Author Topic: semi-active bridge rectifier  (Read 1332 times)

0 Members and 1 Guest are viewing this topic.

Offline exe

  • Supporter
  • ****
  • Posts: 2328
  • Country: nl
  • self-educated hobbyist
semi-active bridge rectifier
« on: September 23, 2020, 12:51:00 pm »
Hi!

For a long time I wanted make a semi-active bridge rectifier. It's half-way circuit between a traditional diode bridge, and a full-fledged circuit using lt4320.

Photos, circuit, etc is here: https://github.com/kopchik/semi_active_rectifier . Here I just post the results of quick evaluation with DC current and compare it with a traditional bridge in GBU package.

Measurement current: 2A, 4V. My circuit has dropout voltage of 0.365V vs 1.6V for traditional silicon diode bridge. At voltages below 4V the mosfets are not conducting, making drop increasing up to ~0.6V (two diodes conducting, I'm using FERD diodes from ST).

Why not using lt4320? My circuit works below 9V :). Other than that, I was curious for years about it, and finally implemented it.
 

Offline bob91343

  • Super Contributor
  • ***
  • Posts: 2052
  • Country: us
Re: semi-active bridge rectifier
« Reply #1 on: September 24, 2020, 12:01:28 am »
Well you are smart enough to do this and make it work so I congratulate you.  However, now that it's done I doubt that many useful applications exist.  The parts count and cost make it impractical for most purposes, I think.
 

Offline exe

  • Supporter
  • ****
  • Posts: 2328
  • Country: nl
  • self-educated hobbyist
Re: semi-active bridge rectifier
« Reply #2 on: September 24, 2020, 07:06:58 am »
I guess it depends on size of the project. My next power supply contains ~250parts (each channel). In this case +10 parts is not significant. I'll also won't use a separate board, the rectifier will be soldered on the main board.

edit: typo
« Last Edit: September 24, 2020, 01:59:34 pm by exe »
 

Offline Siwastaja

  • Super Contributor
  • ***
  • Posts: 4201
  • Country: fi
Re: semi-active bridge rectifier
« Reply #3 on: September 24, 2020, 01:53:19 pm »
Well you are smart enough to do this and make it work so I congratulate you.  However, now that it's done I doubt that many useful applications exist.  The parts count and cost make it impractical for most purposes, I think.
LT4320 is freaking expensive. The total cost of exe's solution, even taking into account the increased BOM lines, increased P&P setup and pick&place time, and slightly increased PCB area, is likely significantly lower. If it works reliably and provides loss reduction that is good enough for some applications, then I do see the appeal over the solution based on LT4320.
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 15599
  • Country: za
Re: semi-active bridge rectifier
« Reply #4 on: September 24, 2020, 04:56:30 pm »
You could actually dump D2 and D3 and rely on the MOSFET body diodes to do low frequency rectification as well, as they typically have the same current and voltage ratings as the mosfet, just really poor turn on and off times, which here with 50/60Hz mains use would not be an issue.
 

Offline Siwastaja

  • Super Contributor
  • ***
  • Posts: 4201
  • Country: fi
Re: semi-active bridge rectifier
« Reply #5 on: September 24, 2020, 05:08:07 pm »
You could actually dump D2 and D3 and rely on the MOSFET body diodes to do low frequency rectification as well, as they typically have the same current and voltage ratings as the mosfet, just really poor turn on and off times, which here with 50/60Hz mains use would not be an issue.

The body diodes have much higher Vf than the FERD types used by the OP. This would be important in case of low input voltage when the MOSFETs won't fully conduct.
 

Offline MasterT

  • Frequent Contributor
  • **
  • Posts: 536
  • Country: ca
Re: semi-active bridge rectifier
« Reply #6 on: September 24, 2020, 05:58:56 pm »
And what happens at zero-crossing, what makes MOSFETs to operate? Unlike  SCR regulator with 150 peak voltage, conducting phase angle for MOSFET seems too big and depends on this voltage too much
 

Offline exe

  • Supporter
  • ****
  • Posts: 2328
  • Country: nl
  • self-educated hobbyist
Re: semi-active bridge rectifier
« Reply #7 on: September 24, 2020, 06:18:14 pm »
And what happens at zero-crossing, what makes MOSFETs to operate? Unlike  SCR regulator with 150 peak voltage, conducting phase angle for MOSFET seems too big and depends on this voltage too much

I'm not sure I get the question. At low voltages mosfets are off, but they are shunted with diodes. At higher voltages they are on. Since they do not block reverse current when on, the upper diodes (on the schematic) block it.
 

Offline MasterT

  • Frequent Contributor
  • **
  • Posts: 536
  • Country: ca
Re: semi-active bridge rectifier
« Reply #8 on: September 24, 2020, 06:29:35 pm »
And what happens at zero-crossing, what makes MOSFETs to operate? Unlike  SCR regulator with 150 peak voltage, conducting phase angle for MOSFET seems too big and depends on this voltage too much

I'm not sure I get the question. At low voltages mosfets are off, but they are shunted with diodes. At higher voltages they are on. Since they do not block reverse current when on, the upper diodes (on the schematic) block it.
The question when MOSFET starts to conduct. For example common Si diode starts at 0.7V.
 

Offline exe

  • Supporter
  • ****
  • Posts: 2328
  • Country: nl
  • self-educated hobbyist
Re: semi-active bridge rectifier
« Reply #9 on: September 24, 2020, 07:34:56 pm »
Depends on mosfet. I used CSD18543Q3A (declared Rds(on) is 12mOhm at 4.5V typical), at 4V the measured Rds was 33mOhm (which agrees with the datasheet), so it starts conducting somewhere below 4V. It's a bit too high for my application as, ideally, I need mosfet to conduct at below ~2.5V. But I didn't find suitable mosfet with such low Vgs(th), and Vds > 30V (I need at least 50 for reliability reasons). So, I'd say this approach doesn't work for voltages below 4V (unless one finds a proper mosfet).

One caveat is that mosfets have to have quite low Rds(on). That's because the peak current into a capacitive load my simulation is 20A (3A average current into 10_000uF capacitor). Since FERD drop is only ~0.55V (I used FERD15S50), in order mosfet to "steal" current from the diode, it needs to have voltage drop below that of the diode (otherwise most of the current will go via the diode). This means that Rds should be << 0.55V/20A=0.027 Ohm. That's why I'm using 10mOhm fets.


PS for no particular reason I'm attaching a pic of my diy power supply's rectifier board with the rectifier module installed.
 

Offline mag_therm

  • Regular Contributor
  • *
  • Posts: 193
  • Country: us
Re: semi-active bridge rectifier
« Reply #10 on: September 25, 2020, 02:22:28 pm »
For control on the upper diodes, you could use a dual optocoupler?
 

Offline exe

  • Supporter
  • ****
  • Posts: 2328
  • Country: nl
  • self-educated hobbyist
Re: semi-active bridge rectifier
« Reply #11 on: September 25, 2020, 02:49:51 pm »
For control on the upper diodes, you could use a dual optocoupler?

You mean to create bias voltage to drive the gate? It could also be done with just p-fets. The problem is, when a fet is open, it conducts current both ways. So, to avoid the capacitor discharging via fets we need a controller (or a comparator) that senses voltage accross the fets and closes it avoid "backfeeding" from capacitor. Here is a more detailed explanation: https://www.eevblog.com/forum/projects/bridge-rectifier-made-with-mosfets/
 

Offline mag_therm

  • Regular Contributor
  • *
  • Posts: 193
  • Country: us
Re: semi-active bridge rectifier
« Reply #12 on: September 25, 2020, 03:29:48 pm »
I was thinking of a storage cap with blocking diode on each of the upper diodes.
The opto inputs would be between R1 (etc) and ground.
The opto outputs would drive the upper fets from their respective storage caps
 

Offline strawberry

  • Frequent Contributor
  • **
  • Posts: 381
  • Country: lv
Re: semi-active bridge rectifier
« Reply #13 on: September 25, 2020, 04:53:18 pm »
How about this integrated circuit TEA2208
 

Offline exe

  • Supporter
  • ****
  • Posts: 2328
  • Country: nl
  • self-educated hobbyist
Re: semi-active bridge rectifier
« Reply #14 on: September 27, 2020, 08:34:29 am »
How about this integrated circuit TEA2208

Yeah, found it a few days ago too :). Looking closer, I see differences from lt4320. The most obvious to me is that it has a separate capacitor for internal supply. It also requires two external capacitors for for high-side drive. Here is the reference design: https://www.nxp.com/products/power-management/ac-dc-solutions/ac-dc-controllers/active-bridge-rectifier-controller-demo-board:TEA2208DB1576 .

I added one to my order. Although, not sure if I'll have time to play with it.
 

Offline iteratee

  • Contributor
  • Posts: 41
  • Country: us
  • Computer Engineer
Re: semi-active bridge rectifier
« Reply #15 on: September 27, 2020, 12:40:27 pm »
The LT4320 gets used for mains rectification? How is it biased to keep its supply within voltage limits while still being able to drive the gates throughout the full cycle? I've dismissed using it because I couldn't think of a good solution. You'd need to pull some supply bootstrapping shenanegans to make that work.
« Last Edit: September 27, 2020, 12:58:05 pm by iteratee »
 

Offline exe

  • Supporter
  • ****
  • Posts: 2328
  • Country: nl
  • self-educated hobbyist
Re: semi-active bridge rectifier
« Reply #16 on: September 27, 2020, 01:50:48 pm »
How is it biased to keep its supply within voltage limits while still being able to drive the gates throughout the full cycle?

The datasheet says it has internal charge pump. Sorta makes sense, considering the drive current for top transistors (120-425uA). May also partially explain why it's so expensive, IC capacitors may take a lot of real estate.
 

Offline zitt

  • Regular Contributor
  • *
  • Posts: 98
  • Country: us
    • Pinball-Mods.com
Re: semi-active bridge rectifier
« Reply #17 on: September 29, 2020, 06:55:52 am »
cool. I've been using a much simpler design for some Pinball mods. Having said that; mine doesn't function at low voltages either. No zeners. No additional diodes. Just using body diodes.

I've been experimenting with charge pumps to see if I can get better VGS ... but right now only in follow on LDO stages. I'm wondering if a charge pump could be shoehorned into the design to get better transistor profiles.
 

Offline exe

  • Supporter
  • ****
  • Posts: 2328
  • Country: nl
  • self-educated hobbyist
Re: semi-active bridge rectifier
« Reply #18 on: September 30, 2020, 10:08:11 am »
I've been experimenting with charge pumps to see if I can get better VGS ... but right now only in follow on LDO stages. I'm wondering if a charge pump could be shoehorned into the design to get better transistor profiles.

It might be easier to use p-fets for high side, as they biased below the source == no charge pumps needed :). I understand you struggle,  for me it was hard to find the balance between price, part count and complexity. All solutions have their own problems.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf