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.

[SOLVED] Best topology DCDC adjustable output

Status
Not open for further replies.

RandomElectrons

Junior Member level 3
Joined
May 25, 2019
Messages
29
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Location
Knoxville, TN
Activity points
277
Good day to yall and pleasure to join up after reading up on here for a year!

Might as well break in with a general guidance request and go from there. My attempts to design a 400VA adjustable DCDC SMPS has been very troublesome. Two versions have been made but are either unstable or < 75% efficient and heavy. Summery: 25VDC in ( to isolated) 50VDC to 1KV adjustable. 400 watts across the range.

Both attempts are push pull. The 1st using 50KHZ and output multipliers results in instability and high loses in the bridge multiplier and front fets due to high current pulses to regulate between output stages. The 2nd takes the 400hz approach. Stable, but huge I^2R Xformer loses and heavy Xformer and caps.

Oh, I definitely have a more to learn in order to fine tune this, but it seems like a good start is to revisit the basic topology choice again. Any guidance will be heavily considered with a much thanks in advance!
 

Hi,

One very good point to note here is that feedback loop closure is made at worst case. If you are able to determine the worst case conditions for your converter and close he feedback loop at these conditions, then your feedback loop would be stable.

See this http://danyk.cz/reg60v_en.html
 
please provide a schematic of what you built, with control and switching waveforms
and other pertinent information

use as high a frequency as reasonable possible - 50 kHz is good, higher is better - this reduces the size of the magnetics
(its also outside the range of human hearing and is audibly quieter)

50V to 1000V is a large output adjustment range - you might want to use a tapped push pull primary
two transistors at n turn taps and two transistors at N turns. watch voltage capabilities - crossover is tricky

alternatively, you can build multiple secondaries, say, three secondaries, each adjustable between 50 V to 150V
one has regular diodes and is always on, the other two get controlled rectification, or a an on/off
switch in the form of a controlled FET.
connect them in series and turn on what you need to get the desired output range. crossover is tricky

worst case operation is usually at low input line and high output load, or high input line and low output load.
and you want to safely survive an output short and an output open
 
Spanning a 20:1 output range and demanding high efficiency is a difficult problem. As you know, parts don't get chosen based on watts they get chosen based on volts and amps. Fets suited for 1000V/0.5A are very poor at 50V/8A. Caps rated for 1000V are 400X larger than caps rated for 50V (same with inductors and current).

Flybacks have some good characteristics for large voltage spans at constant power. Phase shifted full bridges control easily over a wide output range.

But still difficult.

Relaying transformer taps, secondaries or entire converters in series or parallel to suit the requested voltage is a solution that's 'guaranteed' to work...if you can tolerate the complexity of the switching.
 
push pull - with slope compensated peak current mode control ...?
 

A flyback operating in DCM has a lot of advantages when you need to operate over a wide range. If you're worried about efficiency, using a two-switch flyback will help.
 
400W is a little on the high side for a single flyback - 3 interleaved would be more manageable - all 3 DCM or resonant ( resonant best for the HV rectifying diodes ) - how much current do you need at 50V ?
 

Thanks for the responses everyone!

In quick to answer a few. This is mostly a personal project that is powering HV DC devices . Buying or ordering a custom version is so expensive because, like yall said, very wide range! I feel so close though to doing it myself. The general approach has been center tapped push pull open loop current mode.

50khz.jpg

This has been the general approach. Bench prototypes are more refined then this. This PS can sag under high pulsed loads, so long as it recovers in between to desired voltages and can hold 400 VA avg across the range. Without feedback and crude voltage stages, it has been successful. Utilizing feedback with any config has been a disaster. So I have already made decent progress in sizing passives to get target output performance, just can't get this continuously adjustable and stable!

There are few pointers missed from some of you that are worth further investigation, but wanted to elaborate more before diving in.

Peace!
 

feedback and control will be an issue if you are trying to rectify 500 - 1000 V - RFI will get in - you may need to turn your fets on more slowly - and you will definitely need lots of real snubbing across the diodes 9 and some on the fets ) to reduce RFI noise - and stress on the diodes...!

You have no o/p choke which will cause issues at startup and moderate to heavy loads .... large peak currents in the fets ...!

- - - Updated - - -

also you need slope comp if going above 45% gate drive ( out of 100% gate drive ) - else you will get sub-harmonic oscillation in the control - with a 1,21,2, small then big gate drive and current signal ...

- - - Updated - - -

* change all your 100nF decoupling caps to 1uF, 50V X7R and put 10uF near the 3856 and the gate drive chip
 
Yes, RFI has been a beast, as are reflections back to the primary. By keeping around 35% drive and increasing fet turn on time, alot of nasty RFI is suppressed. Snubbing, decoupling and compensation all need to be addressed as you noted. This first pass is missing plenty.

But back to the most basic question. Do you think this general choice of topology is the best?
 

you are making a bunch of engineering choices
every one has its upside and its downside

there is always another way to do it

some are better than others, but there is no one best

do not let perfect be the enemy of the good
and at some point, its good enough, box it and ship it
 
it's probably one of the easier to get going - be better if you had two 500V wdgs and rectified these and put in series - with fast 800V diodes

with enough snubbing you will be able to quieten it down, still need current mode control for push pull unless you have a few uS dead time guaranteed to allow transformer reset if un-balanced ...
 
given that you are switching 16A on the pri - you need bifilar wound primaries - and very significant snubbing on each pri fet ... e.g. 22nF & 2E2 ( 3W )
 

Yeah, multiple wdgs was entertained along with a 4 stage multiplier. The multiplier has some nasty losses though and many claim they are near impossible to FB properly. So, lets say a single 500V wdg approach was taken. It can be doubled later. Some simple math and spice show some concerning issues FB regulating that from 50V to 500V. At 50V, PWM is like 2 to 4% resulting in huge primary and 2nd currents. It will pretty much be under constant current cutoff limits. 4x primary fets and immaculate snub/FB logistics right?

This was my next step before coming on here and asking around. Most FB examples are at one voltage, and if not, vary 20% at most, not 1000%. It seemed like another dead end.
 

With my first pass multitap PS, the current limits on the primary fets are set at 33A. When starting up the supply using the 700V winding, the duty is around 8 to 10% until 630V. So, if a single 500V winding was chosen, it seems like current limits would always be reached way before pwm duty. Maybe not, correct me. So, in order to maintain 400VA across the 500V, currents at lower voltages are very high.

There is no o/p choke. This has not be entertained yet. Is that the silver bullet to the above conundrum?
 

Of course the currents at 24V in are high, watts out require watts in - so for 400VA, Iave in = 17A, for 24Vin, an o/p choke limits the di/dt and hence the peak currents while the Vo is building up, at 500V 400VA, Io = 800mA - if your load draws high current pulses - this will be seen on the 24V side ...

if you have an 24V to 1kV winding on the Tx then 800mA out = 33.33 amps, turn on current spikes will, of course, be higher than this ...
 

Open current loop mode definitely prevents any failures, but also slows start up and recovery. All of what you said makes sense, just not sure how to properly implement a choke. It seems like an output choke would help lower the primary side currents during startups and high load pulses. Sound like a just statement? A lot of schematics on here and net do not have chokes. The ones that do have it right on output of Xformer, others after rectifier diodes. Given my basic push pull full bridge topology, what would you recommend?
 

" a lot of schematics on here" ? from where? the choke does after the rectifiers - but you will then need snubbers on the rectifiers - esp if the Tx has high leakage inductance ...
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top