I now have 600uH and 85uH and Cr is 10nf.
I have tried to calculate the timing components but am getting confused where the numbers come from.
I get:
Fr1 to be 172KHz
Fmin to be 60800.
Fmax I don't know how to calculate although the app note says 150khz which is below 172Khz so cant be right.
I worked out Rt to be 1k2. The app note gets 18k for Rt with 60800Hz so I am clearly wrong.
For some reason with Rmax I get a negative result !
I got a grade A at GCSE maths so my maths isn't that bad.
I think the problem is deciding what the input values are in the first place.
The app note is pretty poor.
I have tried to calculate the timing components but am getting confused where the numbers come from.
I get:
Fr1 to be 172KHz
Fmin to be 60800.
Fmax I don't know how to calculate although the app note says 150khz which is below 172Khz so cant be right.
I worked out Rt to be 1k2. The app note gets 18k for Rt with 60800Hz so I am clearly wrong.
For some reason with Rmax I get a negative result !
I got a grade A at GCSE maths so my maths isn't that bad.
I think the problem is deciding what the input values are in the first place.
The app note is pretty poor.
Last edited:
Of course your calculations are correct, it is the input that is incorrect.
Have you recalculated the values in the app note so you get the same number ?
Some app notes are plain wrong because of typos etc.
About your startup problem, see page 10-11 here:
http://www.onsemi.com/pub_link/Collateral/AND8311-D.PDF
Have you recalculated the values in the app note so you get the same number ?
Some app notes are plain wrong because of typos etc.
About your startup problem, see page 10-11 here:
http://www.onsemi.com/pub_link/Collateral/AND8311-D.PDF
I don't get 18k as per the app note circuit.
If I cant calculate from their numbers I will never get my own right.
I have asked IRF for some help.
Its not obvious where some of the values come from or even what they are.
They talk about the resonant frequency but there are two to choose from.
I don't know where Fr1 comes from, it could be 60800 or 172000.
I looked up a simpler SMPS half bridge IC app note and the transformer was very similar to my own, 600uH + 85uH with 33 turns primary and 9 secondary.
I went through the app note again and I am getting negative values for the Rmax due to using too high a frequency. If I change my Cr to 100nF then the numbers turn positive.
You can only use values from the application note if all your values are the same.
"If you change one you must change other,
the values are dependent on each other." (poem from the XXI. century)
Also the minimal switching freq and the lower resonant freq is TWO DIFFERENT THING. The minimal switching freq MUST BE LARGER than the lower resonent freq to avoid ZCS switching. Its right in the beginning of the application note (page 4, figure 4).
In the spreadsheet there are multiple sheets. See the attached picture for help.
Attachments
Last edited:
You can only get negative values for the timing resistors if you timing capacitor (not the resonant capacitor) is too big. The timing capacitor determines the dead-time (and also the switching freq).
This is bad in an aspect that these components linked to each other, however it also guarantees that your dead time is not big relative to your switching freq.
(In the applicaton note is said that that first harmonic approximation is used, and dead-time is neglected in the calculation of the duty cycle. Dead-time is only used for ZVS parameters calculation. So if you choose too much dead-time the calculation approximations are not valid)
Calm down Nigel, I am not saying you suck at maths, I only say you did those calculations bad.
It's interesting, I specially found this appnote really good, and I was able to understand the operation, although english is not my mothern language.
However to help you I will try to sum the operation of the LLC converter in the followings:
Look at page 3 figure 3, and page 4 figure 4.
The Lm, Lr and Cr dictate two resonant freqy. (Lm+Lr) and Cr dictates the lower (the top of the red curve at figure 4). Lr and Cr dictates the higher (the top of the cyan curve on figure 4).
In order to have the capability of ZVS switching, the network must be inductuve, not capacitive. This is true for two areas:
- Area (1), freq s over the higher (MAIN) resonant freq (Fr1)
- Area (2): this are is load dependent. At every load value you can combine the Lr Lm to an acutal inductance. If the switching freq is higher then this, then you are in inductive mode
(inductive mode let you have ZVS switching, but does not guarantee it)
Lets look at two example:
Lets assume Fr=100kHz
- We have low load, and our input voltage is high (330V). Then we need a gain < 1. We are on now the red curve. To have a gain < 1, we have to use a switching freq greater than Fr1.
- We have high load, and our input voltage is low (300V, the priamry votlage sinks to this at load conditions). Then we need a gain >1. Wr are on now the pink curve. To have a gain >1 we need to have a switching freq about 70kHz.
The cyan curve stand for short circuit. At short circuit the magnetizing inductance is totally shunted. The red curve stands for low load, or idle. At idle the leakage and magnetizign inductance is totally summed.
To increase the gain the converter has to decrease switching frequency, and to decrease the gain the converter has to incresease the switching frequency. (only in the range of inductibe mode)
The frequency where the gain curve has a maximum is dependent on load. The available gain decreases as load increases. So we have look at the maximal projected load (curve pink). We have to calculate the frequency where it has its maximum, and SET IT AS miniminal switching frequency.
If you switch at a lower frequency you will get to the left side of this curve, and will get to capacitive mode, and loose the ZVS. Also the converter feedback will loose controll.
Last edited:
I worked through the spreadsheet and input all my values:
Lm 800uH
Lr 85uH
Cr 100nf
k 7
Fmin 19229
fmax 110000
Fr1 54589
I used 66k Rmin, 15k Rmax.
It still overloads and stops.
I powered up the circuit on 30 volts dc and I get the correct low frequency.
When I short out the opto coupler I get the correct high frequency.
Just to eliminate power up timer problems I used 47uF to stretch the power up delay.
Lm 800uH
Lr 85uH
Cr 100nf
k 7
Fmin 19229
fmax 110000
Fr1 54589
I used 66k Rmin, 15k Rmax.
It still overloads and stops.
I powered up the circuit on 30 volts dc and I get the correct low frequency.
When I short out the opto coupler I get the correct high frequency.
Just to eliminate power up timer problems I used 47uF to stretch the power up delay.
I looked at the power up pulse train and the 27951 is resetting well before the power up sequence ends.
I checked my transformer in case I had wound the secondaries out of phase but it is ok on the scope.
But what is your soft start frequency? What is the soft start frequency setting resistor (Rss)? Without a resistor a soft start capacitor (Css) does nothing!
The circuit wont power up off mains but it does run on 90 volts DC.
I can see the high frequency power up sequence working then the frequency slowly drops to 25KHz. It is going to the lowest frequency as the 90 volts doesn't give enough voltage to make the opto-coupler output go low.
My problem is I don't have an isolation transformer to look at the primary side of the circuit.
I can see the high frequency power up sequence working then the frequency slowly drops to 25KHz. It is going to the lowest frequency as the 90 volts doesn't give enough voltage to make the opto-coupler output go low.
My problem is I don't have an isolation transformer to look at the primary side of the circuit.
If you have an Lm=800uH and a Lr=85uH, how can you have a K=7 ?
Your K is 9,4
(The spreadsheet uses function to calculate values. Cells with input are marked yellow. Don't write into cells with function)
From these values (k=9,4) i get a fmin of 33kHz. (19kHz will lead you to capacitive mode)
The primary peak current is 2,70A. This means you need a FET with Rdson < 0,74 Ohm, which is true for you. The IRS27951 starts with swtiching on the lower FET, and puts a bigger impulse out first.
What is your timing capacitor? 270pF?
Now its getting interesting...
If your Cr Lr, and freqs right, it should start up without a problem now.
With 600uH - 85uH - 100nF combination the primary peak current is 3,6A. But it would also work with your 0,4Ohm FET (its on the limit but should).
Do you power with no load at output?
When you power up does the output voltage increase to some level?
I have 660r across 45 to -45 volt lines.
I currently also just have 100uf per rail for smoothing.
I get a very brief but small voltage on the output.
The 27951 is resetting on the power up high frequency.
Its not getting a chance to get to the normal working frequency.
If I run off 90 volts I can see the power up frequency is fine and then it slowly settles down to about 21000Hz.
The IRS27951 first low output impulse has a double length (datasheet page 15). Do you use one 100nF capacitor, or 2x47nF split capacitor?
If you use a single, then at startup it has no voltage on it. And this (and the IRS27951 longer first impulse) causes a current transient, which triggers the OC of the IRS27951.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.