Circuit Advice Requested - power supply, kind of like GSM/GPRS

Timothy D. Swieter

Hi Guys!

I'm working on the power supply section for a new device I'm cooking up. I'm a little unsure of the architecture of the power system, so I thought I'd get some feedback from the forum.

The device's power behavior is best described as low current mostly, with pulses of higher current for known periods of time. The best analogy for this that I am finding is GSM/GPRS modems, which also have similar pulses of high current for radio transmissions. Another analogy would be circuits that need to "ride through" in that there is some local power storage to help with the times of high power demand.

Take a look at the attached picture. It is a block diagram of a potential power system as well as graph describing the current.

The device will have an I/O and Power header, as well as a USB connection. The user's application can use either or both connections for communication and power. So, that is, there could be two power inlets (A) and (B). I'd like for the header connection to accept a range of voltage while obviously the USB connection is ~5V. Block A and B then are voltage regulators (maybe also current limiters?). Perhaps Block B doesn't need to exist and 5V is passed straight through. Block C is a power OR'ing circuit to select one or the other supply. The default supply would be the header, if both supplies are attached. I'm thinking something like Linear Technology LTC4412 or LTC4415 for the OR'ing.

After the OR'ing, there is a section D, which could be several different things. One requirement could be the current limiting occurs here, instead of at Block A and Block B. The default would be current limit of 100mA (for USB purposes), but I'd probably make solder jumpers or something similar so the current limit can be increased by the user if they so wish. This section would also have large capacitors, maybe super caps, for the handling of the higher current pulse loads. I've looked at lots of different links on GSM/GPRS modem designs and also super capacitor charging circuits. Section D could also be combined with Section E and F, just depending on the solution. For Section D, I've considered Liner Technolgoy LTC4425 of which I'd have it charge super caps to 5V, and have LDOs for the digital and analog section. I've also looked at LTC3619B, which nicely integrates two voltage outputs (eliminating the LDOs of E/F), but it looks like the minimum current limit is 200mA, and I was hoping to stay at 100mA, just to be compliant with USB minimum.

Various questions in my mind to stoke discussion:
1. Block B, is a voltage regulator really needed here?
2. Should I stick with 100mA current limit requirement (to try and play nicely with USB) or should I forget about it and pick some other limit? I wanted to have a limit, just so that the user employing the device knows how to plan power feeds to device.
3. Is this all more complicated than it should be?
4. Block A, I guess depending on the current limit, may or may not need to be a switching power supply. I'd like to keep it easy though. Thoughts?

Thank you for having a look and sharing any thoughts, questions or experience.

Timothy D. Swieter

I am thinking that for the current limit, I need to do some simulation and calculating of the super capacitor sizing in order to see if the circuit is even sustainable at 100mA with the known discharge/recharge times. It is possible it may not be, or at least, the super caps need to be rather large, which can be fine.

Lawson

The usb spec says the voltage from usb should be between 4.75 and 5.5 volts. If that's an acceptable range, I'd just put a noise filter in block B. I'd integrate the cap charging current limit into blocks A, B, and/or C if you can. Voltage drop in these circuits will greatly effect the size of the super-capacitor array. Assuming you can charge the super-caps to 5v with a 1v usable voltage swing, it looks like a 7.5F or greater capacitor array is needed. It may turn out that a boost supply in block B and a ~7.5v maximum capacitor voltage works better. (Note: for 2.7v rated super-caps, a blue led in parallel forms a cheap and effective balance network.)

Also, your maximum average current is about 150mA. Something has to give to fit that within the "safe" usb current draw. (though many PC ports are paired and protected with a 1A PTC fuse, so ignore power profiles and will source 2A for several minutes)

Marty

Franklin

The Arduino uses a power supply sharing setup. Perhaps you could check out how they accomplish that.http://arduino.cc/en/uploads/Main/arduino-uno-schematic.pdf

Timothy D. Swieter

Thank you Lawson and Franklin for the responses.

A long time ago, I recall looking at the Arduino schematics. In fact, I was even looking at their GSM shield schematic a week or two ago. Thank you Franklin for inspiring me to do that again. This time though, I looked at the schematics of all the other boards to see what their power schemes were and how they handled the powering OR'ing. I still like the use of the LTC4412/4415 as it makes the circuit neat and obvious, however I will have to consider the added costs. I like learning by reviewing designs that are implemented and working. For instance, in the Arduino Yun schematic I see the Micrel current limiting device, this is cool and I hadn't come across that before.

Marty, thank you for your thoughts too. I like the idea of say buck/boosting for a higher base voltage on the capacitors and will consider this further. You stated that the average current is 150mA, I'm curious how you determined that. I don't have to absolutely have 100mA current limit, I could work with another limit. Perhaps my desire to play nice with USB is outweighed by the fact that so many devices don't play nice and the "defacto" standard is that USB hosts are more tolerant of "demanding" devices if the current draw is 200mA or 300mA without asking for it. For instance, on the Arduino Yun device, the current limiter looks to be 500mA and I'm not sure I see where it is enumerating, but perhaps it is as I don't have a board.

I'd love to have some more discussion of this power scheme and thoughts for my research and tinkering. Perhaps my next step is to draft a schematic, post it up, and get some comments on that.

Lawson

As far as I know, Apple devices are the only ones that strictly enforce the USB current limits. So if it works fine with a Mac, it should work with everything else. Also, I think there are some ways to passively signal a computer that a device needs >100mA.

The 150mA estimate was a simple weighted average of the worst case you layed out. i.e. (15sec * 0.5amp + 60sec*0amps)/75sec + 50mA idle current. For best accuracy, add up how many Coulombs are in each section of the power cycle then divide by the cycle period to get the average current. (i.e. Coulombs/second)

Marty

Timothy D. Swieter

I've cooked up a basic schematic and even more basic circuit description of the power section. I need to run some more calcs on the super capacitors and current limits. This schematic gives me some foundation to work from and tweak if needed any values or parts.