Controlling a NatSemi adjustable smart switcher... digitally?

[aussiepoof]

Hi folks,

So I have to admit that while I'm pretty good in the digital world, I'm really not so hot when it comes to the analog thang!

I have an idea to digitally control the output of a NatSemi simple switcher by inserting an digital pot driving an opamp into the feedback loop of the switcher. The result would be a digitally-controlled switchmode power supply

I have used a similar circuit to successfully control an LM317... great except it has the standard issue with being a linear regulator in that it dissipates too much energy as heat. The digital pot has a 0-5v swing, so I amplify that with an opamp to create a 0-14v swing in the adjust leg of the LM317.

So... is this even possible with a switchmode controller? Are there any potential "gotchas" when it comes to inserting a similar kind of circuit into the control loop? I have done some extensive reading and know that switchmode supplies are easy to tip into unstable feedback states.

Here's a link to the NatSemi site for the LM2595:

**broken link removed**

Thanks in advance for any ideas or links!

Regards,
Paul

 

[justDIY]

I imagine it would work, but the question is, would it work well??

Usually the switcher expects a certain sized inductor for a certain input to output ratio. During the design, input variations can be averaged to give a "good enough" inductor size. But what happens when the both Vin and Vout are variable.

I don't know enough about switchers to help you, but I'm curious to see what others have to say.

 

[aussiepoof]

I imagine it would work, but the question is, would it work well??

Very much my own thought... it could all end in tears!

Usually the switcher expects a certain sized inductor for a certain input to output ratio. During the design, input variations can be averaged to give a "good enough" inductor size. But what happens when the both Vin and Vout are variable.

Most switchers cope with quite a wide range of inputs without trouble, provided you allow for that in the initial inductor choice (lower input voltages require a larger inductor.) Indeed in my playing with them so far they seem immune to inductor choice provided you go for more mH and not less I assume that it just switches on less frequently in this case.

The NatSemi simple switcher comes in an adjustable version, and it's a matter of adjusting just one resistor to get an output that runs from 1.2v or so up to Vin-1.2v without trouble. Maybe I should try a mechanical pot first

I don't know enough about switchers to help you, but I'm curious to see what others have to say.

Thanks anyway!

P.

 

[justDIY]

National has free technical support - you could try asking them. Could try asking Linear and Texas Inst as well. In fact, linear has the free LTSPICE simulator - you could try building your circuit and then adjusting the feedback voltage without changing any of the other components

 

[mneary]

One fairly recent development in switchers is the so-called "synchronous switcher". It uses a MOSFET instead of a diode in the lower leg. Although you can look at it as just an enhancement over the diode, it can also be modeled as a PWM with an average output equal to Vin times the duty cycle, with a ripple equal to the input voltage. The task then is to filter the ripple (not easy, but possible).

 

[aussiepoof]

National has free technical support - you could try asking them.

Yeah... tried that and got this reply:

With regard to your enquiry, I would suggest you to choose another simple switcher LM25576 because it's emulated current mode and ease of loop compensation. If you are interested, you can go to its product webpage for more information and do design with our online design tool WEBENCH. If you still want to use LM2596, you must be very careful about circuit because the feedback loop will impact stability.

Yay for pat answers from a script... sigh

Could try asking Linear and Texas Inst as well. In fact, linear has the free LTSPICE simulator - you could try building your circuit and then adjusting the feedback voltage without changing any of the other components

That's an excellent idea... now to brush of my SPICE skills

P.