Is the PMP8372 suitable to generate ±12V output with 106dB PSRR 5uV/V?

Mariano,

the PSRR depends on the frequency and you need to specify at which frequency you want to get a specific PSRR.

Also you can increase PSRR significantly by cascading the SMPS with a linear regulator.

However I doubt, that you'll be able to get 106dB at any frequency at all.

Regards

Carsten

Hello Carsten,

PMP8372 has the TPS7A4700 LDO and negative-input TPS7A3301, if these linear regulators aren't not enough is not possible to modify the existing circuit to improve the ripple adding for example come extra capacitors?

Moreover in the specs of the device is not reported any frequency about PSRR, just 5uV/V (-106dB).. the bandwidth of the device is 0-3Khz

Mariano,

I'm sure a colleague from our linear power group will be able to comment specifically on PMP8372.

I have mentioned frequency, because 106dB is a very high target, 5uV are almost noise levels and you can reach extremely high PSRR (how much ever) only for a limited bandwidth.

Regards

Carsten

Hi Mariano,

Carsten is right. 106dB is a very high PSRR for any single LDO. How much ripple at the input are you trying to reject? The output ripple shown in the test report shows almost no ripple when setting the oscope the lowest volts/div we can.

I recommend looking into the TPS7A4700 and TPS7A3301 datasheets to see how PSRR varies with frequency under different operating conditions. 

Increasing CNR, COUT, and CIN, and VIN typically increase PSRR but not as high as 106dB. PSRR will also depend on the load. A higher load usually means lower PSRR. I recommend ordering the TPS7A3301 and TPS7A4700 EVMs to see how well these LDOs perform in your system.

Hello,

just to show you what i'm trying to do, this is the device i want to power up. On page 5 under "Electrical" is written this: Power supply rejection ratio 5μV/V (-106dB)...

I think that is a specification for the magnetic field sensor, not for your power supply.  I read that as saying that the sensor output will reject power supply input ripple at its input by 105 dB.  I think the reference design should work for your application.

Hi John,

thanks a lot for you suggestion, i was going crazy to find something with that PSRR, and i didn't realize that it was referring to the device and not to the PSU that must give power to it.

Please can you provide me information about how changing resistor/other components values to obtain ±12V DC output? i haven't found info about this @ http://www.ti.com/tool/pmp8372

Thanks again

MR

Hello John,

thanks for the information.

Jason Arrigo has shipped the module to us.

Is it expected the board should be modified to obtain desired voltage? (i mean if it's easy to make this changes on it) in my case ±12V output...

Only small few questions:

I've checked datasheet for U1 it results that R8 must be 215K from table and different GND-Float configuration for U2..

There are a couple different configurations for U2 that will result in 12 V output.  I just picked one that wold work.  i edited the previous post to make R8 = 267k.  That Vout of 12.4 V is on the porder between 800 and 900 kHz operation.  But since U3 is operating at 800 kHz recommended, I left U1 at 800 kHz as well.  I do not know if Jason has already modified the EVM.  You will have to check with him, but he probably had it shipped unmodified directly from stock.

Hi John,

we're received the PMP 8372 today, very fast delivery! thank you!

But we have a problem: substituting resistors should be easy, but suddenly i realized that modifying U2 pin configuration is a pretty hard job, is it possible to realize this on delivered sample? or it must need to redesign a new PCB to have ±12Vdc?

Is it possible to have Protel or Altium Designer Project of this EVM ?

Thanks

MR

Hi, I'm wondering also if it is possible to buy somewhere directly this module designed to deliver ±12Vdc output, is it available to purchase?

I looked at the PCB files.  To get that 12 V output you will have to cut some etch or lift pins and "blue wire" the correct pins to GND.  It looks like there is only one version of that board and it is wired for +/- 5 V as you observe.  The files are in PADS format.  Can you use those? 

Hi Mariano,

This EVM was designed using PADS software and not Altium/Protel so the original documents may not be very useful for you. It looks like the output port of the TPS84250 is accessible. I recommending ordering the TPS7A4700 EVM (where the output can easily be adjusted to 12V using jumpers) and connecting the output of the TPS84250 to the input of the TPS7A4700 set to 12V to evaluate this performance. Below is a link its product page.

https://estore.ti.com/TPS7A4700EVM-094-TPS7A4700-Evaluation-Module-P3072.aspx

Regards,

Darwin

Hello,  

Reworking actual Qfn U2 chip is pretty hard job for me,  

Using the 2nd evm could be a temporary solution.. 

The main problem is that we need ASAP a solution to 

Test and deliver +-12v to our equipment, we thought we already solved with this PMP but we don't expect this issue in modify QFN pins configuration 

so main question is: 

Where is possible to purchase PMP already designed to generate 

+-12v? On TI store there is no availability of it..  Maybe you have third party company that sell this item? Please let me know or we have to find another way 

To solve this problem.... 

Thanks 

Hi, 

also if you have the possibility to sell us only 1 sample of this PMP to let us evaluate product as we need it ready to deliver +-12V would be great!

We after can modify PADS project of EVM board and start production of this board ..

Usually those PMP designs are developed for a specific requirement, not as a general marketing tool, so they may not be set up to be easily modified.  I cannot find any +/- 12 V versions.  Even though it is somewhat difficult, I think you should be able to modify the existing board with careful work.  Or you can try the other suggestion of combining it with the LDO EVM.  Let me know if you want the CAD files.  You could modify the PCB file and get new PCBs in a few days time.  Maybe someone in our design servies group could modify the file for you and just send you the gerber files.  I'll ask on Monday.  let me know.

Hello John,

For sure we'll need modified PCB file, so i really appreciate if your design group could send us new PCB  files.

But for the meanwhile actions i'd like like to ask you more details, and sorry if they sounds obviuous but i'm not an expert in this field:

1. Is it the PCB 4 layers or 2 layers?
2. When you say to take the other EVM with jumpers should i just remove U2 from PMP and connect VIN_LDO_POS of TPSA47RGW of the 2nd EVM to pin2 of J4 of PMP?
3. the extra voltage of U1 and U3 of 0.4V are needed from U2 and U4 so that they can generate stable +-12V? what about if i use directly U1 and U3 output? how much greater is their ripple/noise when they generate +-12v? 

It looks like that is a 6 layer board.

I think you can just power the EVM from the 12.4 V and connect it to your load instead of the LDO on the PMP8372.  If you are worried about the power consumption, then you might want to remove U2 and its associated external components.  But I think the easiest rout is to modify the existing PCB to make U2 provide 12 V.

You will get quite a bit more ripple and noise if you use just a switching regulator.  You application may well work with that, but you will have to test to be sure.

I tried to contact the designer of that reference design, but there is no valid email for him.  He may no longer be with the company.  I'll see if i can find someone from that producr line who supports this.

Hello John,

i have tried to remove U2 to modify pins connection but it was pratically impossible to desolder with hot air and an experienced technician, also passive components don't go away..

Finally ldo get damaged.. so we're waiting for modified gerber files... The problem is that it is also a 6layers PCB...so it is very expensive to produce and cost for every PCB will easily be more than 250€ ... The PMP is a very small circuit so if there is available also the same circuit with a 2 layer PCB would be perfect..

Please let me know.

MR

I'm going to forward this to someone who may be able to help.  I don't actually support any of those particular ICs, but I often answer forum posts anyway.

Hey Mariano,

Please email me with the load requirements. both the power module and LDO can output 12V so we should be good; might have to cut some traces however.

my email is sarmad@ti.com

thank you