Antec SL-400 5vsb Mod with NCP1200

[PeteS in CA]

Re: Antec SL-400 5vsb Mod with NCP1200

Your scope display doesn't indicate, in the first image, that the BW is limited to 20MHz. I'm not quibbling, I'm just used to Tek scopes, which put "BWL" next to any channel which is set to 20MHz BW. But I certainly understand that Tek scopes are not inexpensive ($5K-$6K for a 100MHz, 4-channel TDS-3000C, IIRC).

Unless you are measuring ripple/noise (which is usually specified at 20MHz BW) or are working in a distractingly noisy environment, I would suggest using full BW all the time.

I'm guessing that if you did 10 or 20 single-sweep captures that you would find the amplitude of that first spike varying from under 40V peak to about 50V peak. So your schottky diodes are probably getting taken out by cumulative damage rather than one big blast. The real world of electronic circuits has all kinds of interesting limitations and parasitic effects. More basic electronics classes tend to ignore them, for simplicity. But in the real world they can bite you. But just think of it as increasing your understanding of how things work. And as opportunities in the future to help others.

[ben7]

Re: Antec SL-400 5vsb Mod with NCP1200

Your welcome!

[LDSisHere]

Re: Antec SL-400 5vsb Mod with NCP1200

Ben7 and PeteS_in_CA it looks like you two were on the money with your prediction. I am attaching three graphs, (hopefully they will have thumbnails this time) the first one is with the bandwidth limited to 20Mhz, the next one is the same as the first but using the full bandwidth of my scope and the third one is with a narrower time band.

With the limited bandwidth everything looks like it is good, it is only with the full band width that you can see the voltage reaches a magnitude over 40V (-40V) which goes a long way toward explaining why the diodes are dying. I think I will be looking for some 60V diodes for future testing.

Thanks for pointing me in the correct direction for the source of this problem.

[LDSisHere]

Re: Antec SL-400 5vsb Mod with NCP1200

I don't have a scope or even used one, but see

EEVblog #279 – How NOT To Blow Up Your Oscilloscope!

http://www.eevblog.com/2012/05/18/ee...-oscilloscope/

I had watched that video before my scope arrived. I thought it was very informative and explained some things I had questions about. It is also why I am wanting to be very careful about how and to what I connect the scope. I do not want to release any smoke from my new scope nor myself. At this point I am just taking baby steps in the little bit of time I have had to work with it.

[LDSisHere]

Re: Antec SL-400 5vsb Mod with NCP1200

My only "contribution" to this thread is to point out that .BMP files are 1.3MB whereas a JPG version of it is only 66KB and I can't see the .BMP as a thumbnail.

I was very surprised there was no thumbnails of the .bmp files after I made the post. I do not see this format too often these days but it is what the scope dumped to my thumb drive and I really did not give them a second thought before posting. I will try to figure out a "better way" of posting them in the future as I like the thumbnails myself. Using this scope is going to involve a big learning curve on my part and this is just another step on that path.

[retiredcaps]

Re: Antec SL-400 5vsb Mod with NCP1200

I am trying to be careful not to do anything stupid and blow it up as I know that is a distinct possibility when playing with power supplies.

I don't have a scope or even used one, but see

EEVblog #279 – How NOT To Blow Up Your Oscilloscope!

http://www.eevblog.com/2012/05/18/ee...-oscilloscope/

[tom66]

Re: Antec SL-400 5vsb Mod with NCP1200

The Owon scope does not natively output PNG.

PNG is for screenshots, graphic line art, etc.
JPEG is for photos.

[retiredcaps]

Re: Antec SL-400 5vsb Mod with NCP1200

I thought you might be onto something here then I thought why not use my new oscilloscope to actually "see" what the circuit was doing.

My only "contribution" to this thread is to point out that .BMP files are 1.3MB whereas a JPG version of it is only 66KB and I can't see the .BMP as a thumbnail.

[LDSisHere]

Re: Antec SL-400 5vsb Mod with NCP1200

Your only looking at the forward voltage drop of the diode. Instead, look at the ac waveform from the transformer secondary.

Another thing is, your diode might be overheating.

I will take a look at the secondary AC side as soon as I get a chance. I think by attaching the TO-220 package directly to the heat sink I will have eliminated any heat problems I was getting from the cylindrical packages of the other diodes I was using.

I had similar thoughts as ben. Maybe use a 50V or 60V rectifier. The V(f) will be a little higher, but it will tolerate spikes. Unless your scope and probe are 100MHz or better (and the probe equalization is properly adjusted) the spikes might be there, but the scope can't "see" it. The unpleasant part is that the 11pF-22pF capacitance of the probe might also be enough to diminish spikes. OTOH, even a short probe ground lead could pick up noise. BTW, another frustrating thing, different types and brands of rectifiers and MOSFETs will have different parasitics (i.e. ringing).

My scope is a Owon SDS7102 100Mhz which I have only had for 1 week so I am still trying to figure it out. I am trying to be careful not to do anything stupid and blow it up as I know that is a distinct possibility when playing with power supplies. I have an isolation transformer on its' way to me which I should have by Wednesday.

What you're doing is finding the things limiting the circuit's maximum current performance. The output rectifier, switch MOSFET and current sense resistor are three factors. Several others are the MOSFET gate drive (does the turn-on and turn-off get sloppy at higher currents), the transformer primary wire, the transformer secondary wire, and the PCB trace width and thickness. And if there is a separate O/P inductor, the gauge of its wire.

Yes, these things are what I am curious about finding out about and what effects modifying them have on the rest of the circuit. The trick I suppose is to figure out where the cause is located that is creating the effect.

One extra note regarding transformer and inductor wire. With AC, a characteristic called "skin effect" limits the current carrying capacity of wire. As the frequency goes up, current increasingly concentrates in the outer part of the wire. Skin effect is relevant as low as 400Hz (the frequency used for power for US military electronics). At 30KHz or 200KHz, that AWG #18 wire good for 10A DC may only be good for 3A-5A. There are a couple of ways to work around this, both of which increase the amount of "skin". One is to use copper foil for higher current secondary windings. Basically, foil is all "skin". The other is to use Litz wire, which is strands of smaller gauge magnet wires (i.e. each strand is insulated from adjacent strands) twisted together. If you're thinking that neither option is cheap, well, you're right. But using AWG #12 or #14 wire for a 10A O/P isn't cheap, is bulky, and is not fun to wind (try winding AWG #10 on a 1.57" OD toroid and you'll experience what I mean - BTDTGTTS, a drawerful in fact) or to terminate to transformer bobbin pins.

You have provided me with a lot of very good useful information here that I very much appreciate. I could follow most of this but then I get lost. I have an understanding of the skin effect but not a good one. I will have to do some reading on this subject.

[PeteS in CA]

Re: Antec SL-400 5vsb Mod with NCP1200

I had similar thoughts as ben. Maybe use a 50V or 60V rectifier. The V(f) will be a little higher, but it will tolerate spikes. Unless your scope and probe are 100MHz or better (and the probe equalization is properly adjusted) the spikes might be there, but the scope can't "see" it. The unpleasant part is that the 11pF-22pF capacitance of the probe might also be enough to diminish spikes. OTOH, even a short probe ground lead could pick up noise. BTW, another frustrating thing, different types and brands of rectifiers and MOSFETs will have different parasitics (i.e. ringing).

What you're doing is finding the things limiting the circuit's maximum current performance. The output rectifier, switch MOSFET and current sense resistor are three factors. Several others are the MOSFET gate drive (does the turn-on and turn-off get sloppy at higher currents), the transformer primary wire, the transformer secondary wire, and the PCB trace width and thickness. And if there is a separate O/P inductor, the gauge of its wire.

One extra note regarding transformer and inductor wire. With AC, a characteristic called "skin effect" limits the current carrying capacity of wire. As the frequency goes up, current increasingly concentrates in the outer part of the wire. Skin effect is relevant as low as 400Hz (the frequency used for power for US military electronics). At 30KHz or 200KHz, that AWG #18 wire good for 10A DC may only be good for 3A-5A. There are a couple of ways to work around this, both of which increase the amount of "skin". One is to use copper foil for higher current secondary windings. Basically, foil is all "skin". The other is to use Litz wire, which is strands of smaller gauge magnet wires (i.e. each strand is insulated from adjacent strands) twisted together. If you're thinking that neither option is cheap, well, you're right. But using AWG #12 or #14 wire for a 10A O/P isn't cheap, is bulky, and is not fun to wind (try winding AWG #10 on a 1.57" OD toroid and you'll experience what I mean - BTDTGTTS, a drawerful in fact) or to terminate to transformer bobbin pins.

[ben7]

Re: Antec SL-400 5vsb Mod with NCP1200

I thought you might be onto something here then I thought why not use my new oscilloscope to actually "see" what the circuit was doing. I have been want to use it for something practical and this was my first chance.

After putting the SBL1040-E3/45 in the supply, I attached my scope to the output side of the diode. It does not look like the voltage is spiking above 8V and these images were captured with the load at 1.1 ohms. I of course have no idea what the wave form means one way or another as far as if the circuit is operating normally or not. I understand the base voltage is approximately 5V that swings a much as 3V above and below the base line during the periods of oscillation. I did not run it very long at this load as I did not have time.

Your only looking at the forward voltage drop of the diode. Instead, look at the ac waveform from the transformer secondary.

Another thing is, your diode might be overheating.

[LDSisHere]

Re: Antec SL-400 5vsb Mod with NCP1200

My guess is that larger switching spikes from the larger currents involved is killing your diodes. Try either a higher reverse voltage diode, or a snubber across the diode.

I thought you might be onto something here then I thought why not use my new oscilloscope to actually "see" what the circuit was doing. I have been want to use it for something practical and this was my first chance.

After putting the SBL1040-E3/45 in the supply, I attached my scope to the output side of the diode. It does not look like the voltage is spiking above 8V and these images were captured with the load at 1.1 ohms. I of course have no idea what the wave form means one way or another as far as if the circuit is operating normally or not. I understand the base voltage is approximately 5V that swings a much as 3V above and below the base line during the periods of oscillation. I did not run it very long at this load as I did not have time.

[ben7]

Re: Antec SL-400 5vsb Mod with NCP1200

My guess is that larger switching spikes from the larger currents involved is killing your diodes. Try either a higher reverse voltage diode, or a snubber across the diode.

[LDSisHere]

Re: Antec SL-400 5vsb Mod with NCP1200

Well I put a VS-90SQ040 (9A 40V) diode in place of the SB340 that had shorted in my previous test. Last night I got a chance to run it so I started where I left off which was a 1.1 ohm load on the 5vsb. After about 45 minutes of it running fine I noticed that the voltage had suddenly dropped to basically zero.

Today I decided to see where the problem was this time but I started my checks with the diode I replaced. Much to my surprise this new 9A diode was shorted just like the one it had replaced.

I have a SBL1040-E3/45 to try but I will have so squeeze in a heat sink with it if it has any hope of lasting longer than the previous diode. I thought for sure something else would pop before the 9A diode that I used would be a problem. My goal is to get this supply to run 5A on the 5vsb for one hour and I am almost there, I may just need to point a fan at it when it is running.

[LDSisHere]

Re: Antec SL-400 5vsb Mod with NCP1200

Everell I will have a look at those diodes before I place my order.

I used the SB340 I had laying on my table to replace the shorted diode. The 5vsb came back to life and it seems the ringing went away. This diode may have been damaged in my earlier experiments before finally failing completely. I did not try any heavy loads this time, those will be saved until I get a higher rated diode.

[everell]

Re: Antec SL-400 5vsb Mod with NCP1200

I replaced my SB340 with a SB540 diode. So far it has not blown. Should it fail, I also bought from Digikey the SB1040-E3 in TO220 case. If I install it, I want to install it to the secondary heat sink.

The SB340 is directly beside and touching the C36 filter capacitor which usually is bulging. Judging by the amount of heat from the diode in the experiments, I am not surprised that the capacitor doesn't like the location.

When my SB340 diode failed, it was too hot to touch!!!!!

[LDSisHere]

Re: Antec SL-400 5vsb Mod with NCP1200

You may see these experiments as an exercise in futility. I see it as how much juice can I get from this lemon or maybe how to have an indoor fireworks display. It is also a good learning exercise. I was able to trace out and modify the two transistor 5vsb and then discover where the other weak spots in the circuit are located. So after I get a better diode I will then see what pops next. If I melt down the transformer, I may try my hand at winding a replacement or just move on to another supply which I will torture back to the death from which I previously revived it like I have done with this one. Also one thing this experiment does prove is that you can get more from a 5vsb circuit using an IC to drive a FET than you can using a one chip PWM like the DM311 or Viper22A, etc. and the heat is better dissipated because the 5vsb FET is already attached to a very large heat sink.

The squealing/ringing was occurring under no load and would actually get quieter as the load increased. It seems to be directly related to the current sense resistor that goes between the 5vsb FET and ground. As the value of this resistor value gets lower the noise gets louder, as the load increases it gets quieter.

[Toasty]

Re: Antec SL-400 5vsb Mod with NCP1200



You're overloading the 5vsb, so what do you expect?

The SB340 is rated for 3A and you're pushing 3.8A through it...?

The PSU is rated for 2A on 5vsb.

Lowest resistance (for *max* current) would be 2.5Ω. Using a "rule of thumb" for (continuous) static load testing, I would use nothing lower than 4Ω. That's a 60% load.


No wonder it's ringing/squealing. You would too if overloaded by 90%...

Unless you're redesigning the trafo and other components, this is an exercise in futility.

T

[LDSisHere]

Re: Antec SL-400 5vsb Mod with NCP1200

Can't wait to see the carnage LOL

I hate to disappoint you but the carnage, if can call it that, was minimal at best. The diode was shorted and it was a SB340 which is the same as what Everell had fail in his Bestec experiment. I am actually surprised that it held up as well as it did considering the load that was going through it and what it is rated to accept. I have a good SB340 that I may throw in it just to verify everything else is working until I get a better replacement.

Following Petes_in_CA's advice in Everell's Bestec thread, I am looking at ordering some 9 to 10A diodes to replace this shorted one. I also may look through some of my parts piles to see if I can find a heavier rated one to use until I place my next order. These are the diodes I am looking to order:

http://www.digikey.com/product-detai...Q040-ND/352058

http://www.digikey.com/product-detai...5GI-ND/2153531

I also wanted to make a correction to my previous statement. The 5vsb FET is still a 5n60C. I had ordered some 7N65C to use for the 5vsb but I had forgotten that I decided against using it when I swapped out the resistor a few weeks ago. (It was a while before I got around to doing these tests.)

[ben7]

Re: Antec SL-400 5vsb Mod with NCP1200

Can't wait to see the carnage LOL