variable DC DC converter

[dmta]

I want to make a variable DC DC converter using lm2678. The output voltage and maximum current should be adjusted using digital methods (no pots). I have thought of the following circuit. I cannot yet implement it as the inductor will take some time to arrive.

Will this work? Do you have any other good designs ?

Regards

[Marco]

I know very little about DC DC converters, but I know this isn't a good idea ... the DC DC converter normally adapts pulse width smoothly, you are forcing it to turn full on/off and with a significant delay at that (LM358s as comparators are not going to be fast).

IMO you need a PWM controller which allows an external reference and the current "limit" should be handled differently (depends a bit on the IC how, it might have a disable pin which a comparator can drive for instance).

Do take into account that if you limit current like this you only limit the current the converter delivers, you don't limit the current the output capacitor delivers ...

[nickm]

You should simulate it is LTspice.  TI has a spice model that you can use for the 2678 and you can use a voltage controlled voltage source for the high side current sense.  The way you have it set up the feedback comes through all those opamps so if they are net behaving you could have problems.  It might work in steady state, but the real test is to see how it behaves under overcurrent and all the abnormal conditions.  Also the current loop looks like it will have an influence on the output voltage even if the current is less than the limit.  You might want to use a comparator or blocking diode to change that.

[Zero999]

The pin numbers for the power on one of the op-amps is wrong.

[dannyf]

Will this work? Do you have any other good designs ?

1. what does the datasheet say about it?
2. I would suggest that you take a look at the datasheet, particularly how the voltage on the output follows the voltage on the FB pin and how that's done in your design.
3. good luck.

[penfold]

In principal it should work... however, you will more than likely run into many stability problems.

Now, there's nothing wrong with what you're doing but it it is worth giving it a think on whether the ease of use of the lm2678 is going to be compromised by the external circuits you're adding along with their idiosyncrasies causing problems.

[Marco]

Something like the SG/UC/LM3524 has the error amplifier inputs external, this allows you to use your own reference voltage (from a DAC).

[dannyf]

Now, there's nothing wrong with what you're doing

Let's say that you are at an equilibrium. There is a small disturbance on Vout -> it goes up a little.

What will happen to the output of U1B / FB, and subsequently Vout?

There is where reading the datasheet would be tremendously helpful.

[dmta]

Hi all,

Thank you for the comments
It looks like this is not a good idea. So how do I use the lm2678 as a tracking pre regulator?
Thanks

[codeboy2k]

... So how do I use the lm2678 as a tracking pre regulator?

You use the LM2678-ADJ and set the feedback resistors as needed for your desired output voltage.
I would not try to do all that stuff with the opamps for your tracking pre-regulator. 

The datasheet for the LM2678-ADJ has the feedback resistors R1 and R2 outside, and suggests an R1 (bottom) value of 1K, and has calculations for the R2 value to set the voltage.

So I would use 1 of 2 methods: (1) assuming you have a micro controller , then use a digital pot for the top resistor, and set it in a few discrete steps.  Or (2), use digital switches and switch in some fixed resistors at discrete steps (like transformer taps in the old linear PSUs). The digital pot method would actually allow you to track it better, though, you could program the pot calculation into the micro and set it for the correct R2 (top) resistor needed for X volts above the setpoint.

If there is no microcontroller, then use comparators with your set voltage to get 4 steps or so, and use digital switches to switch in a top feedback resistor at discrete points, like 5V, 10V, 15V, 20V.. etc.

Now, according to the datasheet, there are different inductors needed for this, and the one you picked is actually the smallest.  For those voltage steps I listed, you need 10uH, 15uH and 22uH.  Your diagram shows 10uH, but I would use 22uH. I think 10uH will saturate early and won't be able to deliver the full load current.

I did the calulations for you in Smath Studio.  The first one is with max current =4.5A and max input voltage =40V. The second one is from your circuit, Imax = 5A and Vin=25V.  In both cases you can see the R2 (top) resistor needed and the inductor needed.

[dannyf]

It looks like this is not a good idea.

Good idea and bad (=fundamentally wrong) implementation.