The SMPS modding thread (aka. The "Ask Dr. Steve" Thread) / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

The SMPS modding thread (aka. The "Ask Dr. Steve" Thread)

1 2 next

Move Thread

LAN_403

Dr. Shark

Thu Feb 09 2006, 12:30PM

Registered Member #75 Joined: Thu Feb 09 2006, 09:30AM
Location: Montana, USA
Posts: 711

We had a coulple on the old board, and we need one here!

I'll start with a question that evil Dr. Conner did not want to answer by email, so I'll repost it here. Maybe he's got more time now or someone else can jump in to fill the gap:

"On a PC PSU (200W AT) which I am in the process of modding, the output transistors of the TL494 provide pulses at only 2V PP, even though Vref is correctly at
5V and Vcc at 12V. This is of course insufficient to drive the output
transistors into saturation, so the PSU is not running on its own. I
have tried swapping the 494 for a new one, same problem. After lots of
scoping around, you are my last resort for a hint, so I hope I don't
bother you too much by asking you to do my work

Steve Conner

Thu Feb 09 2006, 12:48PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

The TL494's output stage is open collector (or emitter follower if you wire it the other way round but we won't go into that here.) It can only sink current and needs an external pull-up. So, if someone says a TL494 is giving out a too-small pulse, straight away I think of that.

Usually, the collectors are connected to the ends of a centre-tapped GDT winding, and the centre tap is connected to 12V. This makes a classic push-pull transformer drive for the power devices. In the case of a supply using bipolar power transistors instead of MOSFETs, there may be a current limiting resistor between the centre tap and the 12v rail. So I would check that resistor had not burnt out. Probing one of the TL494 outputs with a scope would show that your 2V p-p waveform was lying down near 0v.

I would also check, if the power supply used MOSFETs, that one of the gates had not failed short circuit. This would short out the gate drive.

Finally, if you changed the gate drive circuitry, maybe you forgot the need for a pull-up, and left it out?

Dr. Shark

Thu Feb 09 2006, 12:59PM

Registered Member #75 Joined: Thu Feb 09 2006, 09:30AM
Location: Montana, USA
Posts: 711

Thanks! Its great how you are sharing you knowledge even with "doh!'s" like me.

uzzors2k

Thu Feb 09 2006, 06:28PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

Might as well keep things going, so...

I never metioned it on the old forum, but my modded PSU is actually quite stable while being overvolted. I've been running my ZVS setup at near 200W on one rail without any failuire( yet ).

Suggested reading for SMPS modders

Dr. Shark

Fri Feb 10 2006, 02:40PM

Registered Member #75 Joined: Thu Feb 09 2006, 09:30AM
Location: Montana, USA
Posts: 711

Yep, thats a good one, and of course the TL494 Datasheet is recommended reading.
I am also attachting a schematic of an AT Power supply, because these are more easily available for free than the newer ATX ones.

Marko

Fri Feb 10 2006, 02:52PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Seen this lots of times.
I once tried to make larger version of small standby SMPS, and failed miserably.
In most schematics its mosfet driven and main drivers are bipolar transistors, exceptions can happen offcourse %-6

I used larger MOSFET (later bipolars) and transformer from monitor (SMPS, internal feedback)

I studied the oscillator and realized its just simple blocking oscilator controlled by throwing base/gate to negative (for output voltage regulation)

When couple of mosfets exploded (even with original schematic) I switched to bipolar.
One small transistor held long and blew up when I removed the load from transformer (motor).

After that I blew two 2SC2625 and left the stupid idea (seems impossible with their current and voltage ratings, but they died in seconds)
All protection diodes were fast schottky (originally from SMPS) and current was limited to around 200mA by isolation transformers, and voltage was set exactly to 220V AC..

Few times I got it to work with optocoupler regulation but that turned in LF, parasitic relaxation oscillation (few Hz) of output and could not be helped.
Transistor also got red hot when driven too low by optocoupler..

Transformer was of original SMPS use in monitor, and worked for loong time in similar setup..

Something went terribly wrong and I dont know what.
Al protections were included, I rebuilt exactly same schematic from that ATX supply, later i followed tracks on monitor and it was too bocking-like oscillator.

I may open new topic and include few schematics, if anyone interested...???

uzzors2k

Fri Feb 10 2006, 04:10PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

I'm interested, just keep it coming on this thread...

Marko

Fri Feb 10 2006, 06:44PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

This is what I tried to 'mod' (use bigger transformer) and it didnt work, even basically (i had to remove zener diode and increase gate cap to get it working, and without zener at 300V mosfet was torn in pieces ultimately.

Bipolars worked better but also died misteriously after few seconds...

Even after I copied almost exactly SMPS from monitor, same transformer, transistor and much conponents but I didnt make it.
Even when got working I couldnt establish stable feedback regulation like that in schematic.
It doesnt seem to affect frequency and duty cycle a lot, just lowers gate/base voltage too much and transistor stops to open fully..

such tiny 300VDC powered self-oscillating SMPS is what never worked good for me, I tried also one other schematic (home-made at all) and this one worked very good but one resistor had just to disipate 20wats... resistor R3 that limits current to entire multivibratorheats up dramaticaly, and it must be small enough to provide high base current for driver.

Super simple astable multivibrator of transistors, duty cycle precisely controlled by optocoupler and output voltage was very stable (rated voltage of zener + drop on coupler diode)...

Dr. Shark

Sat Feb 11 2006, 05:12PM

Registered Member #75 Joined: Thu Feb 09 2006, 09:30AM
Location: Montana, USA
Posts: 711

I think we had this on the old board, but I just got a reminder: Make sure you replace the electrolytics on your PSU if you mod it for a higher voltage! I was just running my AT PSU at 24V (it turned out that in fact the switching transistors had blown, the 2Vpp form the TL494 are perfectly normal - I keep forgetting bipoolars work on current, not voltage) for a few minutes and suddenly it goes bang, lets out its smoke, and it was the output filter electrolytic that had exploded.

Marko

Sat Feb 11 2006, 05:35PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

and it was the output filter electrolytic that had exploded.

Without load and feedback regulation of output (I tried it on my astable circuit) by ratio of turns output voltage can reagc ~ 80V.
Surely low voltage caps will blow with that (obivously you are talking about modding feedback for more voltage, ofcourse you must watch the capacitors).

Im interested has anybody built working mini-smps (like gsm cell charger) ?
(without using low voltage side oscilator, TL494 or so) ?
As I said it seems very hard..

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob