Author Topic: Half-Bridge 0-30V, 0-20A Feedback loop problem  (Read 10624 times)

0 Members and 3 Guests are viewing this topic.

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 3034
  • Country: au
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #50 on: July 19, 2019, 12:37:51 pm »
Adding the extra filter board and passing output current through the CMC would likely be making things worse.
Did you try a bifilar wound CMC  from the output capacitors to the U sense input on the control board? As shown in Reply #28.
You can do a test to eliminate the feedback loop completely. Use a low value pot connected to the 5V reference and use the wiper voltage to directly control the duty cycle by connecting directly to pin 9 or to the input of the SG3525's amplifier configured as a voltage follower as it was originally. Pin 9 can be bypassed with a 10uF capacitor to be certain that there is no noise getting into the PWM comparator.
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #51 on: July 20, 2019, 07:27:03 pm »
Have you tried bifilar CMC from the output capacitor to the input sensor U on the control board? As shown in the answer # 28.

Yes, but the inverter behavior was even worse. The waveforms had more distortion and the transformer worked louder.

I did a test with a potentiometer connected to pin 9 of SG3525 as you wrote, and I noticed the following dependence:
The converter turns on at about 0.6V on pin 9, which, with my load of about 1ohm, gave 5V at the output and 5A. The transformer worked quietly all the time, but somehow the output runs were quite weak. Pretty noisy. Maybe I proved myself that the 47uH choke is not a good move. At 25A it's about 10uH is not much.

At the Type 3 compensator I also noticed that the transformer is pulsing and buzzes when the output voltage goes below 5V
that is, when the voltage at pin 9 drops below 0.6V and the inverter switches off.

I also noticed that the current loop worked much worse from the voltage through the INA213 operational amplifier. Maybe it has too small bandwidth, but the amplifier itself made the loop much worse. I omitted the INA213 and 4 ceramic resistors 5W 0.1 ohm connected in parallel, which are directly connected to the loop amplifier.

Below are some photos from the work of direct control of pin 9 of SG3525 from the potentiometer.
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #52 on: July 20, 2019, 07:28:24 pm »
2
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #53 on: July 20, 2019, 07:30:34 pm »
3
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #54 on: July 20, 2019, 07:31:54 pm »
4
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #55 on: July 20, 2019, 07:32:47 pm »
5
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #56 on: July 20, 2019, 07:34:09 pm »
6
 

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 3034
  • Country: au
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #57 on: July 21, 2019, 06:08:46 am »
The 100Hz ripple is normal because there is no feedback loop to correct it.
 The main thing is that the PWM timing is stable with no jitter or missing pulses.
All appears fine to me and proves there is nothing odd affecting the drive.
I have no experience with a direct current control loop.
For a 40A battery charger project, I used fast pulse by pulse limiting and a slow loop that decreased the reference voltage to the voltage error amp.
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #58 on: July 26, 2019, 08:47:43 am »
Have you met with a situation where pin 5 (CT) is connected to pin 1 (INV AMP) through a resistor?
Maybe you know why it's used?

 

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 3034
  • Country: au
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #59 on: July 26, 2019, 11:23:03 am »
Have you met with a situation where pin 5 (CT) is connected to pin 1 (INV AMP) through a resistor?
Maybe you know why it's used?
No, do you have an example? I guess that it could improve noise immunity.
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #60 on: July 26, 2019, 11:49:02 am »
The connection I mentioned is in the schematic of the power supply, which I put in the attachment.
« Last Edit: July 26, 2019, 11:55:55 am by paladyn »
 

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 3034
  • Country: au
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #61 on: July 26, 2019, 11:09:59 pm »
The connection I mentioned is in the schematic of the power supply, which I put in the attachment.
I don't understand the description, another language of some sort.
He is possibly trying to partly cancel 66Khz ripple that's getting into the error amp.
Something looks odd with the MOSFET Gate drive transformer. I think that the high-side MOSFETs would get half of the drive voltage.
« Last Edit: August 01, 2019, 04:04:46 pm by xavier60 »
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #62 on: July 30, 2019, 05:33:21 am »
I made an experiment and connected  pin 1 to pin 5 (SG3525) with an adjustable resistor 500 kohm, no visible effects.

EDIT:

I just realized what a blunder I made.

After all, I do not use the error amplifier in the external SG3525 only. Maybe you would have to connect pin 5 to the COMP9 pin output
« Last Edit: July 30, 2019, 06:42:24 am by paladyn »
 

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 3034
  • Country: au
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #63 on: July 30, 2019, 09:03:04 am »
Does it run smoothly by controlling PWM by applying filtered DC to pin 2 of the SG3525's error amp?
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #64 on: July 31, 2019, 07:07:57 am »
Does it run smoothly by controlling PWM by applying filtered DC to pin 2 of the SG3525's error amp?
pin 1 and pin 2 of the SG3525 system are not used and not connected. Pin 9 is connected to an external operational amplifier that realizes the loops.


I am using an external operational amplifier, but yesterday I did a test where I connected an adjustable 500kohm pin 5 resistor with pin 9 of the SG3525 chip

I must admit that I did not notice the difference in the operation of the inverter. At a small value of the resistor some soft anomalies already appeared in the noise at the inverter output and I stopped reducing the resistances.

Overall, I noticed that the problem occurs when SG3525 can no longer reduce PWM because it has almost reached zero. Perhaps the use of a huge choke would help?

Summing up the problem of squeezing the transformer occurs mainly at low resistance loads and low output voltages.
 

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 3034
  • Country: au
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #65 on: July 31, 2019, 07:42:30 am »
Does it run smoothly by controlling PWM by applying filtered DC to pin 2 of the SG3525's error amp?
pin 1 and pin 2 of the SG3525 system are not used and not connected. Pin 9 is connected to an external operational amplifier that realizes the loops.


I am using an external operational amplifier, but yesterday I did a test where I connected an adjustable 500kohm pin 5 resistor with pin 9 of the SG3525 chip

I must admit that I did not notice the difference in the operation of the inverter. At a small value of the resistor some soft anomalies already appeared in the noise at the inverter output and I stopped reducing the resistances.

Overall, I noticed that the problem occurs when SG3525 can no longer reduce PWM because it has almost reached zero. Perhaps the use of a huge choke would help?

Summing up the problem of squeezing the transformer occurs mainly at low resistance loads and low output voltages.
That's right,the pulse width can't go below 600ns so the SG3525 has to skip pulses and causing some transformer noise. I can't hear it with the covers fitted.
My choke is about 80µH at high load current.
Driving the MOSFETs too hard can cause extra interference.
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 

Offline mrjoda

  • Regular Contributor
  • *
  • Posts: 80
  • Country: sk
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #66 on: July 31, 2019, 01:13:30 pm »
hello guys.

At the beginning : paladyn, very nice project, i will definitely watch it.

I am not able to help you with compensation of feedback, but ... we are working with smps and converters in work and we are using network analyzer for feedback loop. You can simply check compensator pole/zero locations according to nyquist criteria and tune the feedback loop to sweet spot. Without it, it is only shooting to dark, because you don't know the amount of phase margin. Is there any chance for you to borrow one or visit some lab with correct instruments and tune it correctly ? It will save you a LOT of time.


At first, i thought the smps are easy to design... now,when i know some background, i don't have balls for my own design :)
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #67 on: August 01, 2019, 07:46:50 am »
mrjoda
Thank you for the words of appreciation. Indeed, the design of impulse power supplies is a difficult matter. At first it looks like it's quite a simple thing, but the amount of detail determines how the power supply then works. As you mentioned, testing the loop without network analyzer is like working in the dark. I will try to look for a person or institution that would have such an analyzer. It is a pity that they are so expensive, because I would buy one.



There is certainly a lot to improve in the current project, but this is the purpose of the prototype. The inductor has too low inductance, current radiators are only suitable for testing also PCBs must be improved to reduce parasitic inductance. I also noticed that the input capacity is too small. With a power output of 750W, the capacitor voltage is only 280V instead of 320V. Of course, such a decrease can be taken into account in the amount of winding of the transformer itself, but the current one works at the limit of the possibilities.
 

Offline mrjoda

  • Regular Contributor
  • *
  • Posts: 80
  • Country: sk
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #68 on: August 01, 2019, 03:41:32 pm »
its comes to my mind... what about ADALM2000 ? it costs around 120€ and one of its features is network analyzer https://www.analog.com/en/design-center/evaluation-hardware-and-software/evaluation-boards-kits/adalm2000.html


i want to buy it soon. Of course it will be like half blind but maybe, just maybe it could help.
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #69 on: August 19, 2019, 06:45:18 am »
Honestly? Somehow I am not convinced that you can do something with this equipment. If you buy, please report, because maybe I am wrong.
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #70 on: March 06, 2020, 02:46:16 pm »
Hi everyone
After long break i would like to resurrect topic.
I changed LC filter to 300uH inductor and total capacity 9000uF plus common mode choke. Also i bought Analog Discovery 2 to measure bode plot.

Quick recap:
Topology - Half Bridge
PWM Controller - SG3525
Mosfet Drivers - HCPL3120
Transistors - IRFP460
Rectifer - 4 x 60CPQ150
Voltage after diodes on max PWM - 43V
Target output voltage 0-30V (regulated)
Feedback OP-AMP - TL3474C
Used compensators - Type 2 , or Type 3
Load = about 1ohm

LC_Filter schematic
944612-0

I have charts bode plot with out compensator, and few charts for compensator type 2 and type 3. Can you evaluate where put zero and pole looking to uncompensated chart. Also i would ask you to evaluate attached charts for compensator type 2 and type 3.

Feedback loop uncompensated Output = 5V
944586-1

Feedback loop uncompensated Output = 12V
944600-2


Type 2 compensator Output = 5V
944578-3

Type 2 compensator Output = 12V
944582-4

I have hope that toogether we will resolve feedback loop compensation problem, and acquired knowledge will be usefull for others designers

« Last Edit: March 06, 2020, 03:38:11 pm by paladyn »
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #71 on: March 06, 2020, 02:50:52 pm »
Type 3 compensator VOUT = 5V
944592-0
Type 3 compensator VOUT = 12V
944596-1

 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #72 on: March 06, 2020, 02:53:47 pm »
I suppose it's too late to recommend an average current mode control...

The CMC in the filter by the way, can't possibly do anything if it's shorted between GND and GND; indeed if the circuit is common ground, the solution is easier because common mode noise can be controlled by layout alone.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline paladynTopic starter

  • Contributor
  • Posts: 42
  • Country: pl
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #73 on: March 06, 2020, 03:01:49 pm »
Sorry, I made mistake during drowing schematic. I will correct atachment above

T3sl4co1l
can you tell more about average current mode control. Why is it better ?
« Last Edit: March 06, 2020, 03:04:09 pm by paladyn »
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17428
  • Country: us
  • DavidH
Re: Half-Bridge 0-30V, 0-20A Feedback loop problem
« Reply #74 on: March 07, 2020, 12:36:04 am »
Sometimes I am amazed that anybody gets a voltage mode variable input or output switching regulator to work.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf