LM2940C: Replacement for UA78M05 / Heat / Output Cap

Hi Helmut,

What type of capacitor are you looking at for the 22uF? LM2940 has an ESR requirement that exceeds that of ceramic capacitors; the most straightforward approach is to use a ceramic cap with a series resistor to add the ESR artificially. If you are interested in this approach, a 100mΩ series resistor with a 22uF X7R or X5R (depending on your application ambient temperature requirement) ceramic capacitor would be sufficient, but finding a 22uF 0603 X7R capacitor may prove to be difficult if you don't want to go a case size up. 

The thermal metrics reported in the Thermal Information are intended to compare the thermal performance between devices that are evaluated at a standardized PCB layout condition. If you were using the old design for the UA78M, then the LM2940 will perform slightly worse than the UA78M. If you were using the new UA78M chip in the design, then you can expect that the thermal performance will be significantly better with the LM2940C. With that said, with (24V - 5V) * 100mA = 1.9W (as a worst-case - I know you said the load current is less than 100mA), I doubt that it was the new chip because the junction temperature rise would limit this device to be used only at cold temperatures. So, the LM2940 will not solve this issue. Most LDOs cannot dissipate nearly 2W at elevated ambient temperatures, except for some devices that come in very large packages with large thermal pads or tabs. It's likely that you will have to spread out the heat somehow - you can add another series LDO to drop some of the voltage before the final regulation, add a series resistor to drop some voltage that way (if there's a lot of load transients then this isn't ideal), or use a buck regulator to more efficiently get the voltage down. If you don't do any of these then the LDO will run very hot all the time and its expected lifetime will likely be degraded. 

Regards,

Nick

Hi Nick,

thank you for your response and I apologize for the delayed reply - I was on vacation.
If I understand you correctly, there are two revisions of the UA78M, but I don't know how to identify it.
The exact part number should be UA78M05IDCYR, does this help?
So, regarding the temperature, if this is the new part number, then the LM2940 should be an improvement, otherwise it will be slightly worse.

If we have to do a redesign of the board, do you think using a buck converter will improve the heat-up?
I understand that an old-school regulator will have to dissipate power given by the voltage drop times the current draw, as you had calculated, but I am unsure if the same formula applies for a switching converter, and I don't know how to calculate the expected power dissipation based on the efficiency curve. Maybe you can explain?

Anyways, if a buck converter will improve heat-up, can you recommend a part that can operate from 12V to 24V and outputs 5V (fixed or adjustable, both OK), with an output capability of at least 500mA (see UA78M)? I found some parts with integrated inductors (like a TPSM33615), which sounds nice, but I am not sure if this will worsen the heat-up.

Thanks for your support,
Helmut

Hi Helmut,

Perhaps I glossed over it in my initial response, but since you said you're looking to use the LM2940C as a drop-in replacement, it's an old-ish design, right? If it's a year old or more then it's certainly the old device, and the thermal performance would be slightly worse with the LM2940C (53C/W for the old UA78M vs. 59.3C/W for LM2940C). 

A buck converter would definitely be more efficient, if your system can tolerate the additional switching noise. I'm no expert with switchers, but I can assign this post to the buck team for help with efficiency calculations if that's the route you want to go. An alternative is to add another LDO in series with the one already present or a power resistor to spread out the power dissipation. This option is obviously less sophisticated but could suffice if the total power loss isn't a concern for your application. 

Regards,

Nick

Hi Nick,

In fact, the board design is relatively new, but the power supply circuit is re-using an older design. The UA78M05 are relatively new (board assembly was less than half a year ago), but I do not know a date code. Would it have a date code on the chip, or is there really no way to identify old versus new chips?

Yes, please forward my request to the Buck Converter team. I need to quickly resolve the temperature issue when our board is supplied with 24V, getting a lot of pressure.

Thanks,
Helmut

Hi Helmut,

I see. The top marking doesn't indicate the old/new device, but the shipping label would have the information needed to figure it out - if the CSO on the label is RFB, it's the new device, and if it's anything else it's the old device.

I'm reassigning the post to the BMS team. Good luck!

Regards,

Nick