Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

A Good Layout?(SMPS)

Status
Not open for further replies.

Overclocked

Member
Per Data sheet instructions, I Manually routed The Output section (ie, the Inductor, capacitor and diode. According to one of my books, the output section has to be placed a certain way, so I place the output section together per their instructions.

My question is; How do I place the other components? Is the way I have it OK? The rest are just bypass capacitors and for soft start. My ground also connects to power ground in one place.

Note: This is the PCB for my flashlight that Ive asked about in other threads. It uses a boost converter.
 

Attachments

  • Flashlight.zip
    57.9 KB · Views: 163
all the zip file contained were lbr files (eagle, I presume). not particularly useful for commenting on a layout.

to answer your general question - keep the bypass caps close to the the device's gnd and Vcc (or what ever they are called). Softstart is probably not critical but, in general, I'd try to keep the components close as possible.
 
the rest of the eagle files are in there too, but the files are named such that they are incompatible with most file systems, as they contain wildcard characters.

to add to philba's comments ... you want to keep all the high current traces as short as possible, if short is not possible, than make sure they're thick. high current travels along the ground, the switch node and the inductor.

you also want to isolate the feedback circuit from noisy areas like the switch node, diode and inductor. protect the trace that returns the feedback voltage to the chip with grounded guard traces if it must run near a noisy area.
 
Sorry, I am running Kubuntu 7.04 and I used eagle. I used ark to compress the files...

I lib files are needed because eagle; by default; doesnt come with the 2 parts I need. I'll get on Kubuntu ASAP and fix that.
 
OK heres the [hopefully] fixed version. The readme is a text file.
 

Attachments

  • Flashlight.zip
    57.9 KB · Views: 139
Last edited:
forgive me if this was covered in another thread, but if this is for a flashlight, why operate the converter in constant voltage mode (FB tied to ground), instead of constant current mode (FB tied to a current sensing resistor). I recall you mentioned there is no other logic in the flashlight, just the power supply and led? how are you limiting current to the led?
 
justDIY said:
forgive me if this was covered in another thread, but if this is for a flashlight, why operate the converter in constant voltage mode (FB tied to ground), instead of constant current mode (FB tied to a current sensing resistor). I recall you mentioned there is no other logic in the flashlight, just the power supply and led? how are you limiting current to the led?

When FB is tied to ground, Vout = 5V. I drop the rest of the voltage with a resistor (opps! I forgot to put it in...) which also sets LED current. I'll go over my schematic and make sure everything is connected properly.
 
I suppose if you're dropping less than 1.25v across your limiting resistor, then you're better off keeping things as is. however, if you're dropping more than 1.25v, consider constant current mode, it'll give you a slight boost in efficiency.

because Vfb is 1.25v, that's the most you'll drop across a resistor configured for current sensing, and then you can eliminate the current limiting resistor.
 
Is using a ground plane a good Idea? I just found out how to use the polygon tool, and I like it! The hatched part is Ground. The current limiting resistor drops 1.5V. I dont think this IC can be used in constant current mode. But either way you look at it, wont the same power be used?

BTW, heres what the LED board looks like
**broken link removed**

I used this tutorial on how to make a ground plane:
https://www.muzique.com/schem/eagle.htm
 
Last edited:
I love using the polygon tool, but I found the hatching doesn't work will with home made pcb fabrication, so I go with a solid pour, your mileage may vary.


how are you running 20 white leds with only 5 volts ... sounds like a massively parallel array (bad idea). each string (of one led?) in your array will need a current limiting resistor to force current sharing, you can't just use one resistor.

One suggestion is to use a boost converter with an external switch, and aim for a higher output voltage. For example, if you take ten of your leds in series, you need Vf=33vdc and If=50ma to light them up, then take another ten and run that in parallel, totals being Vf=33vdc If=100ma. The MAX1522 supports Vout of up to 100vdc, and Iout of up to 1a. Now you have a mostly series array, and the benefit will be increased efficiency (two resistors instead of twenty) and more balanced light output, thanks to a constant current applied across the entire string.
 
make sure you turn on thermals in eagle for your ground plane. otherwise you may have to overheat the parts while soldering them. The thermals prevent the plane from sucking all the heat away from the pad(s).
 
justDIY said:
I love using the polygon tool, but I found the hatching doesn't work will with home made pcb fabrication, so I go with a solid pour, your mileage may vary.


how are you running 20 white leds with only 5 volts ... sounds like a massively parallel array (bad idea). each string (of one led?) in your array will need a current limiting resistor to force current sharing, you can't just use one resistor.

One suggestion is to use a boost converter with an external switch, and aim for a higher output voltage. For example, if you take ten of your leds in series, you need Vf=33vdc and If=50ma to light them up, then take another ten and run that in parallel, totals being Vf=33vdc If=100ma. The MAX1522 supports Vout of up to 100vdc, and Iout of up to 1a. Now you have a mostly series array, and the benefit will be increased efficiency (two resistors instead of twenty) and more balanced light output, thanks to a constant current applied across the entire string.

I was going to use a ZXSC300, but it didnt have the current capabilities I needed (max of about 350mA). This is what my next design will use, But I'll be powering a 1W LED using 1 AA Battery.

Yes, it is All parallel. Ive done preliminary testing, and by adjusting the resistor, it does limit the current, although, some LEDs seem to vary in color and brightness. I doubt this has anything to do with configuration, since they came from a China wholesaler. However, overall it is bright.
 
The long thin trace to C1 isn't good. You need to position it so it can supply its current to L1.

I go with ground pour whenever I can. You still have to know where your current is flowing.
 
C1 can be moved to the right slightly and the trace can be routed in the over direction straight to the power connector.

Also it might be easier to put the connectors on the opposite side to the tracks in order to make soldering them easier, seems as they are through hole. Remember that if you mount the components on the opposite side, the IC must be reversed so that pin 1 is in the top-right.
 
C1 is a input capacitor to slow down input transients.

Better? I dont think I uploaded the version with a ground plane.
 

Attachments

  • snapshot2.jpg
    snapshot2.jpg
    124 KB · Views: 193
Status
Not open for further replies.

New Articles From Microcontroller Tips

Back
Top