Non-Isolated DC/DC

Forum Post: LM3481 INPUT/OUTPUT Capacitor Selection

takishin — Fri, 24 May 2013 11:14:00 GMT

Our customer has some questions about LM3481.
I'm posting the questions for a customer:

++++++++++++++++++++++++++++++++++++++++++++++

(1)

The datasheet(SNVS346E) says ;
"The input capacitor should be capable of handling the rms current. Although the input capacitor is not as critical
in a SEPIC application, low values can cause impedance interactions. Therefore a good quality capacitor should
be chosen in the range of 100uF to 200uF."

How to decide the capacitance in the range of 100uF to 200uF with LM3481?
(Are there calculating formulas for the capacitance of INPUT/OUTPUT capacitors ?)

(2)

The datasheet(SNVS346E) says "Use capacitors with low ESR and ESL".

Specifically, what ohm is better for low ESR and ESL of the capacitor with LM3481?

++++++++++++++++++++++++++++++++++++++++++++++

thanks in advance.

Forum Post: RE: TPS63001 accuracy

Chris Glaser — Fri, 24 May 2013 11:07:00 GMT

It is important to measure the voltage properly--with a calibrated meter and at the sense point of the device (FB pin). Otherwise, IR drops on the board can cause a false reading.

Per the link you sent, there is an additional 0.5% accuracy due to line and load regulation, each.

Wiki Page: TPS63001 accuracy

Chris Glaser — Fri, 24 May 2013 11:05:00 GMT

Q: TPS63001 accuracy by 7382

What is the specified accuracy of the TPS63001 and TPS63002? It is not specified in the data sheet- only the adjustable feedback tolerance for the TPS63000.

A: Re: TPS63001 accuracy by 244698

The accuracy for TPS63001 and TPS63002 is:

Feedback 1%

Line 0.5%

Load 0.5%

This is in PWM mode.

Bill Johns

A: Re: TPS63001 accuracy by 1129650

Those values are over the temperature range -40C +85C with power save mode disabled.

Forum Post: TPS54383, 24V- 12V burns at no load

Stephan Wille — Fri, 24 May 2013 09:42:00 GMT

Hello,

we are developing a DC DC converter with the TPS54383. In first test runs the circuit worked fine but lately a TPS54383 burned down two times at no load. We have used a boot capacitor of 33nF and a boot strap resistor of 3Ohm.

If we measure the SW1 voltage we have voltage peeks up to 31V with a Duration of approx. 10ns. In the datasheet the max. voltage for SW1 is rated with 31V. So first the question if this voltage (supplied by the boot capacitor ?) is a Problem for the TPS as it's rated up to Vcc max. 28V ?

What could be a possible problem that we have killed the TPS ? We are using a 47uH coil.

Best regards,

Stephan Wille

Forum Post: LM5010A NC-Pins

Stefan Dreyer — Fri, 24 May 2013 08:27:00 GMT

Hi guys,

I wanna use the LM5010A in my application. Concerning the NC pins the datasheet states:

NC - No internal connection. Can be connected to ground plane to
improve heat dissipation.

In the evaluation module, I saw that you connected the pins with their neighboring pins, which massively simplifies the layouting.

Just to confirm: Is it safe to do so?

Cheers,

Stefan

Forum Post: RE: LM5119

EricLee1880267 — Thu, 23 May 2013 20:41:00 GMT

Electrically, your connection is right , but by connceting SW node to GND, LM5116's dead-time might be affected, the dead-time have to be guaranteed by UCC27211 operation. Diode emulation will not work well, FPWM is recommended.

Regards,

Forum Post: RE: TPS53219/ TPS53219A

Nancy Zhang (CPM) — Thu, 23 May 2013 19:19:00 GMT

Hi Kai,

Please contact Sue Xu(xiao_xu@ti.com) directly. She is supporting this part now.

Thanks,

Nancy

Forum Post: RE: TPS61251 problem in 100% duty cycle mode

Chris Glaser — Thu, 23 May 2013 18:32:00 GMT

One other thing you might try is adjusting the value of your feedforward capacitor. Since you increased your FB resistors (too large, by the way), you should decrease the Cff to keep the compensation zero the same. Increasing it might also have some benefit for this noise, so I would test out both.

Forum Post: RE: LM27402, LM22678, TPS63700 - ψJT

S.Satoshi — Thu, 23 May 2013 17:16:00 GMT

Hi Chris-san,

Thank you for your support. Please update when you have information for these devices.

Best Regards,

Sonoki

Forum Post: RE: TPS5120 prebias start-up

K.Hamamoto — Wed, 22 May 2013 23:02:00 GMT

Hi Nancy,

Thank you very much for quick response.

This is the answer that I wanted.

Regards,

Forum Post: RE: Shut down of LM3481 not working

Akshay Mehta — Wed, 22 May 2013 20:18:00 GMT

Hello Rob,

I apologize for the delay. Here are the files.

Regards,
Akshay

(Please visit the site to view this file)

Forum Post: RE: TPS6122x Minimum ON time

Chris Glaser — Wed, 22 May 2013 17:39:00 GMT

This is a good device for your power needs. Your customer will just need to test it, on the EVM, to see if its noise performance is acceptable for their system. Another device to look at is the TPS61070.

The TPS61220 is not a fixed frequency device. Per the D/S, it regulates the ripple current in the inductor. So, it changes its frequency to do this.

It also contains a power save mode, such that the effective operating frequency decreases as the load decreases. It skips switching pulses as needed. Figures 21 and 22 show typical operation.

There is no specific minimum on time. It will do what to needs to do to keep the output regulated, per its control topology. There will be some ripple of course, as it is a switching regulator.

If you like, I can move this post to our models forums where they can explain how the software tool came up with the numbers it gave you.

Forum Post: RE: TPS55340 not switching; input equals output

Matthias Joerg — Wed, 22 May 2013 15:44:00 GMT

Hi Anthony

Thanks for your response.

I’m working together with Goran on this boost converter:

I checked the FB network; component values are correct (187k / 10k for an output of 24V). If the switcher starts up correctly, the FB voltage is approx. at 1.2V.

We see these large noise spikes on the output and the FB pin; we hopefully can reduce this by optimizing the layout according to your recommendations and the example we have from the EVM. However, also on the EVM there is quite some noise on these signals (see scope pictures in section 3).

Let me quickly summarize before I might ask additional questions:

Here, we have the schematics of the circuit:

1. Design no.1:

We produced a first version of the PCB (no.1, old design, 4 layers), which is starting up with any load. It was designed following the proposal of the webbench designer (Vin 10-14V / Vout 24V, 1.5A inc. ripple filter). The webbench design was slightly adapted towards more efficiency (optimization dial factor 4) and then regarding components which we already had on stock. This design works fine, even with significantly more noise on the FB pin in comparison to TI's EVM:

On two additional layers there are GND planes that completely cover the area under the switcher components.

The following scope pictures show the startup procedure for two different load cases:

- CH1 (yellow): VIN
- CH2 (green): SW
- CH3 (blue): FB
- CH4 (purple): VOUT

Lower half of the scope screen shows details (@2us/div) of the upper half (@2ms/div).

Load 15mA:

Load 370mA:

2. Design no.2:

We wanted to optimize design no.1 regarding the layout of the GND plane: It should not only have local GND planes under the switcher components but cover the whole PCB area (165x115mm). This was the only change regarding the switcher; we did not rearrange the components or change component values:

This version no.2 did not start at all: Output voltage remained equal to input voltage. We then tried to play with the switching frequency and the compensation network: At a higher frequency (Rfreq = R37 = 100k instead of 127k) the circuit starts up with very light load (~15mA), but still refuses to start with higher load (~65R, i.e. ~370mA). The values for the comp. network we have chosen according to the Excel spread sheet provided by TI.

The following pictures show the startup procedure:

Load 15mA:

Load 370mA (not starting, VOUT remains at VIN):

When starting up with light load (15mA) and then attaching the higher load (370mA) the circuit works and looks like this:

Load 370mA (after starting with light load):

Besides the startup problems we see a clear difference in the startup sequence at light load: While design no.1 starts up within approx. 5ms (VOUT from 0V to 24V), the circuit no. 2 needs approx. 7ms and shows a large initial overshoot of VOUT.

Since we produced a couple of these PCBs we also found one that starts up with the 370mA load, using the initial components regarding the comp. network and Rfreq (= 127k). However, startup is delayed even more:

Load 370mA (starting with delay, approx. 14ms):

3. Comparison to EVM:

We took TI’s EVM TPS55340EVM-017 as provided (working at ~600kHz) but added the ripple filter we use in our designs. The EVM shows similar startup behaviour than our design no.1. Noise on FB pin is also present but considerably smaller than in our designs.

Load 15mA:

Load 370mA:

4. Different chip versions:

While comparing our two designs and the EVM we found that each version has a different batch of TI’s TPS55340.

Design no.1:

Design no.2:

EVM:

We do not expect that this really matters but it was a difference we noticed.

5. Questions:

a) Could it be possible that only the change in the GND planes makes such a difference? Maybe even design no.1 is at the edge of stability? We are quite confused because we didn’t succeed to fix the startup problems with design no.2. Do you think, it is mainly because of the PCB design? How can we be sure that the next design attempt will be successful?

b) How do you explain the difference in startup time between our two designs?

c) How can the compensation network be designed properly? We know the formulas and TI’s Excel spreadsheet, but if possibly the layout has such a big influence, is there a way how the network can be optimized by practical (measurement) means on the PCB?

d) We noticed that – regarding the compensation network – there are differences between the results from webbench and the results from the Excel spreadsheet. What shall be taken as a reference? How critical is this compensation network? How well designed and tested are the webbench outputs?

e) Does the additional ripple filter need to be considered when selecting the component values for the compensation network? If yes, how? Do recommend using such a filter at all?

f) As can be seen from the Excel spreadsheet a 10uH inductor is rather too small for operation at 380kHz (however, this was webbench’s recommendation). In your opinion, would it make sense to keep the inductor’s value but increase the frequency to e.g. 600kHz? Would this help to make the circuit more stable?

g) Would it make sense to increase the startup time using a larger soft-start cap? Or even use a separate voltage supervisor on the enable pin in order to further delay startup to a point where VIN has reached almost its nominal value?

Sorry for all these questions but we really want to better understand what went wrong with design no.2 and be sure that our next iteration will be successful.

Thanks and best regards,

Matthias

Forum Post: RE: TPS40210 Current Output Too Low / Early Overcurrent Shutdown

Anthony Fagnani — Wed, 22 May 2013 14:38:00 GMT

Thanks for the update Thomas.

A couple notable differences between your design and the reference design are the switching frequency and the extra circuit on the gate drive. The reference design uses a slower switching frequency 100kHz. Also the addition of the BJTs Q1 and Q3 help to speed up the turn on and turn off of the FET. Comparing the FET to the one you are using, it has very similar specs in regards to the gate charge. Both of these will help to reduce switching loss and typically have more of an effect on efficiency than false OCP tripping.

Another thing to try then would be to add Q1 and Q3 to see if there is any improvement.

Best Regards,
Anthony

Forum Post: RE: TPS5450 Ringing PH(Switching Node)

Anthony Fagnani — Wed, 22 May 2013 14:29:00 GMT

Hi Fernando,

The ringing is typical when working with the TPS5450 and any other buck converters. Using a diode with lower capacitance certainly helps.With other parts I have had good success with diodes in a smaller package like the PDS560.

Another suggestion to tame this ringing is to add a snubber to dissipate the ringing energy and slow the rising edge. The following link gives a good guide to adding a snubber: http://www.ti.com/ww/en/analog/power_management/snubber_circuit_design.html

Lastly adding a resistor in series with the boot capacitor (C3) can slow down the turn on of the high-side FET. 24 ohms is recommended for the TPS5450.

Best Regards,
Anthony

Forum Post: RE: What if PH pin volatge under specification of TPS5432?

JohnTucker — Wed, 22 May 2013 11:58:00 GMT

There is a transient spec of -2 V for 10 nsec. You have to be very careful about measuring PH node waveform you have to use tip and ring method or use 5 mm gnd lead directly to the ground barrel of your scope probe. Any additional ground loop will cause unwanted noise pick up.

Forum Post: RE: TPS63030

Chris Glaser — Wed, 22 May 2013 10:54:00 GMT

You will need to supply some waveforms of Vin, Vout, and L1/L2 to show what you mean by unstable. Additionally, it is recommended to test with a resistive load. Electronic loads can sometimes cause erratic operation with their active control loop.

Forum Post: RE: TPS40304 Soft Start Feature

Adam Elias — Wed, 22 May 2013 02:23:00 GMT

Thanks for the response and confirming what I was seeing in the lab and in my simulations.

Any plans to get the datasheet updated to prevent future frustration by designers who select this part?

ADam

Wiki: Design Notes

Anonymous — Tue, 21 May 2013 21:42:00 GMT

Forum Post: RE: TPS61221 with LDO TPS78230 or TPS72730 in a Row

Thomas Schneider1 — Tue, 21 May 2013 20:45:00 GMT

Oh, we looked to our waveforms and we confused about that value by starting the LDO. It is described in the datasheet.

See attached, yellow chart is output voltage and bright blue is inrush current. Measured without load. Input voltage is the primary cell at 3.6V.

The TPS61260 is a good device, but we are looking for a qiescent current device in the range of the TPS61221. This value is much higher in TPS61260.

thanks for attention!