4
\$\begingroup\$

I am trying to implement a switching voltage regulator from 7.2V to 5V with LM2596 by HTC Korea. Since I don't need an adjustable regulation, I chose typical application circuit from the datasheet which is for fixed output voltage (seen below).

LM2596 Application

I use 1N5817 (20V, 1A, for testing only) Schottky diode for D1, and different values for L1 including 27uH and 33uH. I also change Cout to 100uF, 270uF, 330uF, 470uF and 1000uF (16V and 25V). No matter what I did I couldn't change output voltage. It's always around 1.25 volts, the value does not change if I remove D1, L1 or both. Changing L1 or Cout also does not affect output voltage. If I remove Cout and measure output voltage without a load, it's around 4V.

What am I doing wrong? Is there anything that I am missing here? I also attach the pictures of my test configuration on breadboard below.

LM2596 BreadBoard LM2596 BreadBoard LM2596 BreadBoard

\$\endgroup\$
  • \$\begingroup\$ Is pin 5 grounded on your breadboard? \$\endgroup\$ – markrages Dec 22 '11 at 0:40
  • \$\begingroup\$ No, but it's inverted. According to datasheet, unless input 5 is high, the regulator enabled. \$\endgroup\$ – Emre Yazici Dec 22 '11 at 0:44
  • 3
    \$\begingroup\$ Solderless breadboards are absolutely horrible for switching applications. \$\endgroup\$ – Connor Wolf Dec 22 '11 at 6:53
  • 2
    \$\begingroup\$ Amusingly enough, there was a great column by Bob Pease on issues similar to what you are having. See here: national.com/rap/Story/0,1562,8,00.html \$\endgroup\$ – Connor Wolf Dec 22 '11 at 6:54
  • \$\begingroup\$ Also, you might want to look at the original nation datasheets for that part. The version you have looks to be a cheap knockoff of the real device, which is made by National Semiconductor. Considering that HTC don't even have the courtesy to prefix their cloned parts differently, I wouldn't be too confident in their quality. National's page is here: national.com/mpf/LM/LM2596.html \$\endgroup\$ – Connor Wolf Dec 22 '11 at 6:58
6
\$\begingroup\$

The fact you always get 1.25V sounds suspicious to me. This is often the Vref voltage for adjustable regulators. Checking the datasheet for the LM2596 says it's 1.25V also.

Are you sure you don't have the adjustable version?

One way of finding out easily is to set up your circuit as if it was and see if it works. Add the two resistors in the diagram below - R1 should be 1k and R2 3k for 5V.
The feedforward capacitor across R2 (Cff) shouldn't be necessary with this setup, but you can check TI's better datasheet for info on selecting one for situation when it is necessary (over 10V out or extra low ESR on output)

LM2595 Adjustable

\$\endgroup\$
  • \$\begingroup\$ Excellent answer. The Korean datasheet is a copy/paste of the National/TI one, even the scope shots are the same. \$\endgroup\$ – markrages Dec 22 '11 at 1:44
  • \$\begingroup\$ The exact number that is written on IC is "LM2596-ADJ", so it seems it is adjustable version. But are they different? What I understand from datasheet they are the same. Without R1 and R2 it uses inernal resistor to regulate output. Or am I completely wrong? \$\endgroup\$ – Emre Yazici Dec 22 '11 at 1:44
  • 2
    \$\begingroup\$ Right, but the ADJ version has no internal resistors. And no way to guess your intention is a 5V output. \$\endgroup\$ – markrages Dec 22 '11 at 1:48
  • 2
    \$\begingroup\$ @EmreYazıcı - Yep they are different. As Mark says the ADJ version needs setting with external resistors, whereas the fixed version has these present internally. If you add the resistors suggested you should see 5V (or using the formula at the bottom of the diagram you can set it up to output a different voltage) \$\endgroup\$ – Oli Glaser Dec 22 '11 at 1:52
  • 1
    \$\begingroup\$ I used 2K7 for R2 and 900 ohm for R1. The output is 4.90V as expected from the formula: 1.23 x (1 + 2700 / 900). I didn't face any problem with the solderless board (with or without load the output voltage is healthy). I am just using solderless board for testing. See the picture: img855.imageshack.us/img855/4411/lm25961.jpg \$\endgroup\$ – Emre Yazici Dec 24 '11 at 18:46
5
\$\begingroup\$

Since you are using a National part number, allow me to quote a Bob Pease column:

Last week, Wanda Garrett, our senior applications engineer for amplifiers and regulators, got a phone call from an unhappy customer. He had used one of our ICs to design a switching regulator, and it didn't work well at all. The output had glitches and burps and excessive ripple and noise. The regulation was poor, the loop stability was rotten, and the efficiency wasn't even very good.

After a lot of inquiry, Wanda discovered that this person had built up the switcher on one of those solderless breadboards. OHHH!! Patiently, Wanda explained, that is exactly what you expect when you use one of those solderless beasts. The inductances are awful, the capacitances will cause crosstalk between adjacent buses, and if you try to build a switching-type regulator, of course it will work badly. And the customer replied. "SMWISICDI."

\$\endgroup\$
  • \$\begingroup\$ This hint is definitly valuable for me, but it seems I have another problem with my configuration. \$\endgroup\$ – Emre Yazici Dec 22 '11 at 1:47
  • \$\begingroup\$ "Show Me Where It Says I Can't Do It" :) \$\endgroup\$ – abdullah kahraman Dec 22 '11 at 6:04
  • \$\begingroup\$ The breadboard is OK at 150KHz. \$\endgroup\$ – abdullah kahraman Dec 22 '11 at 6:07
  • \$\begingroup\$ @abdullah kahraman - No, it really probably is not. Even if the manufacturer claims otherwise. \$\endgroup\$ – Connor Wolf Dec 22 '11 at 13:57
  • \$\begingroup\$ @FakeName - It will work, I didn't say it will be efficient or clean on the output, or no cross-talk or such but it will show the programmed voltage on the output. He should definitely design a PCB that suits SMPS design rules. \$\endgroup\$ – abdullah kahraman Dec 22 '11 at 14:06
4
\$\begingroup\$

While solderless breadboards can have issues, I don't think that explains what is wrong. You may have it wired up wrong. It's hard to see in the picture since we can't seen the legs of the critical part. Is the diode the right way around?

However, one thing that jumps out immediately is that you are using electrolytic caps. That might be OK for bulk storage, but there should be at least 10 µF or so ceramic immediately on the input and the output.

Also, have you checked that you really have the 5V fixed version? Have you tried it with a genuine National part, not some knockoff of suspect origin?

What!?

You told us you were using a fixed 5V regulator before, but now you are saying it is the adjustable version. Of course that will produce something like 1.25V. That is probably the fixed voltage the feedback is compared to. For adjustable regulators, you have to add a voltage divider from the output to the feedback pin such that it results in the fixed feedback voltage when the output is at the desired voltage. Let's say the feedback regulation voltage is 1.25V and you want 5V out. That means you want to divide the output by 4. A 3 kΩ resistor from the output followed by 1 kΩ to ground should work.

\$\endgroup\$
  • \$\begingroup\$ I have uploaded additional pictures to my question. The diode is connected to GND and 2nd pin. The regulator is not fixed, but can be used as fixed, the diagram from original datasheet for fixed configuration. It's from HTC Korea and have the same characteristics with the NS one. The oriinal datasheet link of mine can also be found in the question. \$\endgroup\$ – Emre Yazici Dec 22 '11 at 0:57
  • \$\begingroup\$ I previously misunderstood the "Fixed" and "Adjustable" are just different configuration of the same component. I have already mentioned about that under Oli Glaser's answer. Thanks for additional clarification. \$\endgroup\$ – Emre Yazici Dec 22 '11 at 19:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.