SG3525 loop

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Topology

Lorenzo

I just recently completed an SG3525-based 2-Phase DC-DC non-isolated Upconverter. Both inductors are wound on the same #26 powdered-iron toroid and rectified into a 25A 45PIV dual-schottky. Output is 19-20V at something like 8-9A. (insert Tim Allen grunt here)

For the uncompensated feedback, use a voltage divider like electrone suggests. For loop-compensation, according to Brown, a good choice for this topology is "in-band with gain-limiting", which is just an RC paralleled combo, running from the '3525's pin 9 (compensation) to signal ground. For a frequency of ~100kHz, I chose 49.9k-ohm paralleled with a 1000pF polyester capacitor.

Here is a pic of my converter.

Cheers,

Steve
 

Attachments

  • sg3525 dc-dc uppie 1 smaller.jpg
    sg3525 dc-dc uppie 1 smaller.jpg
    87.2 KB · Views: 474
Any reason why you choose to use the SG3525????
It is ancient and full of problems... The error amp is very crapy..
This is a Voltage Mode part....
The Boost converter is not trivial when closing the loop... If you don't get it right it will blow up on you...
First need to know the switching frequency you are operating at..... Is this 2-Phase????
Is the output inductor calculated for Continous or DisContinous Operation????? If it is Continous Mode then you will have a RHPZ...
Need to see your schematic with values...and ESR of output cap...
Next procedure is to determine were to set the Cross-Over frequency based on your switching frequency and location of the RHPZ, to put it below this point...
BUT since the Error-Amp is very bad...you are forced to Cross-Over at a much lower frequency then normal...leaving you with a slow response loop....
The transfer function of the OPEN lOOP system for a boost is well known..then you choose typically a TYPE-3 error amp configuration.... The catch is that a BOOST circuit will contain a RHPZ, RIGHT HAND PLANE ZERO, this needs to be dealt with in the error amp compensation, you need to determine at what frequency is RHPZ occurs in your system.....
I will be more than happy to assist you if need be...
Checkout the LT3782 , it is a 2-Phase Step-Up DC/DC Controller........

Chris
 
The LT3782 might be a great chip, but it is 10 times the price, not broadly available, and only comes in a SSOP-28 package. The SG3525’s are $0.35 each in quantities of one at Newark and can be obtained in a DIY-friendly DIP package. I guess we need to know Lorenzo_cy’s application…

PS: Hello fellow San Diegan Chris!
 
Re: Topology

cerrem said:
Any reason why you choose to use the SG3525????
It is ancient and full of problems... The error amp is very crapy..
This is a Voltage Mode part....
The Boost converter is not trivial when closing the loop... If you don't get it right it will blow up on you...
First need to know the switching frequency you are operating at..... Is this 2-Phase???? ........... I will be more than happy to assist you if need be...
Checkout the LT3782 , it is a 2-Phase Step-Up DC/DC Controller........

Chris


N-Channel said:
Lorenzo

I just recently completed an SG3525-based 2-F DC-DC non-isolated Upconverter. Both inductors are wound on the same #26 powdered-iron toroid and rectified into a 25A 45PIV dual-schottky. Output is 19-20V at something like 8-9A. (insert Tim Allen grunt here)

For the uncompensated feedback, use a voltage divider like electrone suggests. For loop-compensation, according to Brown, a good choice for this topology is "in-band with gain-limiting", which is just an RC paralleled combo, running from the '3525's pin 9 (compensation) to signal ground. For a frequency of ~100kHz, I chose 49.9k-ohm paralleled with a 1000pF polyester capacitor.

Here is a pic of my converter.

Cheers,

Steve


Ya' need to read the posting a little more closely. ;) Sorry- could not post the schemo. Either I'm a little too tired, or I just haven't figured it out yet.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.