Solution Finder Get Tech Support

30v to 400 v isolated design without opto

Posted by: bougnoul on

How do I lower the input voltage to 30V for an isolated 3.3v design with only 500mA output? My input will go up to 400V. Do not

want to use optocoupler. So LINKSWITCH  SOIC is an option but the PIExpert wont run down to 30V. It does so from 60V

any ideas?

Comments

Submitted by PI-Terry on 10/21/2014

Hi,

 

Are you thinking to use LinkSwith II for your design with flyback? The constraint of the minimum input voltage is more from the internal shunt that supply the BP pin voltage during the startup by taking into account all the tolerance of the components value under the worst conditions like hold and cold temperature. However, you should still be able to use the IC with a wide input voltage range and lower input voltage if you can select the compoenents with more margin.

In the PI expert software, particularly you could input the input voltage into VMIN and VMAX (dc input voltage), and check if any IC can reach the design requirement. Kindly remind to check all the parameters especially at low line voltage, the Dmax, IP...ect.

Submitted by bougnoul on 10/21/2014

the model works down to 50V & I think I can use that

dmax comes out to 58%

thnx a lot

Submitted by bougnoul on 10/21/2014

teri

could you comment on my scheme of enable/disable function of the LNK626 based converter? The circuit is per PiExpert tool. Only thing I am adding is the EN function.

Which one would you recomment?

I need to go into PCB fab using LPFK machine as soon as I hear on this from you.

thnx

Submitted by PI-Terry on 10/27/2014

Hi, 

 

If you short the FB pin to ground, the power supply will enter into open loop fault condtion. Here is what will happen:

The sensed FEEDBACK pin current during the forward period of the conduction cycle (switch “on” time) falls below 120 mA, the converter annunciates this as an open-loop condition (top resistor in potential divider is open or missing) and reduces the auto-restart time from 450 ms to approximately 6 clock cycles (90 ms), whilst keeping the disable period of 2 seconds.

Please refer to the datasheet of LNK626 in page 3:

http://www.powerint.com/sites/default/files/product-docs/linkswitch-ii_family_datasheet.pdf

 

Is this something that you expected to happen? 


Submitted by bougnoul on 10/27/2014

Terry

The referenced document states that the duty cycle will be like .01%, which is almost "shutdown" condition.

Since  there is no conflict with this mode, I find it is ok & can be used as enable/disable of the output.

I have the option of using the  BP pin instead of the fb pin. Is that  better for "disable" function of the converter?

Unless, you can suggest some other enable/disable circuit for  this LINKSWITCH which can compleltley keep it on shutdown mode.

pl see attached schematic I am routing right now.

thnx

Submitted by bougnoul on 11/03/2014

Terry

When I open PIExpert, I do not see LNK6773 device at all.

Can I use results from LNK626DG design?

Hi,

Unfortunately, the LinkSwitch HP products have not been included in the PI-expert software yet, and you can not use the results from LNK616DG, as they are very different in terms of operation. You could use the PIxls, which is part of the PI-expert software to help you design, the LinkSwitch H products have been included in PIxls.

 

Thanks 

Submitted by bougnoul on 11/04/2014

Terry

I could not use the PIXLS & could not figurte out why.

Also, the default spread sheet only has AC input. NO DC?

Mine comes from 3phase rectified & filtered aircraft power.

I found Digikey only has LNK67774 & I am going to use that.

Now, yes, I figured f=132kHz will make a lot of difference.

The schematic remains the same but values will change.

I designed the xmfr using Colorado U flyback design rules(most likey from Robert Erickson).

What else would be different?

.The spread sheet does not seem to have edit capability.

My PCB is in fabrication now: it is 4-layer board. My returns  are very low inductance because I am using copper planes in 2 inner layers- for heat sinking as well.

BTW: I am using LDO to get low ripple output. LT3080-1

is that ok?

thnx

Submitted by bougnoul on 11/04/2014

Terry

I looked at the  rdr321 FILE. From this I can say that my schematic is right. Values seem to be  also close as far as bp,pd , cp & fb pins are concerned. Clamping may be ideally with different values but with low power I have, I doubt if that will matter a lot. Because I am not restricted to have the lowest cost, I can go fancy & make very low leakage transformer. I have no electrolyic but only large MLCC/X7R caps.

Onyl thing is transformer pr, secondary & bias windings. They comes out pretty much from simple scaling. I can olny use RM7ILP due to ht restrictions.

But of course I would like to verify using the PI XLS if I know how to use it.

b

Submitted by PI-Terry on 11/06/2014

Hi,

 

I am not so sure what exactly the issue of using the PIxls, you may try to uninstall it and install it again.

You can just input the DCmin and DCmax, then it is related to the DC input voltage. The good thing for spreadsheet is to help you do the design, it follows all the basic theory and rules. Some worst cases (tolerance of the inducatance, currnt limit tolerance...ect,) has been also taken into account. You can definitely design the transformer by yourself if you feel more comfortable.  

Submitted by bougnoul on 11/06/2014

Terry

I got it working.

But seems strange!

I have a 3.3V @ 330mA DESING THAT CAME OUT WITH LOTS OF TURNS.

SEEMS IT TRIES TO KEEP CCM

I would keep same primary winding, just change secondaries.

15V came out ok excepting that I cannot  find the right size core that will fit in my vol available.

My xmfrs  are done & pcb is in machine. we will find out next week, thnx 

lpri= 4mH or somethign

leakage<5uH

cww=770pf