LM5001-Q1: DCDC Converter selection

Hello Monique,

Thanks for using the e2e forum.
3.6V to 36V is a boost factor of x10, so it might exceed the duty cycle capabilities of a simple boost controller.
An alternative option would be use of a flyback topology, which allows buck and boost and can support higher boost ratios by adjusting the winding ratio.
However, this topology is commonly not used for more than 100W power. As you application needs 180W peak power, you may see high losses due to the voltage drop on the output diode.
Here another option would be using the boost controller stages.

Is 5A load only required for a short time, or is this the common operation that needs to supported over longer timespans?

Thanks and best regards,
Niklas

Hi Niklas,

5A is the peak current, and would not be sustained. It would likely be about 3A sustained for around 8 minutes, and I plan to characterize this better. I would be interested in how to implement the flyback and boost controller stages; what would the differences be between those and the benefits of each one?

Thanks,

Monique

Hi Monique,

Thanks for the feedback.
Allow me to check if I find a suitable reference design. I will get back to you tomorrow.

Best regards,
Niklas

Hi Monique,

Sorry for the long delay.
Unfortunately, I was not able to find a reference design including both boost stages or a flyback of this size.
In fact, there where some additional problem I encountered with these parameters.
With 3.6V input voltage, most of our controller will not turn on, as the require a higher voltage for startup.
LM5123 might be the closes device, which can start up with 3.8V minimum supply.

There are workarounds for starting up with low supply voltage.
This reference design uses an additional start-up circuit and can turn on with just 1V supply voltage.
https://www.ti.com/tool/PMP31011
However, the application might get a bit expensive this way, as you would require two boost stages and a start-up circuit.

You mentioned you will use 3.6V batteries for the supply.
One consideration would be to use two batteries in series, to create 7.2V.
With this supply, you can achieve 36V with just one power stage and no start-up circuit would be required.
This would safe cost and you can use one LM5122 or LM5123 device.
Would this be a possibility?

Best regards,
Niklas

Hi Niklas, 

I was actually looking at the LM51770 using 3.6, and it seems to do okay? However, the excel sheet and the online power designer give very different values for the inductor. Would that be a possibility?
Using 7.2V, I am going to test with the LM5123 as well.

Thanks,

Monique

Hi Monique,

LM5177x would be suitable as well, I agree.
It is possible that you receive different recommendations on inductance by different calculators.
This comes from the fact that inductor current ripple and component size is a trade-off and the load at which the regulator switches between DCM (discontinues conduction mode during light loads) and CCM (continuous conduction mode) depends on design preference.

Essentially both calculation tool recommendations should lead to solid designs, but if the values vary a lot between calculators, I would recommend to go with the quickstart calculator, as this tool takes more inputs from the user.

Best regards,
Niklas

Thank you!

For the LM5123, I was using the Eval Board and substitued the components for the ones in the quickstart spreadsheet I used. Below is the BOM for the replacements. Could you take a look? When I turned it on, something smoked, so something is wrong, but I am not sure what...

Hi Monique,

Thanks for the update.
The component ratings all look okay to me.

Did you make measurements of the switch node during operation?
If the switching is unstable, you may have chosen a compensation that is not stable.
In this case I would recommend to choose a slower compensation and check if the system stabilizes.
You can also attach the quickstart spreadsheet you have filled in and I will double check the entries.

Best regards,
Niklas

Hi! I realized I accidentally had a jumper connecting Vcc and Vin, and I think that was the cause. Why would someone want to connect those pins? I am also assuming my 7.2V input would cause the Vcc pin to overload and break the chip and potentially the Fets

Thanks,

Monique

Hi Monique,

I am glad to hear you found the problem yourself.
Is the design working now, or do you need additional help?

Best regards,
Niklas

Hi,

it did not work, and I was going to use spice to see what could have been incorrect. I imported the PSpice model into LTSpice, but got an error. Do you have an unencrypted PSPICE or encrypted LTSpice model I could use?

Thanks,

Monique

Hi Monique,

The engineer is out of office today. Please expect a delayed reply.

Best Regards,
Feng

Hi Monique,

The PSpice model of LM5123 is encrypted, so it cannot be transferred to LTSpice.
We have an unencrypted model available, which can be shared under NDA restrictions.
This file can then be transferred to LTSpice.

Do you know if your company has an active NDA contract with Texas Instruments?
If yes, could you share this file with me via direct messaging?
I can then prepare the unencrypted files and send them to you.

Sorry for the additional inconvenience.

Best regards,
Niklas