Where are all the parts?

[PeteS in CA]

Re: Where are all the parts?

Glad to be helpful! Sometimes a seemingly simple circuit has a lot going on. I may do a design description post about a "simple" self-oscillating discontinuous flyback.

[momaka]

Re: Where are all the parts?

The forward converter topology is not especially new. It was around and in use when I got into power supplies bock in 1980. In a forward converter, during the switch "on" time, one diode delivers current to the load, also charging the inductor. During the switch "off" time, the inductor is discharging, supplying current to the load, with the other diode completing the current path to the other end of the inductor. IOW, during the "off" time, the inductor is a current source, and the freewheeling diode's function is to complete the current path. The "off" time is never long enough for the inductor to fully discharge. The output capacitors also charge during the "on" time and discharge some to supply current to the load during the "off" time.

Cool
A few years back I was quite puzzled about why there would be diodes going between GND trace and each rail. Never really came back to that question. And now I don't have to.
Thanks for the explanation .

[PeteS in CA]

Re: Where are all the parts?

Also, the output current has nothing to do with the duty cycle. The maximum output current is the diode's rated current, not some fraction of that like they keep saying.

In theory, that is correct. The duty cycle, theoretically, is the product of the input voltage and the transformer ratio. With real world components, forward drops in the switch device(s) and rectifiers and voltage drops in the transformer windings, inductor windings and PCB traces all increase with increased load current, so the regulator increases the duty cycle to compensate.

Hardware secrets also say that on newer designs, only one of the two diodes is used for rectification, while the other is used for discharging the coils (freewheeling).

The forward converter topology is not especially new. It was around and in use when I got into power supplies bock in 1980. In a forward converter, during the switch "on" time, one diode delivers current to the load, also charging the inductor. During the switch "off" time, the inductor is discharging, supplying current to the load, with the other diode completing the current path to the other end of the inductor. IOW, during the "off" time, the inductor is a current source, and the freewheeling diode's function is to complete the current path. The "off" time is never long enough for the inductor to fully discharge. The output capacitors also charge during the "on" time and discharge some to supply current to the load during the "off" time.

[Th3_uN1Qu3]

Re: Where are all the parts?

Hardware secrets also say that on newer designs, only one of the two diodes is used for rectification, while the other is used for discharging the coils (freewheeling). If that is the case, then on a forward design, the theoretical maximum is only half what the rectifier is rated for.

On two transistor forward this is true, it's half wave rectification, and the freewheeling diode only takes 1/3 of the load. Its job is to prevent reverse voltage spikes created by the inductor, and you will see that in a half-wave rectifier powered by normal sinewave mains, this diode is not needed.

So we can conclude that in a single-ended SMPS, the output current limit is given by the forward rectifier, and the freewheeling rectifier can be of lower current rating. That doesn't make HWS's math less misleading - if the duty cycle is different from 30%, all their calculations fall on their ass.

[c_hegge]

Re: Where are all the parts?

Also, the output current has nothing to do with the duty cycle. The maximum output current is the diode's rated current, not some fraction of that like they keep saying.

I know

Hardware secrets also say that on newer designs, only one of the two diodes is used for rectification, while the other is used for discharging the coils (freewheeling). If that is the case, then on a forward design, the theoretical maximum is only half what the rectifier is rated for.

[Th3_uN1Qu3]

Re: Where are all the parts?

It's interesting that HardwareSecrets always say the duty cycle is typically fixed at 30% or so. Obviously, they are wrong if the duty cycle is in fact varied.

They're not saying that. They've just settled for a typical value so they can run their math and post some theoretical numbers.

The truth is that in two transistor forward the maximum duty cycle could be 30%, could be 40%, could be 50%, could be over 50% depending on the design. Unlike half bridge, 2 transistor forward is single ended, so the duty cycle can go beyond 50%. This brings additional complications so it isn't widely used, but it can be done. Usually, the controllers used have max duty cycle limited to 50% tho.

Also, the output current has nothing to do with the duty cycle. The maximum output current is the diode's rated current, not some fraction of that like they keep saying.

[PeteS in CA]

Re: Where are all the parts?

The main reason for the duty cycle varying is line voltage changes. Load changes cause the duty cycle to change a little to compensate for the small voltage drop changes in the switch devices, rectifiers, windings in the transformer and inductor, and PCB traces.

[PeteS in CA]

Re: Where are all the parts?

The duty cycle is varied to compensate for line and load changes. That's how SMPS regulation is achieved.

[Shocker]

Re: Where are all the parts?

Working on PSUs is fun

I added two MOVs and upgraded the +5VSB diode from a 2A to a 5A.

I can use a 1500uF Chemi-con KZH from DPS-350AB-7 (the same PSU I got the MOVs from) for the +12V. I don't have caps for the other outputs.

Also: 50th post!!!!!

[c_hegge]

Re: Where are all the parts?

In SMPS, regulation is done by varying the pulse width, the height (ie peak voltage) stay the same.

So that's essentially varying the duty cycle.

It's interesting that HardwareSecrets always say the duty cycle is typically fixed at 30% or so. Obviously, they are wrong if the duty cycle is in fact varied.

[Agent24]

Re: Where are all the parts?

Thanks guys, that makes sense now

blocking out what someone writes just because you don't like them doesn't really make sense to me.

I have never known anyone to always say things I never agreed with or which weren't useful.

Probably should explain that ^ better - anyone could interpret that as a statement "I don't like you but I'll listen to you anyway" - I have no issue with anyone on this forum, so don't get me wrong and think I'm secretly hating anyone here!

[PeteS in CA]

Re: Where are all the parts?

In SMPS, regulation is done by varying the pulse width, the height (ie peak voltage) stay the same.

IOW, the peak voltage of the square wave the rectifier sees (ignoring ringing and noise) is much higher than 12V, and varies with the AC line voltage. It could be triple or quadruple the O/P voltage, depending on the supply's designed minimum AC voltage and how the transformer was designed. Then there is ringing, which can go higher still. A conservative rating choice would be 100V or higher (and I am aware that the forward voltage will be higher, especially if an ultrafast rectifier is used). 60V will probably work OK, while 40V or 45V would be pushing your luck pretty hard.

[Th3_uN1Qu3]

Re: Where are all the parts?

On the other hand, I really can't see what's wrong with a 40v device on a 12v rail either. Surely 12v is far within the limits of 40v?

It's due to the peak voltage of the transformer. A computer PSU is a regulated SMPS, and to keep regulation it needs to have some headroom built-in. In SMPS, regulation is done by varying the pulse width, the height (ie peak voltage) stay the same. If you disable the feedback loop of a PC PSU and let the controller run at max duty cycle you'll get something like 25v out, but you'll also get a ton of noise over it as there is no more regulation and the noise isn't smoothed out. This shows that there is headroom available at the transformer for double the output voltage or more. However, as load increases this becomes significantly less.

Needless to say, don't try this with the original caps in as they'll pop in your face.

Above those 25v there's also the voltage dropped across the output inductor, so yes, in some cases, the peak voltage available at the transformer can be higher than 45v. There's only one method that will show you what diode you need and it doesn't have to do with any rule of thumb as there are none. Scope it.

[goodpsusearch]

Re: Where are all the parts?

Actually, I remember at that psu that when I used the 30A schottky at 12V, the secondary heatsinks got very hot when I tested the power supply without load and that should have got me suspicious.

[Agent24]

Re: Where are all the parts?

I never saw the point of ignore lists - well I can see the point - but blocking out what someone writes just because you don't like them doesn't really make sense to me.

I have never known anyone to always say things I never agreed with or which weren't useful.

In saying that I am not trying to be judgemental about anyone's potential blocklist decisions, just voicing an opinion!


On the other hand, I really can't see what's wrong with a 40v device on a 12v rail either. Surely 12v is far within the limits of 40v?

[Shocker]

Re: Where are all the parts?

Oh and I know the ignore list. I already put bigcape on it. But I'm not ignoring you...yet.

[Shocker]

Re: Where are all the parts?

If you want to be rigurous, scope the transformer at maximum mains voltage and zero load and see the peak voltage. Then you will know exactly what the voltage rating of the diode should be.

As for me being banned from this forum, let's let the administration decide, shall we? The forum has an ignore list you can use if you don't like to read my opinion.

It's a case where "barely enough" is not enough.

BTW, I'm working on the L&C. PBL405 bridge in place of the 2A05 diodes, the 560uF Panny UPs, SBL2040CT for +3.3V, SBL3040PT for +5V, S20C60C for +12V, added pi coils (from three different PSUs) to all six outputs. The coil I chose for the +5V will make inserting the caps a bit of a squeeze. Don't even ask about jamming 12.5mm caps into there. The input filtering (excluding MOVs) was already in place.

[Th3_uN1Qu3]

Re: Where are all the parts?

If you want to be rigurous, scope the transformer at maximum mains voltage and zero load and see the peak voltage. Then you will know exactly what the voltage rating of the diode should be.

As for me being banned from this forum, let's let the administration decide, shall we? The forum has an ignore list you can use if you don't like to read my opinion.

[Shocker]

Re: Where are all the parts?

At worst it'll create a little extra heat.

Actually THIS can happen:

https://www.badcaps.net/forum/showth...?t=8171&page=7

All in a matter of 40 seconds.

If I do an L&C rebuild, I'll use at least a 60V rectifier. I have an S20C60C lying around.

You mentioned replacing KZG with KZE elsewhere. They aren't even somewhat close, and I wouldn't dare try it.

If you do it anyway, so be it. But if you tell people to do it, you should get banned from the forum.

[Scenic]

Re: Where are all the parts?

Nice! Loving the Windows 3.1 styled clock too...

Attached it just for the heck of it.
If doubleclicking the .gadget file (in Vista/7) doesn't do anything, type "shell:gadgets" (without the quotes) into the start menu search bar thing and hit enter. That opens the folder where all gadgets are stored. Unpack it there and it'll then show up in the Sidebar gadgets thing in windows.

[sorry for OT]