Hello,
We are trying to use the TPS61500 chip for a new project in development. We are wondering if anyone could help us a bit with the DC/DC design since we are not expert in this domain. The schematic of the circuit we'd like to use is given below, and the idea is to get help with this schematic first, and then help with how to place components on the PCB.
Here is the general design we are using so far (you can find attached the picture of the schematics):
- We are using the PWM output from a PIC from microchip to analog dimm the TPS61500.
- The general power supply will be a 12V supply.
- The PIC is powered using an LDO regulator (5V/150mA).
Here are some questions we are not sure about the asnswers:
1) We'd like to use the Voltage output from the LDO to supply the TPS61500 as well (we think it will be more stable as the 12V supply can vary a bit): Is this possible regarding the power required, knowing that the pic is consuming 95mA and therefore the LDO has 55mA available for the TPS61500?
2) We don't really know how to connect the different grounds (AGND and PGND). Can we use AGND for the PIC as well? If yes where should we "star connect" the two grounds AGND and PGND?
3) If anyone could give us additional information on the design with respect to our schematic, it will be very helpful and greatly apreciated !! :)
Thank you all for your time and help.
Best regards,
Guillaume
Guillaume-
Are you sure you want a boost for this setup? What is the Vf of your LED? Is it truly a single LED, or is this representative of a string?
To answer your questions:
1) This should be fine. Per page 5 of the 61500 datasheet, "Connect VIN to a supply voltage between 2.9V and 18V. It is acceptable for the voltage on the pin to be different from the boost power stage input..." Also, in terms of the LDO's capabilities, you should be okay. Page 4 of the datasheet states that Iq-max is 3.5mA into no load. You appear to have plenty of overhead from the LDO. I must say that 95mA to run a PWM pin from a microcontroller is quite a bit, though. If that is truly the number, you may want to consider a lower power microcontroller, like TI's MSP430. And if there are no other inputs to the microcontroller, how do you intend to modify the PWM duty cycle?
2) Will you have a solid ground plane in your PCB? If you do, and you are inexperienced in PCB design, I would highly recommend that you tie all of your GND pins directly to the plane. Don't try to cut the plane if you don't have experience with this. If this is a single layer board, or has no plane, then I would point you to page 14 of the 61500 datasheet. There is a layout example provided that shows where to place components and how to connect the grounds. The MCU may live on AGND as well.
3) I would star-connect the power from your LDO output cap (C6) to the MCU and the TPS61500. You may want to put a small value R or a ferrite bead to force this to occur in the layout and provide some filtering. I would also provide local bypass for both device's supply pins.
I hope this helps.
Regards, Mike
Hello Mike,
Thanks a lot for your reply.
Yes I simplified the schematic: we'll have a LED string with around 15V in total for this string.
1) 95mA is in fact the maximum rating for the Microcontroler. For a security reason I need to stay below 150mA when summing all the max currents from the different components connected to the LDO. I therefore believe I am fine in my case. Moreover, yes we we'll have other connections to the microcontroler that will allow changing the PWM duty cycle, but I did not show them to clarify the schematic. Thanks a lot for your answers on this point.
2) Yes I'll have a solid ground plane (two sided pcb with components on one side only). When saying "I would highly recommend that you tie all of your GND pins directly to the plane", do you mean both types AGND and PGND tied directly to the ground plane without separating them?
3) Ok thanks a lot! I will try to do that.
This was very useful, thanks a lot for your help.
Hi Guillaume. That is correct. Tie both AGND and PGND to plane as quickly as you can. The point of star-connected grounds and split planes is really about trying to minimize the interaction between the power loops of the regulator and the control loops. In theory, split planes are great, but in practice they can cause more harm than good. You can minimize the interaction simply by how you layout the regulator (i.e., where specific components are placed). You want to try to minimize the inductive loop of the high di/dt paths (the power loops) in the PCB, and you want to physically isolate the control path loop such that it is removed from those loops, or at least orthogonal to the power loops, if it must cross.
Thank you very much for these advises! This is very helpfull. Today we made a prototype respecting the rules you gave us last time, and it is working very well. We got a good 92% efficiency (ratio of power consumed by the LEDs to the input power). This is very encouraging!
Thanks a lot, we really appreciate your help.