If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

# TPS62122: Input Capacitor Selection for Long Distance Power Source

Part Number: TPS62122
Other Parts Discussed in Thread: TLV713P, , TPS62125

Hello Forum,

We are designing a 12V to 3.3V (Buck + LDO) Power Supply for Bluetooth MESH network. We are using TPS62122 to step down 12V to 3.6V and filtering the output ripple through TLV713P to get 3.3V.
In total, 30 Bluetooth Nodes will be connected to a 12V line over 100 meters. Average and Peak Current Consumptions of each Bluetooth Node are 5.5mA and 25mA (9.9mA RX, 25mA TX and 1.4uA Sleep).

We request help on the below,

1) What is the recommended capacitor on TPS62122 input (12V) side considering the long distance from power source? (The TPS62122 datasheet recommends using 4.7uF ceramic. Is another 4.7uF in parallel sufficient considering peak load current 25mA)

2) Is 12V for main line a right choice? Need opinion please.

(Let me explain why we chose 12V over 24V or 48V.
Firstly, calculating for the worst-case scenario of all 30 Bluetooth Nodes consuming peak 25mA and TPS62122 working at 80% efficient, the maximum current drawn from 12V source is ((3.6*0.025*30)/(0.8*12)) = 0.281A
Eventhough current drops at each Bluetooth node and only 0.025A reaches the last node at 100m, to simplify calculations, consider full 0.281A travels 100m over an AWG20 wire, this creates 1.87V drop. So, the worst-case Voltage at the 100m Node is 10.13V which TPS62122 can still convert to 3.6V.

Secondly, cost. 12V Buck Converter costs less and takes-up much less space than 48V)

Thanks,
Ravi

• Hi Ravi,

Are 30 nodes connected to 1x TPS62122? Would this then multiply the 25 mA by 30 or has this already been taken into account?

Yes, lower Vin DC/DCs are smaller and cheaper. TPS62122 is a good choice, but you might have a look at the TPS62125. It provides more current and has a precise EN pin which might come in useful in your system for startup, etc.

Yes, you definitely need additional bulk input capacitance to overcome your high source impedance. This capacitance will likely be electrolytic, etc. to give some ESR and enough bulk C. My blog discusses this a little: e2e.ti.com/.../what-is-that-giant-tantalum-cap-on-the-input-of-the-evm

You can use the spice models on the product folder to model your input wire and see the behavior in your system.
• Hi Chris Glaser,

Thank you very much for the detailed suggestions.

Each node will get its own TPS62122. Kindly find the sample connections figure attached.

and since the max node current is only 25mA, we settled for TPS62122.

1. Sure, we will consider TPS62125 for our design.

2. Since the max load current for each DC/DC is only 25mA, we are thinking of using 47uF electrolytic capacitor for each DC/DC in addition to the 4.7uF ceramic. Do you think this should be sufficient? Or do you suggest a larger input capacitor?

(The 12V input power line is 100 meters long AWG20 wire and all 30 nodes are connected over this 100 meters only. Maximum Calculated Load current on 12V line is 0.281A)

3. Our 12V mains power supply is a 12V 2A SMPS converting 220V AC to 12V DC. Considering 30 bulk electrolytic capacitors on line, do you suggest we include a Soft Start Circuit on the 12V line immediatly after the 12V SMPS?

Thank You,

Ravi

• Thanks for the drawing. I think I understand.

The input capacitance requirement depends completely on the parasitics of the cable--its inductance and resistance. This can be simulated or measured. Each DC/DC will need a bulk Cin to decouple it from the input bus.

Yes, I recommend that the 12Vin be soft started to avoid inrush, overshoots, etc.
• Thank you Chris Glaser again.

In you first reply last line, you referred to 'Spice Models on the product folder to model your input wire'. Can you kindly share the link to this page.

We searched the below link but couldn't find an option to input line/source impedance and cable parasitic values.
webench.ti.com/.../switching-regulator

Thank You,
Ravi
• Hi Ravi,

Clicking on the TPS62122 part number in this E2E thread will take you to the product folder: www.ti.com/.../TPS62122

There, you click on tools and software to find the pspice and TINA models: www.ti.com/.../toolssoftware
• Hi Chris Glaser,

Thanks a lot for suggesting the TINA models link above, its greatly useful.

We have been testing various models and the below looks promising to us, so would like to seek your kind support again before we proceed any further.

For 12V transmission, we will be using 100 meters of AWG20 2 Core Power Cable, and the Inductance and Resistance of cable are 247uH and 3.3ohms respectively. (we used online calculators to determine these values)

We referred to TI's Application Report SLTA055 (www.ti.com/.../slta055.pdf) for the input bulk capacitor calculation. In our case with Vin=12V, Vout=3.3V, efficiency=0.8 and Max Output Transient Current 50mA, the Input Transient Current is 17mA.
(Equation 8 in page 5)

Now, with the Line Inductance 247uH, Allowable Voltage Ripple 100mV and Input Transient Current 17mA, the minimum required Bulk Capacitance is 8.82uF
(Equation 9 in page 6)

We chose 22uF for input Bulk Capacitance and below is the Transient Response graph.
We do not see voltage rings during startup and the Vin Overshoot observed is 13.49V

1) Are the Cable Inductance (247uH) and Resistance (3.3ohms) positioned correctly in our schematic?

2)Do we need to add Cable capacitance also to model?

3) From the transient response graphs below, do you suggest we go with 22uF input bulk cap? Or do you suggest any other value?

4) For Input Bulk Capacitor, we chose an MLCC over Electrolytic Cap to save space and do not see much inout voltage overshoot during startup. Can we stick to MLCC or do you still recommend Electrolytic?

Thank You

Ravi

• Hi Ravi,

I'm glad the model proved useful.

At the end of the day, these are your design decisions to make. A big unknown from my side is the accuracy of your model.

The cable sounds fairly damped with 3 Ohm, so an electrolytic may not be required. I would add footprints for both a ceramic and electrolytic.

I would also recommend that you perform a load transient to see the resulting Vin ringing.

Finally, I would also check the closest node (with its different parastics) to see if the results are any different.
• Hi Chris Glaser,

Thank you for sharing your valuable feedback.
Just wondering whether we can use a Thick Film Resistor in series with Ceramic Capacitor instead of an Electrolytic. Two reasons to avoid electrolytic are
1) the height of the circuit in our case needs to be smaller than 5mm
2) we tested nodes at various distances - 1m, 25m, 50m, 75m and 100m, and a single Resistor value is not suited to damp the oscillations at all nodes. For ex, 3ohms suits nodes between 50meters and 100meters; 2ohms for nodes between 25meters and 50meters; 0.5ohms for nodes  before 25meters.
1) Can we use a Thick Film Resistor in series with Ceramic Capacitor instead of an Electrolytic?
2) If Yes, how to calculate the safe Power Rating of this series resistor considering large inrush capacitor charging current?
For example, consider a 3 ohms Resistor in series with 22uF input Ceramic. Input Voltage is 12V and Maximum Input Transient Current is 0.02A. From the simulation below, during start-up, the maximum Inrush Capacitor Charging Current is 1.62A. We understand matching series resistor power rating to this current is not possible (1.62 X 1.62 X 3 = 7.87W), but not sure what is the safe Power Rating.
Is a 125mW Thick Film Resistor sufficient in this case?
3) Can TPS62122 / TPS62125 withstand single momentary input voltage overshoot up to 20V during start-up?
(The absolute maximum input voltages of both as per datasheet are 17V and 20V respectively)
Thank You

Ravi

• Hi Ravi,

Sure, you can use a discrete resistor in series with a ceramic. Or you could use a tantalum cap, which can have higher ESR and be lower height than an electrolytic.

More bulk input capacitance should reduce the overshoot on Vin. We recommend keeping the maximum input voltage within the recommended input voltage rating of the device. If it must exceed this, it must be kept below the abs max level. It would be better if the IC were disabled during this overshoot and enabled after the input voltage reduced to a value within the recommended range. Perhaps you could also lower your supply voltage a little.

It doesn't look like you are measuring the current through the discrete resistor. You need to add a current meter in series with the resistor to see the actual current in the resistor. Assuming this current only occurs during startup, it may be ok since it is so short a duration. You would need to contact the resistor company to understand the transient power limits of their resistor. Slowing the Vin rising time should also reduce the peak current in the resistor. Finally, you can use parallel resistors to distribute the power loss.