Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Register Log in

4 interleaved boost - how to synchronize?

Status
Not open for further replies.
M

Magnethicc

Member level 2
Joined
Feb 7, 2022
Messages
44
Helped
1
Reputation
2
Reaction score
6
Trophy points
8
Activity points
615
Hello everyone,
I was wondering how would you go about interleaving 4 boost converters with 90 degrees shift between each one?
I was thinking of using the LT6906 to generate clock signal (at 4 times the desired boost frequency) and then connect it to CD4017 counter to produce 4 sync signals which will be sent to UCC28C42 Current Mode controllers.

Does anyone have a better idea on how to achieve this feat, or any recommendations?
 

Sort by date Sort by votes
Hi,

we don´t know the requirements, like frequency (range) duty cycle range, duty cycle resolution....

My first idea: a CPLD or FPGA for digital generation of PWM.

Klaus
 
Upvote 0 Downvote
I have only attempted this with voltage mode controllers, but your outer voltage control loop may face similar issues with current mode control.

When using the sync pin, it usually works by cutting the timing ramp off short, and starting a new timing cycle.
That reduces the final peak amplitude of the timing ramp, and where there are multiple controllers all running off the same error voltage, the duty cycles can diverge considerably up near maximum duty cycle.

I overcame that by not using the sync pin, but allowing the ramp to work normally, and modulating the current being sunk by the usual timing resistor. That shifts the frequency up and down without changing the ramp amplitude.

Its then possible to phase lock the controller to a Johnson counter, as you already plan to do, but use a phase locked loop to phase lock each controller via the ramp timing resistor.
 
Upvote 0 Downvote
Peak current mode control with a common error amp, where the end of cycle is the inductor ramp crossing current demand point, just wants the edges.

A thing to beware - your phase field controls alignment of pulse start but pulse end also wants to be kept away from (after) any sensitive events (like current target compare). If one phase alters the next's operation then you can get subharmonic oscillation ("sloshing" current back and forth on alternate cycles scrubs off energy, costs eff%) or polystable falling edge jitter, which you will get asked about. Oh, yes.

So you might want to base number of phases on some "logic analyzer" sketches of where "things than pollute ground" (and if you have Vin-referred sense in the mix, add "things that pollute Vin") should not occur, across the duty cycle envelope for line/load/etc. Might be that three or five phases maps better to a given center duty cycle.
 
Upvote 0 Downvote
Very good and informative answers guys, thank you.

So as I understand from you guys is that phasing the current mode boosts may present more issues than solve - how about moving to a voltage mode boosts rather then current mode?
 

Upvote 0 Downvote
Some of the linear.com controllers will do a 4 phase interleave for you.
They can also all go into the same output, and each error amplifier output gets tied together, since they are txconductance error amps.

so typically, linear.com do a two way phase interleaved chip...and you just connect two of these together, and it shows you how to get 4way.

Remember also the UCC28070A interleaved boost pfc controller can also be set up as a 4way interleaved split. (obviously you dont want pfc though)
They show you how to do it in the datasheet and app notes.
Thats average current mode control so wont be as noisy as pure current mode control, which Dick_Freebird kindly speaks of.

AYK, damping the gate drive switch ON and OFF (wth series gate resistors) makes the switching transisitons less noisy

Maybe average current mode control could solve the problem that Dick_Freebird spoke of.

I actually did it with a ucc2808 on open loop to give the two 180 deg phase shifted pulse trains...but the pulses were too long so i have to put each through a CDXXX monostable to reduce their timewidth.....that gave me the , two 180 degree phased pulse trains, (but now with nice , short 400ns pulse widths) which got used to sync each ucc28070a, and each ucc28070a does its own pair of interleaved boosts for you anyway....so it thus-ly gave 4way interleaved.
 
Last edited:
Upvote 0 Downvote
UCC28C42 has no built-in synchronization feature, but can be synchronized by a short positive pulse using additional hardware, see datasheet 8.3.6. Selection of suitable controller depends on general product requirements, I think.
 
Upvote 0 Downvote
What voltage and power range are we boosting here ?

Direct boost converter may not be most ideal topology.
How about push pull forward converter with secondary side sitting on top of incoming dc ?
Fewer problems, no right half plane zero issues, or high peak current in the switches.
 
Upvote 0 Downvote
Hi warpspeed - 30V to 100V each phase 1300W.
it is settled around 4 interleaved boosts as a front-end.

So I did some work today i can handle the current mode controllers just which ones are the issue.
Does anyone have any experience using a PFC IC without actually PFCing?
As cupoftea suggested I can use the UCC28070 in interleaved PFC but is it okay to use it as a normal current mode pwm controller without the PFCing?
 
Last edited:

Upvote 0 Downvote
Last edited:
Upvote 0 Downvote
cupoftea said:
Sorry no you cant use ucc28070a....its got too many ingrained bells and whistles which are for pfc.


LTC3862 does up to 12 phase step up...

...the last page of the datasheet shows more multiphase step up solutions.
Click to expand...
according to e2e it is possible we just have to pay attention to the multiplier value and the input voltage variations:
 

Upvote 0 Downvote
yes i see your point, but your vin wont be sinusoidal, and the ucc28070a is made for sinusoidal input really.....you coudl hack it to work , but it wouldnt be worth it.

If you want average current mode, you coudl just do it yourself, with external error amplifiers and some pwm controller....Av current mode is just where the voltage error amplifier sets the reference on the current error amplifier input.
 

Upvote 0 Downvote
Simulation showing expected waveforms. Four interleaved boost converters.
4017 IC sends 4 staggered pulses at 25% duty cycle, non-overlapping.

Pulses are inverted to become 75% duty, overlapping. It turns out this arrangement provides more or less what you want.

As an experiment I made one inductor 50 percent greater value than the others. Yet it conducts about the same Ampere level in simulation.

4 interleaved boost converters 4017 IC 30VDC supply 75 pct duty to 100V 48A.png
 

Upvote 0 Downvote
Magnethicc said:
Is it okay to use it as a normal current mode pwm controller without the PFCing?
Click to expand...
A plain vanilla current mode controller such as a UC3846 should work fine.
Connect up all the internal error amplifiers as voltage followers, and drive all of them from a common external error amplifier.
 

Upvote 0 Downvote
Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top