How to decide what smps I need for project hybrid amp.

kees54

kees57
I have no test some, however I need a formula to calculate the optocoupler/reference 431 network, I do not see one who works.

Here feedback does not catch, ecept in the begin but very short, did let iot work by increase the frequenty swing. like from 30 to 100 khz.

feedbackproblem.jpg

working feedback, with with a opamp network..

Regulated voltages half bridge.jpg


resonance sinusoidal voltage primairy.

LCCresonant-primairy-sinusoidal.jpg
 

MicrosiM

Administrator
Staff member
For the people behind this forum, I get errors when posting and a massage the site is not secure from google chrome, it is also very very slow and get message also that when I post I leave the site, but posting do work normally.

regards

Hello

SSL certificate installed, moved to a better hosting.

Please advice if you notice any difference
 

kees54

kees57
Did play with the resonance, does good, have ZVS also with no load, stable sim, put deadtime in to get the ZVS to see, it works. have 20 amps max, self limiting happens on 35 amps or so, not important.

you can see in pics that there is ZVS, as i do read, we have to go above resonance for best effects and emi.

But for a ampplifier, regulation is not needed, but use a chip with current protection, and even this seems to be not needed to protect the supply because resonace is self limiting.

nice stuff, high voltages is less.

softswitching between deadtime.jpg

regulated output.jpg

very low ripple.jpg

very low load still ZVS.jpg
 
Last edited:

kees54

kees57
I need some advice about what to use, mosfets or igbt, there are enough there, to much, but find the proper output semiconductor, I think here are experienced men present
who can tell me.

I have doe the feedback with esim, it does work oke, esim was quite close to calculated output.

regards

feedback-TL431.jpg

Opto-feedback.jpg
 

kees54

kees57
I have change the network it is better this way because I do not need a 3th version.

However, with load the feedback is oke, but with open output or 1K it does lost its track and voltage get to max.

someone now why?.

feedback break through.jpg

feedback.jpg

thanks
 

kees54

kees57
Not much respons here?

I have now try the LCC version of smps, these one feedback did work oke with open output or low load, but suddenly now I do sim again and did change nothing
the simulation fail again, it ramps the voltage to max, and the feedback drops to 1 volt cause the frequency of vco to max, who means the voltage should drop and regulate,
strange, maybe ltspice bug or such, because as you se on pic, it did work yesterday.

Someone do now why this happens, maybe the gain in the LLC versions? or maybe just the sim model who is the simple one.

I have start the pcb, do first the high voltage ones for tube, think this is a better way, I make a vco so i can fine tune the resonace when it is not correct place after calculated it..

regardsfeedback.jpg
 

Silvio

Well-known member
Have you start with some hardware? Try to build something and then put what you are simulating in practice. You will see if your simulations is correct or not.
 

MicrosiM

Administrator
Staff member
I also would suggest that you start building your prototype, because simulations are tricky sometimes.
 

kees54

kees57
I also would suggest that you start building your prototype, because simulations are tricky sometimes.

Yes I go start with that, making a pcb, and does simple.

I need however a calculator for a resonace transformer with 4 outputs,, 320, 150mA 2 x120 250mA and 15Volts /10 a 15 amps, did see in simulation that do more secondairy windings the resoance get much lower
who is normal, but q get worse also. making such a transformer out of tests is not a nice pad, need simething better like a calculator. I did see a resonant forward converter does also the job for this.
 

res_smps

Member
I think you can use the formula in Infineon (international rectifier) application note AN-1160 page 13 and put it in Excel
 

kees54

kees57
i don't know if this ncp1396 worksheet can be used for ncp1395
View attachment 7130

THanks res_smps, appreciate it I go try, I think it will work oke.

For the other stuff I did ask about like feedback runout I have discover the reason, the resonant tank gain is the reason, needs different for type of supply who has low idle load like for amps.

Dit change the ratio Lm and Lr and then I can tune to proper feedback correction see pic three. Ohh I have include now the soft start, and then the ramp up does get much better, for the sim
I have limit it somewhat, so you see still some amp surge.

You now guys I have not so much experience with supply's of this kind, but will to learn, and as such also learn to respect the voltages
go make a proper protection, fast leakage switch for grid or so.

regards

feedback.jpg

feedbacl-full power-softstart.jpg

Mg-setup.jpg
 

Silvio

Well-known member
For protection we usually use a voltage comparator sensing the half bridge through a current trafo or an external inductor, then this feeds a 555 timer which eventually operates the shutdown pin on the oscillator chip. The process is very fast but can be tamed a little with a small capacitor so that it will ignore any noise that might false trigger.
 

kees54

kees57
maybe you can ask onsemi, i downloaded it from here https://www.onsemi.com/support/design-resources/tools?rpn=NCP1396
I also got an error

She did respons one time, saying the ncp get out of production, but after a new mail asking for the error in the sheet, she did never answering that again, this kind of
company's need to be sanctioned.

DIY people are a milestone in designs, she do notm understand that, so I go use a other chip from a other company.

Ohh yes a question, I have seen on resoant Lr and Lm, is Lm the leakage induction? because what is then the real transformer induction>? I have some doubt about that,

regards
 

kees54

kees57
Hi All

I decided to use a open loop resonance version with lineair voltage regulators for the tube filament voltages, I have seen in simulation that such a resonant smps keep voltages quite stable even without feedback.

have not done it so I have the voltages however the resonance change when winding the seconday windings, I need for this a calculator who do can calculate this multiple outputs, doing this out of hand is quite a challence I think, I do not now how the Q amd m will change.

For this open loop, what is best the use in Lm/Lr and Q? I have set it up above resonance, it was around 120 Khz, I go lower because of the induction transformer, the 15 volts is just 1 uH, quite low and just one winding or so.

thanks

resonant.jpg
 

kees54

kees57
I have for me to make it more easy but special to learn more made a schematic contains only the resonant parts, transformer and outputs.

Then made a sweep, I did discover when put more outputs on the transformer the resonant frequency did not change much, really almost
nothing, suprise? or maybe not, because q can also change, but in sweep I did not see that, peak stays intact, with degrading q it should
then drop.

To get the real inductions maybe this is a way to do it.

From all the calculators, much are not oke, damaged files.

Regards

findresonance.jpg
 

kees54

kees57
Hi All

Do somebody now how to calculate the transformer itselfs? I have done this with a excel sheet who is used for a sudden chip I not use But I think it will work
also because resonance is resonance right? I have imput that in a simulator and it was quite good and close. I see the resulting induction was quite low
that is a result of how I does make the Q factor of the system with the k.

Here is a test winded transformer, I did make a slotted bobine for it, measurement did let see a leak of 40 uH on 120uH primairy coil inductance.

Because I do need more voltages I do not know what happens, I have a old ferrite from a pc supply, I think it is better buy new ones, these are not so expenseive
anymore. For the resonant coil, do I need a version with the same spec,s? because to prevent things co wrong like saturation.

I have received now some controllers for the system, one has a direct mosfet drive capabillity without a drive transformer, for bigger power I like a transformer more.

So I can experiment.

DSCN3901.JPG

see pic, Transformer is winded with wire, not the real one because I need to use strains of wire, this is just a test,

thanks.
 
Top