Problems using MC34063 as a tracking-preregulator

[bsunisol]

Hello from Germany,
I want to use the MC34063 as a tracking-preregulator for the LT3080. I simulate it with LTSpice. It's working, but not as supposed to.
Firstly the voltage starts to adapt slowly and secondly the output of the MC34063 starts to oscillate at about 0.3sec.
And last but not least, 2V is to much offset. How can I generate an offset somewhere between 0.5V-1V?

Thanks in advance.
Robert.

[Seekonk]

Isn't power connected to the wrong pin?  Haven't looked at one of these in a long time.

[2N3055]

Isn't power connected to the wrong pin?  Haven't looked at one of these in a long time.

Yep wrong pin ... 

[bsunisol]

Isn't power connected to the wrong pin?  Haven't looked at one of these in a long time.

Yep wrong pin ... 

Thanks, fixed it. The ripples are much smaller now, but it's ramping up to 33V at the beginning. 

I searched for the feedback loop in the datasheet. (https://www.sparkfun.com/datasheets/IC/MC34063A.pdf)
On Page 7 they discribe the step-down configuration and on page 10 the formulas like Vout=1.25*(1+R2/1).

How do I get this Vout 0.5V above the LT3080's Vout?

[dannyf]

Cannot help you with the overshot or simulation but would suggest that

1 you use a buck converter module as the tracking regulator.
2. Use a LED (across the linear regulator) plus a photo resistor (on the tracking regulator feedback divider) as a tracking mechanism.

I have had great success with that.

[2N3055]

Find AN32 from Linear technology.. there is a optocoupler based feedback scheme that can be adapted to MC34063... Or any regulator  ...

[bsunisol]

Find AN32 from Linear technology.. there is a optocoupler based feedback scheme that can be adapted to MC34063... Or any regulator  ...

OK got it. You mean the schematic from the attachment, right?
I don't get the difference? The two diods just create a fixed voltage drop of "X times the Forward voltage", right?
But I don't need to use the output of the LT3080, the set pin has the same voltage and is controlled by a µC. I just need to use a voltage divider with a 1:3 ratio, right?

How is it possible get an overshoot with a buck converter?

thanks and regards.

[prasimix]

Take note that dannyf suggest LDR instead of phototransistor (as part of the optocoupler) and with good reason: that is the real equivalent for ordinary voltage divider in the feedback loop. In reality (not simulation) having active part as transistor without negative feedback could generate huge instability. 470pf or more is required between base and collector. This circuit is shown in Dave's EEVblog #329 (starting from 1:20). If you want to go with dannyf's suggestion it's not a big trouble to make such optocoupler by pairing cheap LDR and LED (fix them using black heatshrink tube) or you can use e.g. NSL-32SR3.

[2N3055]

You are right but only partially...

In theory you could have two regulators in series programmed from same source with first one having offset of few volts more to define dropout on second regulator..
That is actually good thing because that way you  can have good dynamic tracking..

But when you have a short on an output what then.. If you set voltage to 24V preregulator is at say 26V..
Make a short at output, and out goes to ground, LM3080 limits the current to say 1A.. that is 26W over it...  Too much.. It will survive, because it is modern fully protected IC..
But wont like it.

If you make preregulator manage voltage drop OVER linear regulator, you limit worst case scenario say 2-3 W.. Not that's better.

Of course, you could also program both regulators  with uC and put additional circuit to limit voltage over linear regulator only in overload...
Engineering decisions... lots of fun...

 

[bsunisol]

You are right but only partially...

In theory you could have two regulators in series programmed from same source with first one having offset of few volts more to define dropout on second regulator..
That is actually good thing because that way you  can have good dynamic tracking..

But when you have a short on an output what then.. If you set voltage to 24V preregulator is at say 26V..
Make a short at output, and out goes to ground, LM3080 limits the current to say 1A.. that is 26W over it...  Too much.. It will survive, because it is modern fully protected IC..
But wont like it.

If you make preregulator manage voltage drop OVER linear regulator, you limit worst case scenario say 2-3 W.. Not that's better.

Of course, you could also program both regulators  with uC and put additional circuit to limit voltage over linear regulator only in overload...
Engineering decisions... lots of fun...

You're right. I'm trying to build my own PSU based on Dave's µPSU.
It has a current limiting. The same as Dave made. A short on the outputs should be handled by the current limiting curcuit.

Sorry, analog circuits are not my best.....
Ok doing it like in #329 it works, though I don't get why the transistor creates a constant current source.

I found the reason for the overshooting. The initial current through the coil causes the voltage to raise that high. How do I avoid this? With a standalone MC34063 there's no overshoot.

thanks and greetings.

[Kalvin]

Have you tried changing the BSS84 P-MOSFET to a BJT PNP like 2N3906 in order to reduce the voltage overhead?

[bsunisol]

Have you tried changing the BSS84 P-MOSFET to a BJT PNP like 2N3906 in order to reduce the voltage overhead?

I read http://cds.linear.com/docs/en/article/LT3080PR_Article.pdf and http://cds.linear.com/docs/en/application-note/an32f.pdf.
Yes, I tried it like Dave in #329, with a diode and a BJT, two diodes with a BJT, a MOSFET and a fixed 1.25V voltage at the FB pin. In any case the voltage across R1 jumps up to 32V...

What is wrong with this circuit/simulation?

[prasimix]

You can use this simulation as basis for further investigation. It started from 2ms but freely modify it to start from 0s. Add new plot plane and check R3 voltage. You can then replace LTC3864 (that works like charm) with MC34063.

[free_electron]

a MC34063  ? seriously ?

The only good application of a MC34063 is to pound it with a large hammer into dust , douse the dust in gasoline , set it on fire , collect what remains , embed in one cubic metre of concrete, bury that 5 meters down and then put a sign on top of the land saying : the remains of a MC34063 is buried here, enter at own risk.  After about 10.000 years you could grow potatoes on it.

[ZeTeX]

a MC34063  ? seriously ?

The only good application of a MC34063 is to pound it with a large hammer into dust , douse the dust in gasoline , set it on fire , collect what remains , embed in one cubic metre of concrete, bury that 5 meters down and then put a sign on top of the land saying : the remains of a MC34063 is buried here, enter at own risk.  After about 10.000 years you could grow potatoes on it.

[bsunisol]

You can use this simulation as basis for further investigation. It started from 2ms but freely modify it to start from 0s. Add new plot plane and check R3 voltage. You can then replace LTC3864 (that works like charm) with MC34063.

I did and the same result.

BUT, when I simulated the sample file for the MC34063 I noticed a checked box in the settings. See below.
When I activate it, it works. The question now, is this an issue in the simulation/LTSpice or will it happen in real circuit too?