm, with 300mA it gets hot, with the heatsink I can still touch it but without it can't be touched without burning a finger.

A quick note: I've taken some pictures of the scope with my cellphone, and I'll post them as soon as I can download to my pc!

Hope this helps!

Thank you,

TheBlob

 

[TheBlob]

It's possible you wired the circuit wrong, or some parts value caused it to fail to oscillate properly. In that case the mains sw (BC327) will run as a linear regulator and will cook.

If it is running properly the BC327 will run quite cold even with 300mA or more output.

Do you have a 'scope to check the oscillation?

Here are the scope images; I took 3 series with 3 different transistors, which are a darlington BDX34C , a BD140 and a BC327-25

BDX34C with 60mA load (a power LED)

BDX34C with 100mA load (a 1,2W 12v bulb)

BDX34C with 300mA load (a 5W 12v bulb, limited to 300mA by the current limiter)

BDX34C with 200mA load (a 11AH battery + desulfator)


BD140 with 60mA load

BD140 with 100mA load

BD140 with 300mA load


BC327-25 with 60mA load

BC327-25 with 100mA load

BC327-25 with 300mA load


As you can see, there is a sort of "second waveform" superimposed with the first, and I had a lot of triggering problems to make them both appear in focus.

The strange thing is the frequency, I had to put the scope on the 0,5 uS range, and since the wavelength si roughly 3 divisions, I get the frequency is about:
1/0,0000015 = 666666Hz !!!

How can the frequency be so high????

 

[ronv]

That is about 10X to high. Check that you have the correct capacitors and inductor.

 

[ronv]

The current rating on the inductor is also important. It could be saturating.

 

[TheBlob]

That is about 10X to high. Check that you have the correct capacitors and inductor.

Hi ronv, I measured again and it seems It's "only" 5x too high...I had made an error measuring the wavelenght (not from peak to peak)
The inductor is a 470uH rated 0,89ohm and 0,43A
I thought that for a 300mA load it was enough...but I also have some spare 1000uH inductors (2,5ohm and 420mA) I could just parallel those and get a or also some 220uH (0,13ohm 1,9A) that I could put in series to get the correct value (almost!).

Anyway, I'd really like to understand why the transistor overheats, since it should not.
Also, for the sake of learning by example, how do you "see" that an inductor is saturating by looking at the scope's waveform?

Thank you for your feedback

Cheers,

TheBlob

 

[ronv]

Smps

Here is a plot of the waveforms. Seems like your inductor should be OK.
Take a look at your waveform on a longer time scale to see if there are some pulses missing. Your "jitter" may be a clue.
If the inductor is saturating I would expect the triangle after L1 to look more like a square wave.
PS I didn't have the correct zener so the DC level may be off a little.

 

[ronv]

PS
What is your input power supply?
Just noticed the lamps as the load. They can have a very low resistance before they heat up.

 

[TheBlob]

Here is a plot of the waveforms. Seems like your inductor should be OK.
Take a look at your waveform on a longer time scale to see if there are some pulses missing. Your "jitter" may be a clue.
If the inductor is saturating I would expect the triangle after L1 to look more like a square wave.
PS I didn't have the correct zener so the DC level may be off a little.

Wow! You just introduced me to LtSpice!
I tried to reproduce the circuit and , then noticed that I had the wrong resistor to polarize the zener (4,7K instead of 15K).
Guess it remained there after some tries and I forgot...
However by simulating the circuit with LtSpice I noticed that as the resistor value lowers, the frequency gets higher, and in fact with 15K i get the right , slower, frequencies.
I then tried it with my breadboard and the scope confirmed this behavior; also, with 300mA output the BC327 gets only lukewarm and doesn't smoke anymore.

Also, the scope shows the square wave is way more precise and "squared"!

By trying various values in LTSpice, I also understood the significance of the two capacitors that regulate the "on" and "off" periods.
I could even get the right operating frequency while keeping the "wrong" zener resistor (4,7K) and increasing by 5 fold both capacitors (22n and 4.7n).

I have to say LtSpice is a wonderful tool, I had never gotten around it since I had tried before its "cousins" (the various SPICE versions that are around), but never gotten to make them simulate anything useful.
LtSpice instead works like a charm!

Thank you ronv, you just opened a new world to me!

My best regards,

TheBlob

 

[Mr RB]

If the frequency is too high the usual cause is either the inductance is too low or the cap values C1 and C2 are too small.

Please see my page again, the waveform here;
**broken link removed**
is critical as it shows the voltage at the top of the zener. During switching C2 discharges C1 by a couple of volts. This provides a finite time that the transistors are OFF. This mimics the operation of many SMPS ICs that use a "fixed off period" which slows the rate of oscillation and gives much faster and cleaner switching times.

Try to adjust the values of Rz, C1 and C2 to get a very similar looking waveform when running at your typical Vin Vout and Iout. The voltage waveform at point Z should have an amplitude of about 2v to 3v so it turns Q2 base hard off by 2v to 3v.

Once you have done that the freq will be much lower and the waveform at the collector of Q1 should be a nice clean squarewave for best efficiency.

 

[TheBlob]

If the frequency is too high the usual cause is either the inductance is too low or the cap values C1 and C2 are too small.

Please see my page again, the waveform here;
< snip! >
is critical as it shows the voltage at the top of the zener. During switching C2 discharges C1 by a couple of volts. This provides a finite time that the transistors are OFF. This mimics the operation of many SMPS ICs that use a "fixed off period" which slows the rate of oscillation and gives much faster and cleaner switching times.

Try to adjust the values of Rz, C1 and C2 to get a very similar looking waveform when running at your typical Vin Vout and Iout. The voltage waveform at point Z should have an amplitude of about 2v to 3v so it turns Q2 base hard off by 2v to 3v.

Once you have done that the freq will be much lower and the waveform at the collector of Q1 should be a nice clean squarewave for best efficiency.

I'll check the voltage at the top of the zener, thank you. (well, in my case, at the top of the 13V zener + 2 x 1N4001 )
Also, I've made some very quick tries with the corrected Rz value and now Q1 seems to get only slightly warm at 300mA (even at overload when the current limiter kicks in).
In the weekend I'm not home and I can't let the circuit run unattended, I'll get some long-run tests as soon as I get home.

Have a nice weekend,

TheBlob