SMPS current not rising higher despite saturation.

[potcore]

Dear Engineers

Our bench power supplies having been going missing from work (people are allowed to “work from home”) and I have been asked to make some. I have tried out a SMPS power supply with TL497, but have encountered problems in my investigations….

(My scope is a cheap $200 (new) type, and my readings are perhaps a little rough, though indicative.)

I have just designed and tested a flyback SMPS using TL497 controller IC to give V(out) = 8V with V(in) = 5V

The Flyback schematic is as follows………………………………

When the secondary load was ~2K resistive, the circuit waveforms and V(out) appear to be in order.
However, when the secondary load was 24R, the waveforms are unexpected and V(out) drops to ~2V.

I would be grateful if engineers could possibly shed some light….

Primary & Secondary were wound on an RM14 potcore (no air gap)
L(primary) was ~ 18mH (about 50 turns 056mm enam. copper wire)
L(secondary) was ~ 23mH (about 64 turns 056mm enam. copper wire)

Primary and secondary coil resistance ~ 1R

V(out) was intended to be 8V
V(in) was 5V.

The power switch and diode used were those internal to the TL497.
The “negative” rails of primary and secondary circuits were tied together since the voltage feedback from the secondary was direct (no opto-isolator).

Flyback waveforms with Secondary load ~2Kohms resistance…………….

Flyback waveforms with Secondary load ~24R ohms resistance…………….

Since the rate of current rise in the primary when the switch is closed should be 278A/second (= V(in) / L), and since with either load, the switch was obviously never ON for more than 15uS (see switch waveform) , -then I would expect the peak primary current to never be higher than ~4mA.
-However, the primary current sense resistor voltage goes up to 200mV , meaning the peak primary current was 200mA.

This “overcurrent” indicates saturation, However, if saturation occurred, then I would expect the primary current to keep on rising, and not “clamp” at 200mA.

When the load is 24R, The primary current sense resistor voltage shows that the primary current does indeed STOP rising at 200mA and then REMAINS at that value for a while.

I cannot understand why the primary current stops rising at 200mA, (when surely the core is saturated and it should go higher), and then just goes along at that level for a while.

I would be very grateful if any thoughts could be offered.

 

[Madcow]

I don't think your saturating the transformer.
I think you have too much inductance. You probably
want to be around 30uH. 18 millihenry is pretty high
for 15 uS. Here are some calulators.

http://www.national.com/appinfo/power/files/flyback-mathcad_example.doc

http://www.daycounter.com/Calculators/Flyback-Transformer/Flyback-Design-Calculator.phtml

 

[itsmoked]

Measured the transformer resistance??

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[Madcow]

I'm not familiar with the TL497 but it looks like
your set for 500mA peak current limit. So in discontinuous
mode you'll be limited to about 100mA @ 8vdc.
I also noticed the app.notes show the current sense in
the highside input, and you have it on the rtn.
What type of core material? Flybacks like a gap.

 

[potcore]

Hi, Thanks for replying....sorry my diagram is misleading concerning the position of the primary current sense resistor.....in the actual circuit it is between pins 13 & 14 as in the app note.

The transformer is RM14 ferrite core set ungapped,8000nH (AL) N67 (Epcos Manuf No. = B65887ER67)
-the primary and secondary res is about each 1 ohm. (each 0.56mm diam. enamelled copper wire)

 

[BobbyNewmark]

Why not just buy more power supplies?

 

[potcore]

Hi,

Unfortunately , we're not buying more as they keep going missing....not only that, but i think that once ive gotten the hang of it, i'll be able to make them very cheaply to any spec we would need. At the moment we have engineers going to the production dept begging production staff to make them a "5V power supply" because they don't have them and need to test their microcontrollers etc.

 

[itsmoked]

You realize that you can buy Wall Warts that will run most modern micro systems for about a dollar....?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[MacGyverS2000]

I can understand engineers occasionally blowing up a cheap supply with faulty prototypes (though any normal PS should be safe against that), but having an engineer abscond with them over and over again? Put them on a shelf with a serial number.. sign them out to every engineer that wants one. If they blow it up, they need to bring it back in for a replacement. Sounds childish, but if the engineers want to treat test equipment like toys, well...

I'm with Keith... if they're using multi-hundred dollar supplies to power their 5V micros, buy a $2 wall wart at Wally Mart and be done with it (though I'd still have them check it out with a serial number just for spite... children!).

Dan - Owner

http://www.Hi-TecDesigns.com

 

[potcore]

Hi,

Actually some of the power requirements are more complicated than just 5 volts for micros. Sometimes its split supplies or 24V at 2Amps etc etc. In fact, the 24V supply that I used to use disappeared, and I now use a supply which was ripped out of a PC for that 24V. It's considered that it's best and cheapest to just make the supplies ourselves.

 

[BobbyNewmark]

That is an idiotic policy. If somebody is stealing or hoarding equipment, you stop them. If the company wants to outfit home labs, it should do so using proper requisition, asset tagging, and sign-out procedures.

Custom-designing and hand-building hardware is a waste of money. The result will be garbage**, the hardware cost will likely be higher than just buying decent stuff off the shelf. Certainly the cost of working around the custom units' deficiencies will overshadow the supposed savings.

**No offense to you. Designing a good variable power supply is hard.

"i'll be able to make them very cheaply to any spec we would need."

I want slaved outputs; an ammeter and voltmeter for each output; adjustable current limiting; 2% line, load, and temperature regulation; EMI tolerance so that when my power MOSFET starts "singing" the upstream power supply won't decide to squirt out 50 volts; low radiated EMI to not piss off my sensitive instrumentation circuitry; good regulation under transients; reasonably small internal capacitance so that accidents fry as few chips as practical; and a quiet fan if any.

Dan and Keith: A lot of wall warts are ungrounded, and therefore stray capacitance couples ~60 VAC onto the output ground. It's perfect for popping MOSFET gates. This is one reason I use a proper benchtop supply if at all practical. (Laptops have this problem, too. I've fried at least one prototype board because it was hooked up to a laptop with an ungrounded power brick.)

 

[itsmoked]

hmmm, Interesting point. I've never had enough capacitive leakage to ever cause me a problem with small switchers. I'd think if it was much you'd have complains by people touching the barrel and a ground.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[MacGyverS2000]

Same here, no problems to date, though that doesn't mean one doesn't exist. I use wall warts to power all of my dev boards, though I try to limit the amount of circuitry on the proto sections... when it comes time to add extra circuitry, that usually means a custom board.

We had our weekly engineering meeting yesterday and I made a quick mention of this thread. I got a good chuckle form everyone about engineers "losing" the PS's at home... again, and again, and again. We don't get to work from home with our equipment (asset tracking would get cranky if we tried to take a $100k signal analyzer home), but then again, we actually track our assets.


Dan - Owner

http://www.Hi-TecDesigns.com

 

[BobbyNewmark]

The AC current from a wall wart is generally tiny, less than a milliamp. It will not hurt most digital chips because of their ESD diodes, but small-signal MOSFETs, laser diodes, and such may not fare so well.

Then again, it doesn't take many man-days of debugging a zapped component to pay for a good power supply.

 

[itsmoked]

Thanks Bobby.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[zeitghost]

Don't security notice when people walk out with a psu stuffed down their trousers?

 

[itsmoked]

They just think the guy's excited to see them.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[potcore]

Hi,

The need is not generally for really high spec supplies like BobbyNewmark described one.

But in UK, a 0-30V, 2A power supply would cost something around $350.00 and hence its thought worthwhile to look into making them.

There is flexi-time and it'd be difficult to check everyone over (search them for hidden PSU's)all the time.

 

[geekEE]

$350 sounds cheap compared to the man-hours involved in iterations of designing, sourcing parts, manufacturing, and testing. And won't these power supplies go missing as well?

Is it just power supplies that walk out the door or is this just one of many things that go? I've never seen that happen as an ongoing thing at any place that I've worked.

 

[zeitghost]

You can get a halfway decent psu for a lot less than that these days.

CPC seem to do reasonable range starting from about £35... as do Maplin.