LM5175EVM-HP: Adapt the output voltage to 18V with FB voltage dividor

Hello, 

Thanks for reaching out. In order to change the output voltage you need to change the feedback divider, like you said, and check that the compensation components still fit the new design specs. To help with the choice of these component, you might refer to the Excel Quickstart calculator: https://www.ti.com/tool/LM5175QUICKSTART-CALC 
Let us know if further help is needed. 

Kind regards,
EM 

Hello EM,

thanks for your quick reply and I wanted to confirm the new values with you

- LVM5175EVMHD : I have been doing the changes to the feedback divider with LVM5175EVMHD to reach 18V output with the current value of divider and TRIM input and by having R15 to become 13.3kohm I can reach 18V. Now we will try the EVM at 18V/4A to see if it sustains the load at 4A with no issue. Do you believe I could do more current at 4A or there is a limitation somewhere ?

- LVM5175EVMHP : Looking at LM5175QUICKSTART-CALC for 18V/12A output instead of 12V/6A I noticed that the differences between 12V/6A and 18V/12A is only in the compensations components section Rc, Cc1, Cc2.

12V/6A : Rc=2.1kohms, Cc1=18nF, Cc2=68pF whereas 18V/12A : Rc=3.74k, Cc1=10nF, Cc2=27pF

Other than that, one needs to change R11, R14, R22 to configure the 12V output to adjust to 18V.

Could you confirm my above informations ?

Zabieru.

Hello EM,

Some updates on LVM5175EVMHP : I have changed R22 to 13k to get Vout=18V which works correctly (I have not changed the compensation component but I believe there are  R22, C23, C28 when looking at the schematic as they connected to COMP pin).

Now I am using an electronic load to measure the possible drainable current and it reads 18V at 0A but then the voltage drops as the current increases with 16.7V at 1A then at 1.35A the voltage at load drops suddenly (around 21W) where I would like to reach at least 120W nominal like the schematic seems to indicate.

Could you help me to reach the 18V/6A to respect our specs ?

Best regards,

Zabieru.

Hello Karcel,

Please take the quickstart calculator to check the changes needed for reaching your output voltage and power levels. Yes, the feedback divider is the first thing to change, but you might need to change as well the sense resistors and the input and output capacitors (due to the voltage rating) or the inductor due to the current rating. If you fill in the calculator properly, it should work.

BTW, it is possible that the compensation causes the issue that the output power is limited, as the compensation stabilizes the output. Please check with a scope after you changed all necessary components.

Best regards,
Brigitte

Hi Brigitte,

I have been using LM5175QUICKSTART-CALC for the 18V/12A configuration and what do modify on the LVM5175EVMHP schematics and below are the results, if you could confirm those values

- Rsense=6.1mohms 1% 6W
- Cin = 42uF 0V, Cout = 50uF 25V

- Lf = 2.1uF (6mohms) 17A

- Rc=3.74k, C_c1=10nF 100V, C_c2=27pF 50V

We will use those values to do further tests.

My best regards, Zabieru.

Hi Karcel,

a few comments:

- you need also to update the Feedback resistor divider for the new voltage

- the output cap looks very low, but this depends on the ripple you can allow - please double check

- also double check the input cap again

Best regards,

 Stefan

Hi Stefan, Hi Brigitte,

thanks for your feedbacks and basically my understanding of the modifications

- FB divider : this is already good and giving 18V output

- Rsense : from the calculator output we need to put 6mohms for the 18V output we target

- Cin and Cout we will add more capacitors but the LVM5175EVMHP is equipped with quite a lot of capacitors already. We will see the ripple on the output

- Inductance : we need to adapt (from the results of the calculator) from 4.7uH currently on the LVM5175EVMHP to 2.1uH and 17A rating (Wurth 7443551200 looks good for that specs)

- Compensation components (I believe important to maintain 18V when current increasing from message of Brigitte) : we will put C_c1=10n, C_c2=27p, R_c=3.74k

That's basically the components we imagine changing on the LVM5175EVMHP. Could you confirm that changes ?

Regards, Zabieru.

Hi Karcel,

I am slightly confused - the EVM has quite large caps on input and output - this is right, but you gave this:

Cin = 42uF 0V, Cout = 50uF 25V

Best regards,

 Stefan

Hi Stefan,

- From the CALCULATOR results (please see attached)
* Cin    > 42uF(ESR=2mohms) 25V (in our case input will be 13-22V) --> EVM-HP is equipped with 32.9uF/80V : so we add 10uF/25V in parallel we should be correct right ?
* Cout  > 128uF (ESR=2mohms) 25V (we want 18V ouput)                 --> EVM-HP is equipped with 132uF/22V (6*22uF in parallel) with 660uF/16V (2*330uF in parallel) for a total of 792uF : so greater than 128uF so we don't need to change the Cout right ?

- For the other components, do you confirm the desired modifications ? namely the Rsense to be 6mohms to ensure a greater output current right ?

Much regards, Zabieru.


LM5175_Buck-BoostQuickstartTool_18V_12A_Results.pdf

Hi Karcel,

Cin : 25V is to less margin with input range is 13-22V - would at least go for 35 V

Other settings look OK.

Best regards,

 Stefan

Hi Stefan,

thanks for the feedback  and I will target 35V for the add-up Cin=10uF

One remark on the LM5175EVM-HP differences between the datasheet and the actual PCB

- LM5175EVM-HP : 4.7uH/3.5mohms  (XAL1510-472MEB or 74439370047)

- actual PCB : 2.6uH/1.58mohms/31.5A (Wurth 744355626)

So I believe we can leave the equipped L1 on the PCB (instead of 7443551200 = 2.1uH/2.6mohms/23A that we want to install)

Do you confirm we can leave the equipped L1 on the PCB ?

Regards, Xavier.

Hi Stefan, Brigitte,

I have been changing the components accordingly and done some load tests (with an 150W electronic load) and a stable power supply (also limited at 150W). We can reach 18V/3A output so far (before getting limited by the 5A max output of the power supply).
I will extend the tests to 18V/6-10A tomorrow and feedback you.

One question about efficiency, I am currently observing 91% efficiency (displayed power at supply and load). The LM5175 indicates more 96+%.

Should I worry somewhere or is there room for improvement ?

Cheers, Xavier.

HI Karcel,

thanks for  the update.

When doing efficiency measurements, you need to have a high resolution multimeter for current and voltage on the input and output side (so 4 in total). Using the supply and load can give wrong results as this are often not so accurate and you also measure the drop on the supply lines.

Best regards,

 Stefan

Hi Stefan, Brigitte,

I ran more tests at 18V/6A which are looking fine and will push the tests to 12A and feedback, beside this I have several questions

1.If there a protection or error on the LM5175EVM-HP board, does it trigger something on a pin or signal somewhere that I can use for an external GPIO ?
2.If running the LM5175EVM-HP at higher current, we might have some heating issues ? Do you recommend to monitor temperature somewhere on the board and place heat sink ? Any application note for thermal designs on this board ?
3.Our goal is to have 18V/12A (nominal max current) but we may have some brief (few ms) peak current at higher current (20A), is there any limitations on the current design that will prevent from this brief peak currents ?
4.Is there a way to enable/disable the output on the LM5175EVM-HP through GPIO, the datasheet indicates the

Best regards, Zabieru.

Hi Karcel,

1. You can use the PGOOD - this does not cover all faults but if the output cannot be maintained it will be signaled there.

2. For the thermal: Best would be to load the board with the worst case load and condition and check the thermal behavior with a thermal camera.

3. The EN/UVLO pin can be used for that. If EN and UVLO should be used just add a Diode between EN/UVLO and the digital control signal.

Best regards,

 Stefan

Hi Stefan,

1. You can use the PGOOD - this does not cover all faults but if the output cannot be maintained it will be signaled there.
=> for instance I can connect PGOOD to a pull-up GPIO on a MCU which monitors the level. If PGOOD GPIO level is going up then a fault occured right ?

2. For the thermal: Best would be to load the board with the worst case load and condition and check the thermal behavior with a thermal camera.
=> I have run more tests with thermal camera at higher current
- 12A/18V for 10 minutes and the hot part is L1 self (currently Wurth 7443551200 = 2.1uH/2.6mohms/23A) and of course the MOS (but no so hot to worry)
- The circuit limitations was reached at 12.5A (still fine at 12.4A) and then the output voltage dropped but current keeps at 12.4A
- How could I possibly increase the max current at 18V ?

3. The EN/UVLO pin can be used for that. If EN and UVLO should be used just add a Diode between EN/UVLO and the digital control signal.
=> I guess you are replying to question 4 ("Is there a way to enable/disable the output on the LM5175EVM-HP through GPIO, the datasheet indicates ?") ?
The LM5175EVM-HP is equipped with the SW2 manual switch so I wondering how can I bypass this switch to externally control LM5175 EN signal through a GPIO ?

Best regards, Zabieru.

Hi Karcel,

1, no - the other way round, if PGOOD goes low the output is not within the specified range, which could indicate an error.

2, It might reached the current limit set by the R1 connected to VSNSP and VSNSN  (12.5A * 0.004 Ohm = 50mV)

    You would need to reduce R1.

4, yes, that should be 4 - so using 4 again: the SW2 also has a center position where the switch is open

Best regards,

 Stefan

Hi Stefan,

1.PGOOD : ok I understand and will software monitor the low level in the MCU
2.VOSNS+/- R1 : I will try a 0.002ohms/3W to get 25A*0.002ohms=50mV and feedback. Do you see any other limitations somewhere on the EVM to reach 24A?
3.I have confirmed that SW2 on central position and external GPIO bringing EN to GND disables the LM5175, the load will indeed see the discharge of C1, C13, C14 when LM5175 gets disabled.
Best, Xavier.

HI Karcel,

the EVM has not been designed and for sure also not been tested for that current.

So at least these points need to be checked.

- are the PCB traces able to handle that current

- are all components able to handle that current

- when loaded with this current are the components able to dissipate the losses

Most of these points cannot easily be answered and requires a detailed analysis or calculation.

Best regards,

 Stefan

Hi Stefan,

of course an EVM is not designed for high constraints such as 24A current. What are the max electrical ratings of this EVM ?
In terms of design what is the acceptable limits we could go until ?
Our plan is next to design a PCB according to the application specifications but at this time to use the EVM to validate the concept as a Proof-of-Concept design.

Could you get some feedbacks from the EVM design team about the limit they use for this design ?

Best regards, Xavier.

Hi Stefan,

after changing R1 to 2mohms/5W, we tested the EVM until 18V/15A output for 30s and it was running fine. The L1 self is the one component getting hotter rapidly but under 60 degrees from finger touching it.

I would not go too much above 15A not to break the EVM we have. Looking at SV601157A_PCB.PDF document about the layout, it seems that Signal3 and Bottom layer should have some limits in terms of trace width eventually. The other layers seems to be solid polygons. We cannot study more the layout as we don't carry the layout tool in-house but if you have TI inside info about the actual limit of the design it would great.

If we want to start designing the PCB layout for 18V/24A output specs, which document do you recommand for design guidelines ?

Much regards, Xavier.

Hi Karcel,

Sorry, also not more data available what the max current would be the PCB could handle. I just know the data it was designed for, and this are the ones given in the EVM Users Guide.

A good starting point for Layout is this first article (of 4) https://e2e.ti.com/blogs_/b/powerhouse/posts/four-switch-buck-boost-layout-tip-no-1-identifying-the-critical-parts-for-layout

The next one is the linked in the first.

Also here you find some info about PCB (Page77 ...)

https://www.ti.com/amplifier-circuit/analog-engineers-pocket-reference-guide.html?keyMatch=ANALOG%20ENGINEERS%20CALCULATOR

or the tool:

https://www.ti.com/tool/ANALOG-ENGINEER-CALC?keyMatch=ANALOG%20ENGINEERS%20CALCULATOR

Best regards,

 Stefan

Hi Stefan,

1.thanks for the great pointer for PCB layout design and we keep in that in mind for next steps
2.We have connected a GPIO from a MCU directly (No PU/PD resistors) to the EN pin on the EVM and when driving the signal low from the MCU it does drive strongly enough to trigger the EN to low level (whereas putting directly the EN pin to GND disabled the board). Do you have a recommanded circuit to drive EN from external circuit GPIO ?
3.The LM5175 is a single-phase buck-boost architecture, for higher current (~30-40A at 18V) would you recommand a multi-phase architecture product from TI ?

Best regards, Zabieru.

Hi Karcel,

2. Is there a condition where only one of the two parts (MCU / LM5175) is powered?
     Is the drive level matching the levels for the EN input?
     If that is OK and it fulfills your requirement it should be good.

3. Yes, for that you should use a multiphase architecture. LM5177 might be an option if buck boost is required. For a boost only design you may check LM5122.

Best regards,

 Stefan 

Hi Stefan,

2.Both the MCU and the LM5175 should be powered at the same time (maybe cases where MCU is powered and the LM5175 board is powered off then come back to power, is there a concern related to EN connected between the MCU and the LM5175 ?).
The MCU is classical Arduino-like MCU with 3.3V GPIO where we use one of the GPIO as output to drive the LM5175 EN to GND.
When doing that the GPIO seems not solid enough to drive the EN to GND.
Looking at section 8.3.3 of the datasheet (Enable/UVLO) it seems that it may need a votlage dividor. However the goal is very simple and just bring EN to low and then release to latch the LM5175.
Do you have an example circuit on how to connect and Arduino to a EN pin of LM5175 ?

3.For later if LM5175 is not enough for higher current.
Best regards, Xavier.

Hi Karcel,

when the MCU can be powered while the LM5175 is not powered, i recommend to add a series resistor between the MCU and the EN pins to limit the current.

(In this case current can go from EN to the protection diode within the LM5175).

It sounds strange that the MCU IO would not be able to drive the pin low. You and cross check with just putting an e.g.  10K resistor between EN and GND, the EN should then be on 0V. (I assume you have nothing else connected to EN).

I do not know Arduino MCUs very well but using an MSP430 was not an issue.

Best regards,

 Stefan

Hi Stefan,

I have made some tests with different PD (Pull-Down) resistor values (10k, 1k, 100, 1, 1) on the signal between LM5175-EN and the MCU GPIO (MCU=ESP32) and none gives the right value to drive LOW (GND) and HIGH (7V from the LM5175EVM-HP signal).
Maybe we need the line resistor to be driving correctly the EN signal from LM5175EVM-HP ?
Do you have an example of driving circuits between MSP32 and LM5175 (or other circuits ?).

Best regards, Xavier.

Hi Karcel,

can you please share the schematic - esp. the section showing the connection between the MCU and the LM5175 with all components connected to that.

Best regards,

 Stefan

Hi Stefan,
sorry for my late reply, basically the connexions are like below between ESP32 and LM5175EVM-HP
ESP32 (GPIO13) <---> LM5175EVM-HP (EN pin which is connected to R19 93k Pull-Down on the schematic of the EVM)
The goal is to be able to drive to GND the EN pin from GPIO13.
Best regards, Xavier.

Hi Karcel,

R19 is connected to MODE not UVLO/EN

Best regards,

 Stefan

Hi Stefan,
let me check the actual connexion and if actually connected to R19 it seems normal that this is not working. Will test with R18 and feedback later.
Best regards, Xavier.

HI Stefan,
I have checked the connexion and using R18 does not work and the EN signal stays at 2V level. Something I missed somewhere ?
I have not added any more PD resistor other than the 59k existing on the EVM.
Best, Zabieru.

Hello Karcel,

As we did not hear back from you, I expect the issue is solved and I will close this thread now. Please post below or start a new thread for further questions.

Best regards,
Brigitte