EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: jwhitmore on March 08, 2016, 09:44:00 pm
-
SW Head with a soldering Iron, which might explain something ;-)
I've got a LTC3603 DC DC Converter giving me a 5V Rail from a 12V Input. I used it because I wanted a bit of current and just used the datasheet to create a basic step down. Now the circuit is working most of the time but occasionally I solder up a unit which fails when you apply 12V Input to the circuit. It doesn't give any output at all.
When I get this problem if I power down and short out the Tantalum Caps with a piece of wire and start up again it's all good. However the unit will occasionally fail and I don't know why.
I've gone over the datasheet to see if I missed something, and if I did miss something I'm still missing it. Perhaps it's my soldering of a very fine pitch component (The LTC3603) if that's what people suspect at least I'd have an answer to the problem..
Thanks for any pointers to fixing the issue.
-
1. Schematic
2. Physical layout.
-
Recently migrated to kicad but this is one of my old eagle designs. I'll attach an image of the schematic but the layout is a two sided eagle PCB Layout so I'm not sure if that's something I can upload here and expect people to be able to view?
-
I am guessing U1 is an inductor - usually L1.
In general, most problems in SMPS come from layout problems. These circuits are very sensitive to the PCB parasitics that can easily cause failures or poor performance. Before I spend too much time analyzing the schematic - take a photo of the top/bottom of the PCB.
What type of caps are you using on the input / output?
What are D1 and D3 doing?
Trying to figure out why the SYNC pin is tied to the FB pin - looking at the data sheet.
-
Failure to start reliably is a sign of input or output capacitor problems usually.
Either they are too far away or have too much ESR/ESL, upsetting the control loop.
I'd suggest you read the datasheet and run through the calculations for your application, and verify all the values.
If all looks okay there, have a look with a scope on the output for turn on stability. If it's trying to instantly charge the output cap, it's probably collapsing the input supply, it should be soft starting (which is what happens if you discharge the output caps prior to powering).
I note you don't have any HF decoupling either - probably not the root cause, but it never hurts.
Then next place I'd look would be the layout (as suggested). Keep input cap and output cap loops really small and things will be happy.
-
Took a screen dump to give you an idea of the layout I'm using. The grid size is 1mm, if it shows up. The Inductor is on top between two output Tantalum Caps. I've been over the data sheet what seems like too many times but I'll have another go. I'll have to look into HF Decoupling as like I say I'm a SW Engineer I'm not sure about that subject. That and ESR/ESL
I guess I might be learning something, well I hope so
-
Where is your feedback divider R3/R4 with filter C5?
Your input cap C1 needs to be much closer to the Vin along with an HF coupling cap maybe in the range of .1uf-.01uf .
I am assuming D1 is for reverse polarity protection, if so - get it out of the way of the important stuff. Also, that diode will get warm if you are pulling some curretn out if this thing.
You need better grounding.
What is D3 doing?
Where is your loop compensation? C2/C3/R2
if you changed the schematic - I will wait until you post a matching schematic and layout - useless to investigate otherwise. Take the time to 'smash' the parts on the layout and arrange the designator so they can be more easily read. For those that can help, don't make it harder than it should be.
-
Ditto everything noted above.
Looks primarily like a layout issue.
Copy the manufacturers layout recommendation as a first step.
Once you understand the constraints and compromises around SMPS circuits, then deviate.
-
Layout looks quite bad. No ground plane, big loops to both input and output caps. LTC3603 datasheet doesn't seem to have any layout examples. Check e.g. LTC3630 datasheet for a layout example and note the ground plane, a lot of vias to ground plane and the caps very close to the pins.
Read carefully the layout checklist in the datasheet.
Are your tantalum caps suitable for the high ripple currents needed? Is the input cap also tantalum? Note that the datasheet example is using ceramic for the input and ceramic + tantalum for output to get low enough ESR.
-
Thanks a million for all that. Man that is obviously so bad :-(
Was migrating to kicad anyhow so I might try this all again, hopefully putting all of your advice into practice.
One advantage of Eagle CAD is that it allows the creation of a PCB block, (that power supply) and including it in numerous designs. Of course you can still make a mess of it, more times.
Thanks again for your help.
-
I used a similar chip (LTC3600) a while ago - The RUN-Pin has to be fitted with a resistor if the Input-Voltage is above 12V!
Quote from the Datasheet: "When tying RUN to more than 12V, place a resistor (100k to 500k) between RUN and the voltage source."
In my case, I was only about 1V over 12V and destroyed the chip. When testing a second board, I discovered the threshold to be about 12,6V. Beyond that the chips never worked again.
-
That's for 3600, which has 12 V absolute maximum for RUN. For 3603 it is 16 V. But even 16 V is quite low for a 12 V system. Especially with low ESR capacitor in the input. There can easily be 20+ V caused by the inrush current to the input cap. So you may have broken your chip. There is even a warning about it in the 3603 datasheet.
-
Should be a simple add of a resistor to limit the current.
Look at various layout examples from other converters. Even though they are different, you will start to pickup valuable tips on SMPS PCB layout. I always treat the components as HALF the circuit and the PCB as the other HALF on SMPS designs - it's that important. I started about 2 years ago with SMPS designs and had a rough start. I never guessed how much the layout would really steer the ship.
It did not take long to get my designs to start-up, but it has taken a long time to get good performance. It will take even longer to get great performance most of which is learning how to measure SMPS circuits.
Have fun.....
-
RUN pin can easily be protected by a resistor, but protecting Vin is not as easy. There are application notes about this problem. You can use a high ESR electrolytic or a cap + a resistor to dump the peak voltage in this transient. Or you can clamp it with a TVS diode etc.
-
I'm going over this again. In my circuit I've put a protective diode on both the input and output. The intro on the data sheet states that a protective diode isn't needed but I've no idea if this is the input or output it's referring to:
It operates from an input voltage
range of 4.5V to 15V and provides an adjustable regulated
output voltage from 0.6V to 14.5V while delivering up to
2.5A of output current. The internal synchronous power
switch with 45m? on-resistance increases efficiency
and eliminates the need for an external Schottky diode.
-
Basic buck converter needs a diode for providing current while the switch is open: https://en.wikipedia.org/wiki/Buck_converter
LTC3603 doesnt't, since the same functionality is dealt with internally (using two FETs). It has nothing to do with your input or output protection diodes. Are you worried about reverse polarity or why are you using the diodes? They will decrease efficiency and produce heat.
-
Thanks a million for that. Yes I'm worried about miss connection of the input 12V Supply and then the power supply is being used in a board which can be either USB Host or USB Device Mode. In Host mode it's simple we supply the USB 5V power, but in Device mode the board can be self powered or USB Bus powered. In which case there will possibly be 5V coming back at the LTC3603.
Being a SW Engineer and still looking for a Rosetta stone for really understanding electronics data sheets I stuck in the output Cap. I could actually test it out and see what effect USB power has, I Must do that!