Welcome to Laser Pointer Forums - discuss green laser pointers, blue laser pointers, and all types of lasers

LPF Donation via Stripe | LPF Donation - Other Methods

Links below open in new window

ArcticMyst Security by Avery

Planning DIY TPS63000 based driver

jib77

0
Joined
Jun 19, 2010
Messages
718
Points
0
So I've been playing with Eagle PCB and reading TI ap notes. So far I have come up with the following design.

Any suggestions are welcome (Im a software guy and know just enough hardware to get myself in trouble:D).

Features:
+Battery Pad on bottom layer
Through holes for spring or wires
The current setting resistor can be a set value (R1) or a pot(R2)
0.60"x0.60" square (corners can be trimmed to fit the pills for CR123/18650 hosts).

Schem:
3524-drive-schem.png


Top Layer:
3523-drive-top.png


Bottom Layer:
3522-drive-bottom.png
 





Joined
Jun 30, 2008
Messages
286
Points
0
1st:
PCB is WRONG. You're dealing with high currents switching at high frequency. YOU MUST ground to a point or solid ground "plane". Using small traces between leads of components and then to the ground point is a fail, as the traces between component's pads act as parasitic inductances... and we don't want that.

Also, the thermal pad MUST be soldered to solid copper. There must be some copper area surrounding that pad so the IC can get rid of the heat generated inside.

Also, you must make power traces as wide and short as possible, less than 5mm long or so.

And that SPECIALLY applies to the feedback trace. There's a high precission comparator inside the chip taking samples of the voltage in that pin at very high speed. That trace MUST be wide and short, or else, any induced noise would cause inestabilities and ripple in the integrated circuit.
That's not a power trace, but a high frequency analog trace, which also need to be wide and short.

2º The coil is wrong. Dunno what coil you're using but it looks way tooo small. Small coils have high ESR and low saturation current. During startup, there's a current spike that the coil MUST handle without saturating, or else, the driver might not start. Just get a bigger, more massive coil with some serious ferrite core. Something with less than 0.1ohm of series resistance and with saturation currents around 2-2.5A should be fine. You should find those in 5x5 or even 4x4mm packages.

3º The design itself might be wrong. I've tried to make designs like yours with high side current sensing using a differential currtent sensing amplifier and all of them went seriously wrong. They just weren't fast enough to be able to "update" themselfs at the speeds required by the regulator IC, so it didn't stabilize and the current just oscillated between the one it was supposed to be set to and the max current the driver could deliever. A plain and complete fail. Adding a capacitor solved the problem but made the driver go "full power" until the capacitor charged, which happened in less than a millisecond, but enough to kill a diode with a very high current spike.
And yeah, I tried some very fast amplifiers.

4º The sensing shunt resistor will be a very low value one, in the range of 0.1 ohms or so. Try finding potentiometers of those values. You can't adjust that resistor. A better way is to add a potentiometer as a voltage divider in the output of the amplifier.


If you want any design help, or any example of how a PCB for one of these must be designed, just ask for it. :p
 
Last edited:

jib77

0
Joined
Jun 19, 2010
Messages
718
Points
0
This is the inductor I was planning on using: Coilcraft ME3220

The smallest traces are 10mil with most of the power being 16mil ... the longest trace is 3.8mm.

I have the TI formula for Rs at home ... ill look it up as soon as I get off work.

Man that sucks about the current sensing amp ... why the heck does TI have an app note about it (along with a recommendation for the INA193). Sheeesh.


Update:

Formula for Set Resistor -> Rs = Vref/(IxG)
Where I is the desired current and G is the Amp gain
 
Last edited:

jib77

0
Joined
Jun 19, 2010
Messages
718
Points
0
what will be the cost and range of each?

It cost me $34 for 3 prototype boards. Im sure it will be less per board in higher quantities. If I can get the design to work I might consider doing a GB for the PCBs.
 
Joined
Jun 30, 2008
Messages
286
Points
0
This is the inductor I was planning on using: Coilcraft ME3220

The smallest traces are 10mil with most of the power being 16mil ... the longest trace is 3.8mm.

I have the TI formula for Rs at home ... ill look it up as soon as I get off work.

Man that sucks about the current sensing amp ... why the heck does TI have an app note about it (along with a recommendation for the INA193). Sheeesh.


Update:

Formula for Set Resistor -> Rs = Vref/(IxG)
Where I is the desired current and G is the Amp gain

Okay, the traces are too thin. You should use 15-16 mil for signals and around 25-30 mil for power and feedback sense.

And the grounding is still wrong and that's a big issue you MUST correct. Grounding like that would work for lower frequency devices but this one switches at 1.5Mhz. Definitely you MUST provide a solid, wide ground. Else, it might work but not at the desired performance (big ripples and that).

About the coil, well, yeah, that's a power inductor and it's recommended by coilcraft for the TPS63000. So yep, it's just right. Better if shielded but... no big issue there

And about the design, I said it might be wrong, because it didn't work to me, but it doesn't mean it won't work for you. I never tried a TPS63000 nor a INA 193. I'm willing to see if this proyect works for you, and please, could you give me a link to that application note? I wanna check it!
 
Joined
Dec 3, 2007
Messages
54
Points
0
I have the TPS63000 running in a similar setup with a current sense amplifier. I used a Zetex 1009 current monitor.

The TPS63000 and 61200 are the only ones I could get to run in this manner, tried it with some other step-up converters and had the same problems as described by erdabyz.
 
Joined
Jun 30, 2008
Messages
286
Points
0
Good to know i'm not the only one who miserably failed with that configuration... and also good to know that TI's regulators can be driven that way.

It was hilarious when I assembled my first driver with current sense amplifier. I knew it would fail because the simulator said so, but I wanted to see it fail in the real life.
The o-scope showed the current oscillating at around 18Khz and the coil was sounding at that frequency. Yeah, I could hear a high pitched tone. If i increased the current the frequency also increased until I stopped hearing it.

It was a driver fail, but we had a great time checking our auditive capabilities. I can hear up to 18.5Khz +- xDDDD.
 

jib77

0
Joined
Jun 19, 2010
Messages
718
Points
0
Well, I got the prototypes ... not too bad. I wont be stuffing them due to improper design, but hey they are at least useful as a battery contact PCB.

 

Rafa

0
Joined
Feb 4, 2010
Messages
437
Points
0
Erdabyz said thay they might work or not, at least try them, you paid a lot for them.
 

dnar

0
Joined
Jun 7, 2010
Messages
608
Points
0
I'll add that adding R1 and R2 in parallel is bad. You can adjust down to 0 ohms....
 

jib77

0
Joined
Jun 19, 2010
Messages
718
Points
0
I'll add that adding R1 and R2 in parallel is bad. You can adjust down to 0 ohms....

It was never intended to be used in parallel, just have a choice weather to set it permanently with a fixed resistor or use a pot ... Ill have to rethink that since we are talking about very low values (~0.3Ω for 500mA).
 

jib77

0
Joined
Jun 19, 2010
Messages
718
Points
0
Ok I got rid of the pot and just have two places for 1206 sized resistors so you can combine to get even lower values. Fixed the thermal pad issue, and all power traces are at least 32mil.

3648-jibdriveschem.png


3651-jibdrive-v2.png
 

Rafa

0
Joined
Feb 4, 2010
Messages
437
Points
0
That one looks pretty good, surely it will work.

Btw you should extend the thermal pad to all the board.
 




Top