Hi Luka,
That barrier protects MOSFETs, recitifers & output caps from current stress during switching processes.
Generally choke can be overheated by current through windings, or saturation. The solutions: fat wire, & iron powder core. Because Bsat of iron powder cores much higher than Bsat of same sized ferrite cores. The same core (or may be just a little bit biger) I found in filter of my Antec NEO 550 PC PSU. Due to specs of PSU, it able to deliver 3X18A 12V rails. The total output is 550 watt. And such small inductor dealing with it almost 2 years without any problem. Else as i writed above, inductor & trafo are runing cool under 320W load. To be sure, today I ordered a couple of cores from amidon. So will see...
Due to http://www.murata.com/emc/knowhow/pdfs/te04ea-1/26to28e.pdf - yes
You else can to try. Just may be that you will be need to add some turns to you trafo secondary winding, to compensate inductor looses.
That barrier protects MOSFETs, recitifers & output caps from current stress during switching processes.
Generally choke can be overheated by current through windings, or saturation. The solutions: fat wire, & iron powder core. Because Bsat of iron powder cores much higher than Bsat of same sized ferrite cores. The same core (or may be just a little bit biger) I found in filter of my Antec NEO 550 PC PSU. Due to specs of PSU, it able to deliver 3X18A 12V rails. The total output is 550 watt. And such small inductor dealing with it almost 2 years without any problem. Else as i writed above, inductor & trafo are runing cool under 320W load. To be sure, today I ordered a couple of cores from amidon. So will see...
Due to http://www.murata.com/emc/knowhow/pdfs/te04ea-1/26to28e.pdf - yes
You else can to try. Just may be that you will be need to add some turns to you trafo secondary winding, to compensate inductor looses.
Some comments.
1. D7 and D8 are preventing the FETs to develop the 2*Vin (24V) voltage between Drain and Source. Clamping the Drain voltage excurtion can only be done with a decent snubber and for the very high power units- a regenerative snubber.
2. Building of ANY push-pull stage with capacitive load as you did first is forbidden! Heavy, uncontrolled currents will flow. An inductor between secondary rectifier and the output cap is a MUST.
3. By increasing the operating frequency, one reduces core losses, while moving avay from trafo core saturation (if exists).
4. Usage of voltage mode in push-pulls is dangerous, you may use UCC2808, UC2846,UC2825 for this purpose. As you work with EXTERNAL gate drivers, at a very low frequency (50KHz) any of them will be an excellent choise.
5. Lets calculate some things for real:
What we need? A 60Vdc REGULATED NONISOLATED power supply. The transformer Ae is given ~1cm^2. Vin=12V Vo=60V.
Lets settle for 85% DC. Then you need n=1:6.
Given frequency of 50KHz and on time of 17uS we have:
Np=4+4 Ns=24+24. Bp=5.1Kgauss. Lp=64uH. This means 3.2A magnetizing current. Quite a lot!
Np=6+6 Ns=36+36. Bp=3.4Kgauss. Lp=144uH. Im=1.44A. Better!
Lets go for 100KHz:
Np=5+5 Ns=30+30. Bp=2Kgauss. Lp=100uH. Im=1A. Very good!
What about the secondary inductor?
We have DC=85%.
Po=350W 60V, 5.8A.
If we want a nice current ramp for the current mode to work properly: diL=1.5A Diode voltage=1V.
50KHz: L=0.15*20uS*61V/1.5=122uH. A very heavy inductor!
100KHz: L=61uH. Still heavy but easier.
Also, bu adjusting the turns ratio you can increase the DC to 90% (dont venture beyond 92%), decreasing the inductor. This comes at the expense of abuility to handle lower input voltages.
Add INDIVIDUAL snubbers to the drains, instead of a complete dumper across the primary, whoknows.... you may save a couple of watts.
And watch the output inductor for magnetic saturation! Its deadly! Avoid using junk unknown cores for it. Its too vital for the longevity of the power supply.
1. D7 and D8 are preventing the FETs to develop the 2*Vin (24V) voltage between Drain and Source. Clamping the Drain voltage excurtion can only be done with a decent snubber and for the very high power units- a regenerative snubber.
2. Building of ANY push-pull stage with capacitive load as you did first is forbidden! Heavy, uncontrolled currents will flow. An inductor between secondary rectifier and the output cap is a MUST.
3. By increasing the operating frequency, one reduces core losses, while moving avay from trafo core saturation (if exists).
4. Usage of voltage mode in push-pulls is dangerous, you may use UCC2808, UC2846,UC2825 for this purpose. As you work with EXTERNAL gate drivers, at a very low frequency (50KHz) any of them will be an excellent choise.
5. Lets calculate some things for real:
What we need? A 60Vdc REGULATED NONISOLATED power supply. The transformer Ae is given ~1cm^2. Vin=12V Vo=60V.
Lets settle for 85% DC. Then you need n=1:6.
Given frequency of 50KHz and on time of 17uS we have:
Np=4+4 Ns=24+24. Bp=5.1Kgauss. Lp=64uH. This means 3.2A magnetizing current. Quite a lot!
Np=6+6 Ns=36+36. Bp=3.4Kgauss. Lp=144uH. Im=1.44A. Better!
Lets go for 100KHz:
Np=5+5 Ns=30+30. Bp=2Kgauss. Lp=100uH. Im=1A. Very good!
What about the secondary inductor?
We have DC=85%.
Po=350W 60V, 5.8A.
If we want a nice current ramp for the current mode to work properly: diL=1.5A Diode voltage=1V.
50KHz: L=0.15*20uS*61V/1.5=122uH. A very heavy inductor!
100KHz: L=61uH. Still heavy but easier.
Also, bu adjusting the turns ratio you can increase the DC to 90% (dont venture beyond 92%), decreasing the inductor. This comes at the expense of abuility to handle lower input voltages.
Add INDIVIDUAL snubbers to the drains, instead of a complete dumper across the primary, whoknows.... you may save a couple of watts.
And watch the output inductor for magnetic saturation! Its deadly! Avoid using junk unknown cores for it. Its too vital for the longevity of the power supply.
Hi
I forgot to say I had >15v zeners in series with those diods, so they chopped above 24v that is max on D
Other thing is, you could have >24v zeners from D to GND
I forgot to say I had >15v zeners in series with those diods, so they chopped above 24v that is max on D
Other thing is, you could have >24v zeners from D to GND
Hi Tolik,
output inductor is a must for any buck converter SMPS.
In a symetric power supply using voltage mode feedback you normally monitor only one output (normally the positive) to make the feedback.
The result is that if you use two separate inductors you must draw exacly the same current on the positive and on the negative rail to keep the regulation on both rails. This can work for a full bridge amplifier or at the limit for two stereo channels wired out of phase.
If you load only the negative output (using the positive for feedback) you will see that the voltage on the negative output will not be regulated at all and will sink.
If you load only the positive output (using the positive for feedback) you will see that the voltage on the negative output will increase.
There are methods to regulate both rails, the easiest is to use a common inductor for both rails. This inductor in facts limits the di/dt avoiding huge current spikes in the rectider diodes (and thus also on the other side of the trafo, ie. mosfets) while charging your output capacitor bank.
If you plan for something much more easy to tune use a current mode controller. Another option will be to use synchronous rectifiers instead of diodes on the secondary side but at those low power level it will be quite complicated for nothing.
Normally the yellow cores you find in PC power supply are material 26, but there are also a lot of other made in china PSU with unknown cores with of course unknown parameters. Avoid uning a core without knowing what it is.
ciao
output inductor is a must for any buck converter SMPS.
In a symetric power supply using voltage mode feedback you normally monitor only one output (normally the positive) to make the feedback.
The result is that if you use two separate inductors you must draw exacly the same current on the positive and on the negative rail to keep the regulation on both rails. This can work for a full bridge amplifier or at the limit for two stereo channels wired out of phase.
If you load only the negative output (using the positive for feedback) you will see that the voltage on the negative output will not be regulated at all and will sink.
If you load only the positive output (using the positive for feedback) you will see that the voltage on the negative output will increase.
There are methods to regulate both rails, the easiest is to use a common inductor for both rails. This inductor in facts limits the di/dt avoiding huge current spikes in the rectider diodes (and thus also on the other side of the trafo, ie. mosfets) while charging your output capacitor bank.
If you plan for something much more easy to tune use a current mode controller. Another option will be to use synchronous rectifiers instead of diodes on the secondary side but at those low power level it will be quite complicated for nothing.
Normally the yellow cores you find in PC power supply are material 26, but there are also a lot of other made in china PSU with unknown cores with of course unknown parameters. Avoid uning a core without knowing what it is.
ciao
The differential mode output inductor is the way to go.
It does not reduce efficiency, it's the opposite. What you will notice is just that output voltage becomes secondary winding voltage multiplied by duty cycle and it drops a little more with load, but regulation and some extra secondary turns will solve that.
Output inductor core material should either be iron powder, MPP or gapped ferrite. This inductor should be designed to have an high enough saturation current.
It does not reduce efficiency, it's the opposite. What you will notice is just that output voltage becomes secondary winding voltage multiplied by duty cycle and it drops a little more with load, but regulation and some extra secondary turns will solve that.
Output inductor core material should either be iron powder, MPP or gapped ferrite. This inductor should be designed to have an high enough saturation current.
Hi
differential mode? huh I did mix up two, so then common is for line filtering, or will I be wrong about this too?(Will have to read few things again before giving wrong info out
)
differential mode? huh I did mix up two, so then common is for line filtering, or will I be wrong about this too?(Will have to read few things again before giving wrong info out
)Yes, differential mode,
the current is flowing in two opposite direction on the two rails.
This inductance is not only a filter, it is mandatory to have the possibility to regulate the SMPS. It is used to have a fairly constant current trough it +/- the ripple current.
the current is flowing in two opposite direction on the two rails.
This inductance is not only a filter, it is mandatory to have the possibility to regulate the SMPS. It is used to have a fairly constant current trough it +/- the ripple current.
Hello guys 🙂
Alexsch ,
What is diL, and about witch diode you are talking ?
0.15 - as I understand is total dead time %/100 ?
Are you about to get feedback for Isens. from output inductor ?
Explanations with even simple schematic will be very helpful & much appreciated 😉
Luka,
I think that good idea to use TRANSIL instead DIODE+ZENER pair. Its hyper fast acting , pulse mode zener diode wit ability to dissipate pulses up to 1500W.
Mag,
. First idea was that I connected something wrong way. I repaired the board, & rechecked it more carefully... The result of next try - peeEEK, next pair of STP80NF06 from SAME rail gone to ashtray 
.
I think that such topology, in pair with voltage mode & small transformer that running near Bsat. are good substrate for flux walking & core saturation due to current imbalance, & poor regulation of power rails.
So I using feedback that monitoring both rails at same time. In couple with differential mode inductor it working good & stable 🙂
www.amidoncorp.com , www.ti.com & www.irf.com - FOREVER
Eva,
Happy to see you here . You are short & informative as always
I can`t add nothing to you post - everything that you posted I seen on my prototype board & my scope screen.
Thank you all for awesome advice to my project
Alexsch ,
How you calculated Lp for winding, placed on core with unknown Al ? Can you to share complete formula please?
What is diL, and about witch diode you are talking ?
0.15 - as I understand is total dead time %/100 ?
Are you about to get feedback for Isens. from output inductor ?
Explanations with even simple schematic will be very helpful & much appreciated 😉
Free samples of UC2825 from TI are received,& waiting already 😀
Luka,
I think that good idea to use TRANSIL instead DIODE+ZENER pair. Its hyper fast acting , pulse mode zener diode wit ability to dissipate pulses up to 1500W.
Mag,
I`m avoiding such topology like plaque 😱 . I tried it ( a topology 😉 ). SMPS started good, but when I loaded one of it rails, after about 2..3 seconds the SMPS produced a short, high tone, ricing volume noise like peeEEK, for less than 1 s, & a pair of 80A MOSFET`s was
. First idea was that I connected something wrong way. I repaired the board, & rechecked it more carefully... The result of next try - peeEEK, next pair of STP80NF06 from SAME rail gone to ashtray .I think that such topology, in pair with voltage mode & small transformer that running near Bsat. are good substrate for flux walking & core saturation due to current imbalance, & poor regulation of power rails.
So I using feedback that monitoring both rails at same time. In couple with differential mode inductor it working good & stable 🙂
Yellow made -
www.amidoncorp.com , www.ti.com & www.irf.com - FOREVER
Eva,
Happy to see you here . You are short & informative as always
I can`t add nothing to you post - everything that you posted I seen on my prototype board & my scope screen.
Thank you all for awesome advice to my project
Tolik.
Look carefully on the very site you mentioned. The Al=4040mH/1000T. Thats what they say!
Diode voltage is the Vf of the rectifier diode. In order to get 60VDC on the output, the secondary is to develop 60V+VF+Vlosses to compensate for everything. Since i dont belive you can easily obtain 200V schottkeys, i took the freedom to assume UF rectifier on output.
dIL is what happens when you impose a square wave on the secondary that is coupled to the inductor. Go ahead, read appnotes from TI. There is actually one on a 1MHz UC2825 5V 50W push-pull. May be very usefull for you.
Look carefully on the very site you mentioned. The Al=4040mH/1000T. Thats what they say!
Diode voltage is the Vf of the rectifier diode. In order to get 60VDC on the output, the secondary is to develop 60V+VF+Vlosses to compensate for everything. Since i dont belive you can easily obtain 200V schottkeys, i took the freedom to assume UF rectifier on output.
dIL is what happens when you impose a square wave on the secondary that is coupled to the inductor. Go ahead, read appnotes from TI. There is actually one on a 1MHz UC2825 5V 50W push-pull. May be very usefull for you.
Hi Alexsch ,🙂
Good place where I getting the dedicated MOSFET drivers IC, 40A 250 V schottkeys, ferrite toroids etc. for a long time. Prices & shipping rates - much less than in Tel Aviv or Rishon Its in good case that you will find these parts in our country.😉
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&rd=1&item=160237455785&ssPageName=STRK:MEWN:IT&ih=006
yeah, I forgot about it.
I even have downloaded copy.
Going to repeat it 😀
A lot of pepole suggested me to operate at low Fsw, you telling to go to 50KHz. I started at this value. There was a lot of noise & spikes in output voltage. You can to see the screenshots taken from my scope. So I gone to less Fsw,& results became much better...
How you can to comment it ?
Good place where I getting the dedicated MOSFET drivers IC, 40A 250 V schottkeys, ferrite toroids etc. for a long time. Prices & shipping rates - much less than in Tel Aviv or Rishon Its in good case that you will find these parts in our country.😉
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&rd=1&item=160237455785&ssPageName=STRK:MEWN:IT&ih=006
yeah, I forgot about it.
I even have downloaded copy.Going to repeat it 😀
A lot of pepole suggested me to operate at low Fsw, you telling to go to 50KHz. I started at this value. There was a lot of noise & spikes in output voltage. You can to see the screenshots taken from my scope. So I gone to less Fsw,& results became much better...
How you can to comment it ?
Hi Luka,
Its not like 100% zener. See here -
http://en.wikipedia.org/wiki/Transient_voltage_suppression_diode
Transils can be ordered at http://www.futurlec.com/Voltage_Suppressor.shtml
Its not like 100% zener. See here -
http://en.wikipedia.org/wiki/Transient_voltage_suppression_diode
Transils can be ordered at http://www.futurlec.com/Voltage_Suppressor.shtml
Tolik
I cant! I need to see it with my own eyes. But higher frequency does not mean automatically more noise. Its not that you are going from 50KHz to 10MHz... Go for 100-200KHz. Anyway, generally, a problem does exist somewhere, the change of frequency only displays its sympthomes.
You may visit I.R. Metals. Its in Shoham. There are stockpiles of high-quality MPP cores from mW to kW range. They are better than iron powder. Especially if you have to deal with high current ripple as you need.
Why do you need higher ripple? Thats your next question... isnt so?
Because in current-mode you need the high level of current ramp to reduce noise effects on the loop. And you do need the current-mode when working with push-pull.
Know this:
Iron Powder: Bsat=14-15KG (or 1.4-1.5 Tesla). High dB losses at high frequency. Narrow u range: 10-60. 125 maybe.
MPP: Bsat=7.5KG. Much reduced dB losses at high frequency. Very wide u range: 14-300.
You need MPP in range 60-147. You may contact me by phone on that.
And what is this thing in Rishon? I'm kinda confused... Do you mean RS-components HQ in Israel, or what?
I cant! I need to see it with my own eyes. But higher frequency does not mean automatically more noise. Its not that you are going from 50KHz to 10MHz... Go for 100-200KHz. Anyway, generally, a problem does exist somewhere, the change of frequency only displays its sympthomes.
You may visit I.R. Metals. Its in Shoham. There are stockpiles of high-quality MPP cores from mW to kW range. They are better than iron powder. Especially if you have to deal with high current ripple as you need.
Why do you need higher ripple? Thats your next question... isnt so?
Because in current-mode you need the high level of current ramp to reduce noise effects on the loop. And you do need the current-mode when working with push-pull.
Know this:
Iron Powder: Bsat=14-15KG (or 1.4-1.5 Tesla). High dB losses at high frequency. Narrow u range: 10-60. 125 maybe.
MPP: Bsat=7.5KG. Much reduced dB losses at high frequency. Very wide u range: 14-300.
You need MPP in range 60-147. You may contact me by phone on that.
And what is this thing in Rishon? I'm kinda confused... Do you mean RS-components HQ in Israel, or what?
And what all this Zener stuff anyhow?
Lets do some math!
Personally i do nor belive Tolik can wind the trafo with such perfection as to avoid leakage (Lk).
Lets say we do go for a 100KHz supply.
On toroid with 5+5:30+30 turns expect 1-5% leakage. Lets say Tolik will excell in winding and there will be a 1% leakage, ok?
What do we have?
Each pulse the output current's peak will be about Io+1/2dIL=5.8+0.75A=6.55A
Then Iprimary will be 39A.
If we have a 100uH primary, we have Lk=3uH.
What power it will store?
PLk=0.5*I^2*Lk*Fsx=80W
Thats 40W on each MOSFET.
Got it now?
You want to kill it with a Zener diode? Really?
You need a regenerative damper or go for full bridge. In Full-Bridge you have no leakage worries and no complex primary winding issues.
Lets do some math!
Personally i do nor belive Tolik can wind the trafo with such perfection as to avoid leakage (Lk).
Lets say we do go for a 100KHz supply.
On toroid with 5+5:30+30 turns expect 1-5% leakage. Lets say Tolik will excell in winding and there will be a 1% leakage, ok?
What do we have?
Each pulse the output current's peak will be about Io+1/2dIL=5.8+0.75A=6.55A
Then Iprimary will be 39A.
If we have a 100uH primary, we have Lk=3uH.
What power it will store?
PLk=0.5*I^2*Lk*Fsx=80W
Thats 40W on each MOSFET.
Got it now?
You want to kill it with a Zener diode? Really?
You need a regenerative damper or go for full bridge. In Full-Bridge you have no leakage worries and no complex primary winding issues.
Hi
What every the leakage value is, I didn't had any problems with it and nor did I had problems controlling it, with RC (and diode + zener) to limit turn ON spike
What every the leakage value is, I didn't had any problems with it and nor did I had problems controlling it, with RC (and diode + zener) to limit turn ON spike
- Status
- Not open for further replies.