Help design to isolate a 20-36V DC input to 24-30V DC output

[lutus]

Hi, first post for myself.

I have a 20-36 V DC input and need an isolated output of 20-30V DC. This is a low power application, less than 1 amp. I know I could use an off the shelf dc-dc converter but I don't need 30W if I can do it with a couple of chips and a small transformer.

Anyone have any ideas?

Thanks for help in advance.

PS: I also have a 24V TO-220 regulator that I can use on the input, but I need a 24V ISOLATED output. Just throwing that out there.

 

[crutschow]

You say you don't need 30W, but 20-30V @ 1A is 20-30W. So how much current do you need?

 

[dougy83]

You'll want a flyback converter, and you can use pretty much any controller IC that can handle your power requirements. Attached is the basic circuit (hacked out of some datasheet). If the output voltage needs to be more accurately controlled, you can use e.g. a TS431 in its typical arrangement.

 

[Joe Caughey]

Hi, my first post as well. Not even sure this is the correct place to post this. sorry if not.

I'm trying to build a renewable resource lab using small DC motors and small solar panels for a input voltage I'm thinking will be .5V to maybe 2V output maybe. My question is: is there a way I can build a device that takes the output of the students generators and control the variable output of a 3V battery display? So depending on the power generated the output (lights or a meter) will vary conversely? I think this will help the kid understand the concepts but make it much more exciting.

 

[Vizier87]

Joe, you might wanna post a new thread than replying to this thread, since the discussions here will only pertain to the first post. Welcome to the forum, anyway.

 

[lutus]

You say you don't need 30W, but 20-30V @ 1A is 20-30W. So how much current do you need?

I will be sending this voltage to a black box that will be charging the internal battery card of this black box. The trickle charge current will be around 17mA and at full current draw I think it is around 30mA. When I said below 1A I probably should have brought that ceiling down to "less than 50mA". Big difference!

 

[lutus]

You'll want a flyback converter, and you can use pretty much any controller IC that can handle your power requirements. Attached is the basic circuit (hacked out of some datasheet). If the output voltage needs to be more accurately controlled, you can use e.g. a TS431 in its typical arrangement.

View attachment 37689

This flyback converter looks very promising. I have found a link to a document that steps through the process of choosing your values.

Powered by Google Docs

I believe that I should start with a linear voltage off the shelf regulator that can handle up to 36V to normalize my voltage input to a steady voltage (not variable) and feed that output into a flyback converter circuit. What do you think?

A little background. The variable voltage of 20V to 36V is actually a bunch of DC batteries. At full charge they will be 36V but over time they drain until the system shuts off at <20V.

 

[crutschow]

This flyback converter looks very promising. I have found a link to a document that steps through the process of choosing your values.

I believe that I should start with a linear voltage off the shelf regulator that can handle up to 36V to normalize my voltage input to a steady voltage (not variable) and feed that output into a flyback converter circuit. What do you think?

That would significantly reduce your efficiency since a linear regulator just wastes the excess voltage.

doughy83's flyback converter is regulated so the output is basically independent of the input voltage. It should have good efficiency at all input voltages if correctly designed.

 

[lutus]

That would significantly reduce your efficiency since a linear regulator just wastes the excess voltage.

doughy83's flyback converter is regulated so the output is basically independent of the input voltage. It should have good efficiency at all input voltages if correctly designed.

Ah, I see what you mean. After a little more reading I see that these things should be designed so that they can input a large range of inputs.

Would anyone care to suggest a "SMPS IC" and transformer based off of dougy83's schematic? (Something like the Linear LT3573?)

The output voltage does not have to be too tightly controlled as long as it 20V<X<30V.

Thanks for the help guys!

 

[crutschow]

Here's an ap note for a reasonably simple isolated flyback regulator using an LM2577 (https://www.electro-tech-online.com/custompdfs/2010/01/AN-777.pdf). It has 3 outputs, but with proper selection of the winding, you should get the output you want (with the magnetic core shown, a single 80 turn output winding should give you 25V). You also need to change the voltage feedback resistor R02 from 45.3k to 2.61k.

 

[lutus]

Here's an ap note for a reasonably simple isolated flyback regulator using an LM2577 (https://www.electro-tech-online.com/custompdfs/2010/01/AN-777-1.pdf). It has 3 outputs, but with proper selection of the winding, you should get the output you want (with the magnetic core shown, a single 80 turn output winding should give you 25V). You also need to change the voltage feedback resistor R02 from 45.3k to 2.61k.

Thanks Carl!

I have printed out all the datasheets except I can't find this transformer. Is this a particular off the shelf transformer or will I have to find build one?

Also, Vout2 and Vout3 can just be deleted from that schematic and I will use Vout1 with that resistor change to give me the correct voltage?

Which LM2577 package should I use? 5V, 12V, ADJ?

I just need a little more help with the transformer. (If I could use an off the shelf transformer and modify this circuit a little that would be the best. Something like https://www.coilcraft.com/mintrans.cfm#table maybe? The 5V to 24V T6650-DL)

Thanks in advance.

 

[crutschow]

Transformers are always the rub in designing switching inverters. If you notice the LM2577 notes at the beginning calls out a Ferroxcube core where you wind you own transformer.

Rewind--Looking at the Coilcraft website I found a transformer that should work for you Coilcraft Q4438-BL Surface Mount Flyback Regulator Transformer. You would connect both output windings in series to get 24V.

It shows a different IC (LM258X) **broken link removed**. This ap note https://www.electro-tech-online.com/custompdfs/2010/01/AN-1095.pdf discusses using it as an isolated converter (Figure 6). It's actually a simpler circuit than the one using the LM2577

To get 24V output or so you would need to change the LM3411 regulator. Since you are not overly concerned about regulation accuracy I would try substituting a 22V zener diode in series with the output voltage and Ro, to replace the LM3411.

You may have to play with the value of Ro and perhaps add additional compensation as shown in Figure 14 to improve the converter stability.

P.S. Why do you need an isolated output?

 

[lutus]

Transformers are always the rub in designing switching inverters. If you notice the LM2577 notes at the beginning calls out a Ferroxcube core where you wind you own transformer.

Rewind--Looking at the Coilcraft website I found a transformer that should work for you Coilcraft Q4438-BL Surface Mount Flyback Regulator Transformer. You would connect both output windings in series to get 24V.

It shows a different IC (LM258X) **broken link removed**. This ap note https://www.electro-tech-online.com/custompdfs/2010/01/AN-1095-1.pdf discusses using it as an isolated converter (Figure 6). It's actually a simpler circuit than the one using the LM2577

To get 24V output or so you would need to change the LM3411 regulator. Since you are not overly concerned about regulation accuracy I would try substituting a 22V zener diode in series with the output voltage and Ro, to replace the LM3411.

You may have to play with the value of Ro and perhaps add additional compensation as shown in Figure 14 to improve the converter stability.

P.S. Why do you need an isolated output?

Attached I have a hand drawn sketch that replaces the LM3411 with a zener diode. Also, I have included the transformer that you specified from Coilcraft.

The LM2587 will be a +12V version correct?

Is this looking closer toward what I would be looking for?

The reason I need isolation is because the input voltage from the system batteries is anywhere from 20V to 36V. This flyback will be incorporated into a larger power supply design that is using Vicor DC-DC converters for +5V, +12V, and +28V. When the system is powered OFF I still need to get a small amount of voltage onto the secondary side of these converters without having to power up one of these larger Vicor modules. I could draw you a rough sketch but the problem is I don't want to switch grounds or anything like that, I just need a little secondary side voltage.

 

[carmusic]

just buy somthing like this one, it will cost a lot less than building one since you only need 30ma
XP POWER|IL2424S|DC/DC Converter | Newark.com

 

[lutus]

crutschow, I replied with an attachment .pdf but it must be reviewed by a moderator because of my low post count I guess. Hopefully that will come through soon.

carmusic, according to that datasheet the input voltage is "Nominal +-10%". It seems that the 48Vdc input module would allow input of 43.2Vdc and 52.8Vdc. I need an input of 20V to 36V. Unless I am reading that incorrectly I am not sure this would work.

 

[Claude Abraham]

use buck-boost converter

This sounds like an ideal application for a buck-boost converter. The output is in between the values of the minimum & maximum input voltage. TI has plenty of app notes. Download the buck-boost app note by Everett Rogers. There are 3 he authored, SLVA057, 059, & 061, which cover buck, boost, & buck-boost.

A buck-boost requires a simple inductor, or 2 identical units in series. This is an off the shelf part. A flyback converter requires a transformer, which adds complexity, and is not a catalog item. Winding your own can be done, but it is more work and what does it offer? A flyback does not provide better performance than a buck-boost, so why bother with a custom transformer?

Peak current mode control is what I recommend, using a UCC28C42/43 family of control IC parts from TI. I authored an article in EDN back in 2006 which I've attached. It gives some insight into the buck-boost convrter. I've used this approach and I have hundreds of thousands of products in worldwide use utilizing buck-boost converters, all with great success.

Other options include the boost-buck converter, and the SEPIC converter. The SEPIC is tricky to compensate, as it has 3 right half plane zeros and 2 pairs of complex conjugate poles. It's a good converter, but should be used only by highly experienced people. The SEPIC is advantageous at low output voltages due to the single rectifier, as opposed to 2 rectifiers for the buck-boost & boost-buck. At 24V out, the SEPIC is not very advantageous. It can be used successfully, but the other 2 are better suited. The boost-buck has 2 RHPZ & 2 complex pole pairs. Compensation is tough. The buck-boost has 1 RHPZ & 1 complex pole pair making it the easiest to compensate as well as the fastest.

Did I help?

 

[dougy83]

carmusic, according to that datasheet the input voltage is "Nominal +-10%". It seems that the 48Vdc input module would allow input of 43.2Vdc and 52.8Vdc. I need an input of 20V to 36V. Unless I am reading that incorrectly I am not sure this would work.

Have a look at **broken link removed** or **broken link removed**

A flyback converter requires a transformer, which adds complexity, and is not a catalog item. Winding your own can be done, but it is more work and what does it offer? A flyback does not provide better performance than a buck-boost, so why bother with a custom transformer?

Well, it provides isolation, for a start.

 

[Claude Abraham]

Have a look at **broken link removed** or **broken link removed**

Well, it provides isolation, for a start.

I must have missed the isolation part in your OP. Yes a buck-boost provides no isolation. A flyback would be the logical choice. The buck-boost app note from TI also covers flyback operation. It would be helpful.

 

[crutschow]

Deleted due to duplication.

 

[lutus]

dougy83 is the winner!!!

I ordered that XPpower IW2424SA off the shelf item and it works EXACTLY like I needed. For only 10 to 15 dollars a piece this is a no-brainer.

The only problem now is trying to track down 50 or 60 of these things. I already have a call into the parent company.

Thanks to everyone who gave all the suggestions. Never would have found the perfect component without your help.

Clay