Please, check for possible mistakes

[OhMy]

Hi!

I finally decided to try to make an inverter, so, after searching for a manageable circuit, I settled on this one http://danyk.cz/menic230_6_en.html.

There was no PCB, so I made my own for a DC booster part. Most likely it is a bad design, but I'm far from being an expert in electronics.

Can you, please, check with a fresh look it against the schematic for mistakes? I'm asking because, while waiting for an UC3843 to arrive, I took a used UC3842 instead to test the circuit, but it didn't work, and the IC got quite warm, but not hot. Luckily there was no smoke . Supply voltage was 22-23V (something to do with a start-up voltage of the UC3842 at 16V), not 12V.

The voltages at the IC's pins:
1. 1.34
2. 0.68
3. 0.0018
4. 0.568
5. 0
6. 0.0025
7. 23
8. 1.28

Thanks

 

[(*steve*)]

One thing you might want to watch out for is the clearance between your 325VDC rail and lower voltages. Same also between your inverted output and lower voltages.

If the traces need to run close (say to the leads of a capacitor or an optocoupler) then you may consider placing a slot in the board.

 

[OhMy]

Yes, that's the problem of bad (mine ) PCB designing. But so far it doesn't look like there's ever going to be a 325V output from this inverter...
Quite possible, that the UC3842 is damaged.
Maybe someone could comment about the measured voltages of this IC?


Most of my built projects tend to fall into one of the two categories- the ones, that work right away; and the ones, that are just plain dead- no magic smoke either.

 

[jpanhalt]

Your gate drive voltage (Pin6) is lower than I would expect for any duty cycle >25%. Do you have a scope to see what the peak voltage is?

John

 

[(*steve*)]

Do you have a datasheet that shows sample circuits for this chip? The one I found doesn't have that (just an open loop test circuit).

I've looked at the page you refer to, but it doesn't explain the operation of the circuit so I really can't guess at whether these voltages are OK or not.

Is your board layout as viewed from the component side or underneath? It appears to be from the component side.

One obvious problem is that your power rail measures 23V when your schematic says it is 9 to 15V, however I take the point that the chip you used has an undervoltage lockout requiring more than 16V to turn on.

However you need to take into account the fact that other parts of the circuit might be designed for a lower voltage.

One obvious thing is the zener diode protecting the gate of the main switching mosfet. This is a 16V device and the current through it is limited only by a 10R resistor. At a 23V input voltage it would be conducting heavily and would explain the heating of the chip.

However you measure the voltage at pin 6 as 0.0025V. Whether this is because it is at effectively ground level, or if it is because the duty cycle is low, I can't say.

What is the voltage on the output capacitor? Is it 300ish volts?

It is also possible that the zener diode is inserted back to front.

The PCB design fro the pin 6 through the 10R resistor, the zener and the 10k resistor and then on the the gate of the mosfet seems right.

I assume T2 and the associated circuitry connecting it back to the chip are for cycle by cycle current limiting, but I can't comment on what, if anything, might be going wrong here as I don't really know how this chip is typically used.

Beware of getting the phasing of the windings of T2 wrong. Although that's probably not a factor in you problems at present.

Sorry I can't really be of any significant help.

 

[OhMy]

Jpanhalt, no, I don't have an oscilloscope. My most advanced measuring tool (and almost the only one) is a UNI-T UT61E multimeter.

*steve*, you gave some good points to think about.
The datasheet I have for these chips is too large to attach here and does not compress with winrar, so here are the first page and a sample circuit of it:

The board is viewed from component side.
The datasheet says, that the output voltage should be 5V , so I'm not even sure, why that zener is needed at all there. If it was inserted incorrectly, still there should be al least 0.7 or so volts, I guess.
There is almost zero voltage at the 325V side- after connecting the power supply, the voltage slowly rises to about 0.6V and that's all .
About the transformer phasing, well, I tried my best to not make any mistake, while winding it, but who knows... Sometimes it only takes three trees to get lost; or one can spend halve a day searching for a thing, that was in a pocket all the time .

 

[(*steve*)]

The datasheet says, that the output voltage should be 5V

That's possibly the ouptut voltage of a test circuit, not pin 6 of the chip.

so I'm not even sure, why that zener is needed at all there. If it was inserted incorrectly, still there should be al least 0.7 or so volts, I guess.

If you look at the internal circuit for this chip you will notice that the output is a pair of transistors which can pull the output to Vcc or down to ground. Thus it will rise to almost Vcc -- in your case well above the zener voltage.

The zener is thre to stop small voltage spikes from destroying the mosfet. It is not intended to be conducting for any period of time.

There is almost zero voltage at the 325V side- after connecting the power supply, the voltage slowly rises to about 0.6V and that's all .

Well that tells you the circuit simply isn't switching. Whether that is because you've killed the output of the chip, or something else is wrong, it's hard to say.

What current does the circuit draw from your power supply?

About the transformer phasing, well, I tried my best to not make any mistake, while winding it, but who knows... Sometimes it only takes three trees to get lost; or one can spend halve a day searching for a thing, that was in a pocket all the time .

Assuming you get the number of turns right, the issue will be whether the phasing is correct. If it's not you need to swap over either the input or the output connections (but not both). I can't tell you what the symptoms of incorrect phasing might be.

edit: you can probably upload the pdf to some pdf hosting service or give us the URL where you got it from. google for "free pdf hosting"

 

[OhMy]

the output is a pair of transistors which can pull the output to Vcc or down to ground

And again you appear to be right...

I think, it's enough torturing you for a while. The best thing now is to wait for the correct ICs to arrive from China (waiting from 20th March, could be another week or a month, shipping time gets just worse, as time goes), and then see how's it going.

I'll try to update thread with the result of this.

Thanks.

 

[KrisBlueNZ]

I don't think you've answered the question of why pin 7 measures 23V when the specified supply voltage for the circuit is 9~15V.

Also, pin 8 is the reference voltage output from the UC3843; it should be 5.0V.

 

[(*steve*)]

The chip (in common with many like this) has a zener diode across the power connections so it can be suplied via a resistor from a high voltage source.

The absolute maximum voltage for the device is just a tad shy of the zener voltage because a voltage source higher than this is bad news. However the chip will (probably) operate correctly when the internal zener regulates the voltage to a value just above the absolute maximum.

The difference between this chip and others in the range is the required input voltage to enable the inverter. The one the OP tried has a higher (16V) turn on voltage (compared to 8.5 ish) than the original part..

The latter reason justified the higher Vcc rail. In any case 23V is less than the absolute max.

Well spotted on the Voltage reference output. Although a low reference voltage would simply cause the output voltage to be low. I suspect there's more wrong than just that.

 

[KrisBlueNZ]

The difference between this chip and others in the range is the required input voltage to enable the inverter. The one the OP tried has a higher (16V) turn on voltage (compared to 8.5 ish) than the original part.

Well spotted to you

Well spotted on the Voltage reference output. Although a low reference voltage would simply cause the output voltage to be low. I suspect there's more wrong than just that.

There's no telling what else the reference voltage is used for inside the chip. If it's being dragged down externally, that could prevent it from working. If it's not being dragged down externally, there's internal damage, and probably to more than just the reference.

 

[OhMy]

There is a great chance, that this IC is damaged- it could have been from the start, or I burned it. Probably because I just now realized, that C945 has different pinout than C237 or C547...
Either way it's no use to dig deeper, till I get the new chips to put in and burn .

 

[OhMy]

Hello again!

Today I finally received the correct and new IC UC3843, and it worked right away !

At a supply voltage of 10.7V and output 300V this converter draws about 35-40mA unloaded, when an analog voltmeter with a resistance of 270k is connected, the current draw increases to almost 100mA. Any thoughts about, how good or bad that is? I tend to think, that it's not really good, but then again- this is my first working DC booster.

Now for the problems ...
It seems, that my current setup won't be able to deliver promised 150W of output power. Maybe the transformer is too small or the core is from a bad (unknown) material.
The output voltage of ~250V at a steady supply ~12.6V dropped to some 215-220V, when it was loaded with a 100W 230V rated bulb, the current was ~0.41A, so the output power was about 90W, but the voltage could not be kept at a steady level. This didn't happen, when loaded with just a 40W (maybe 60W- letters not visible) light bulb.

The transformer makes some annoying low frequency noises (zip archive), mostly when unloaded or slightly loaded. These noises change, when I touch one of the transformer's output pins or its core (should it be conductive?).

Your thoughts, please?

I know, that these measurements are not lab-grade, but for a quick test they should be OK.

Thanks

 

[(*steve*)]

Firstly, I wouldn't be poking my finger anywhere near the output (or the output transformer) of that circuit.

Secondly, Well done!

Today I finally received the correct and new IC UC3843, and it worked right away !

Excellent!

At a supply voltage of 10.7V and output 300V this converter draws about 35-40mA unloaded, when an analog voltmeter with a resistance of 270k is connected, the current draw increases to almost 100mA. Any thoughts about, how good or bad that is? I tend to think, that it's not really good, but then again- this is my first working DC booster.

(And that analog multimeter reads 300V?) Well, ignoring the quiescent power consumption of 400mW or so, for an output power of 333mW, the input power is 642mW, so you're looking at about 50% efficiency.

That sounds really poor, but efficiency normally increases with load up to some maximum value before dropping again.

The best thing to do is to measure input and output power for a range of loads, and then graph the efficiency. From your statements below, graphing the output voltage would seem to be a good thing too.

It seems, that my current setup won't be able to deliver promised 150W of output power. Maybe the transformer is too small or the core is from a bad (unknown) material.

Well, the transformer is critical. The correct choice of transformer will depend on the voltages, the frequency, and the power you need. If the transformer is "unknown" it may not be suited to any of these criteria -- although clearly you are getting the voltage you want.

The output voltage of ~250V at a steady supply ~12.6V dropped to some 215-220V, when it was loaded with a 100W 230V rated bulb, the current was ~0.41A, so the output power was about 90W, but the voltage could not be kept at a steady level. This didn't happen, when loaded with just a 40W (maybe 60W- letters not visible) light bulb.

When you say "the voltage could not be kept at a steady level", can you describe more exactly what happened to the voltage, and anything else relevant? Did it drop over time? Didi it fluctuate? Did it fall to zero? Did something start to smoke (or get really hot)? Does the input voltage start to fall?

The transformer makes some annoying low frequency noises (zip archive), mostly when unloaded or slightly loaded. These noises change, when I touch one of the transformer's output pins or its core (should it be conductive?).

It is quite likely that at low powers the transformer is called on to provide intermittent bursts of power and it is not until the power exceeds some value that the regulator needs to be running continuously. What you are probably hearing is a single pulse of (say) 20kHz sound. You hear that as a click. When it's running continuously, the sound is probably a lot louder, but you can't hear it.

The sound comes from the transformer. As energy is stored into it and released, parts can flex and this produces noise. You can damp this out by securing the transformer to something solid in many cases.

I know, that these measurements are not lab-grade, but for a quick test they should be OK.

They're lab grade for your lab, and seem perfectly acceptable at this stage

 

[OhMy]

Hi, (*steve*), and thanks for encouragement.

Firstly, I wouldn't be poking my finger anywhere near the output (or the output transformer) of that circuit.

Yes, I kinda know, how it feels to get touched by some more, than just a car battery, but nonetheless I refreshed my memory on that, when I was holding the circuit in place by pressing on the output capacitor's aluminum top, and trying to adjust the voltage regulating pot with a metal tweezers. Before that I thought, that aluminum shell of the capacitor is isolated form its terminals ...

Yes, I used an analog (mechanical) DC voltmeter with a full scale of 300V.

Well, the transformer is critical. The correct choice of transformer will depend on the voltages, the frequency, and the power you need. If the transformer is "unknown" it may not be suited to any of these criteria -- although clearly you are getting the voltage you want.

The transformer's core material is of unknown hopefully ferrite type, but the windings are my own; wound as close to author's description as possible. Although the core is a bit smaller. I had to use one with a round central column of 1.4cm diameter, so it is 1.54 square centimeters versus the recommended 1.8. Could this seemingly small difference make all the difference at available output power?

When you say "the voltage could not be kept at a steady level", can you describe more exactly what happened to the voltage, and anything else relevant? Did it drop over time? Didi it fluctuate? Did it fall to zero? Did something start to smoke (or get really hot)? Does the input voltage start to fall?

When loaded with a light bulb (rated 100W at 230V), the output voltage dropped from about 250V to just ~215V, so this DC booster was unable to handle a load about 100W.
The input voltage couldn't drop much- at that time it was taken from a car battery. The switching mosfet IRFZ44N, cooled by just one TO-220 four-finned radiator, got hot pretty quickly, so I couldn't torture it for extended periods. I probably will change it to an IRF2807 one.

It is quite likely that at low powers the transformer is called on to provide intermittent bursts of power and it is not until the power exceeds some value that the regulator needs to be running continuously.

If you say so... It could be right, because with an extra ~250uF connected to output, these impulses get close to just 1Hz level. I thought, that the UC3843 chip works by varying the pulse width of impulses, not by burst mode.


I should take some better measurements, as you said.

 

[(*steve*)]

Before that I thought, that aluminum shell of the capacitor is isolated form its terminals ...

I bet that woke you up.

The transformer's core material is of unknown hopefully ferrite type

Ferrite is very different from iron, so assuming you can tell one from the other... (So let's assume it is some sort of ferrite core)

Where did it come from. Ideally you grabbed a transformer from a swithchmode power supply, unwound the original windings, and wound your own.

but the windings are my own; wound as close to author's description as possible.

Clearly fairly well because you have a good output!

Although the core is a bit smaller. I had to use one with a round central column of 1.4cm diameter, so it is 1.54 square centimeters versus the recommended 1.8. Could this seemingly small difference make all the difference at available output power?

In general the power is related to the area of the core, so you can get 1.54/1.8 = 85% of what was possible with a larger core (assuming the cores are of the same type)

When loaded with a light bulb (rated 100W at 230V), the output voltage dropped from about 250V to just ~215V, so this DC booster was unable to handle a load about 100W

I'd say that this is effectively your limit, and probably not for continuous use.

The input voltage couldn't drop much- at that time it was taken from a car battery.

Excellent. I was hoping that was what you were using.

The switching mosfet IRFZ44N, cooled by just one TO-220 four-finned radiator, got hot pretty quickly, so I couldn't torture it for extended periods. I probably will change it to an IRF2807 one.

You probably want to stick a much larger heatsink on it too.

If you say so... It could be right, because with an extra ~250uF connected to output, these impulses get close to just 1Hz level. I thought, that the UC3843 chip works by varying the pulse width of impulses, not by burst mode..

Almost any SMPS will go into burst mode when the power being drawn from the supply is less than is supplies when the pulses delivered by the controller are at their minimum width. Once you get there, a single pulse will pump the output voltage above the set limit and it will not have dropped below the desired regulated voltage by the time the next pulse is due. Most often the controller simply does not produce a pulse. Now you're effectively in burst mode.

An SMPS that doesn't go into burst mode with a very small load will increase the output voltage well beyond the regulated limit.

 

[OhMy]

Ferrite is very different from iron, so assuming you can tell one from the other... (So let's assume it is some sort of ferrite core)

Oh, when I see iron, I can tell, that it's iron . But a different story is the 72 millions (approximately) types of ferrites and iron powder cores.
This one is some sort of ferrite from some sort of unknown equipment's power supply, or maybe some other purpose transformer... So, yes, it's a precise science .

P.S. Some advice from an expert (me)- never try to figure out, what would happen, when this kind of circuit got short circuited, while measuring the current draw from a power supply with an ammeter in a 200mA range.
You might get sorry for that...

 

[KrisBlueNZ]

So you were getting about 215 VDC into a 100W, 230V light bulb? That's good!

Is this just using the converter, without the 50 Hz H-bridge section?

You must have been lucky with your transformer core. There's a wide range of them around. Do you have any way to measure the inductance of the windings?

 

[OhMy]

I'm lucky, that finally at least one circuit works, that's for sure .
It's just the converter without the H-bridge, but it shouldn't make much difference, I guess.
No, I can't measure inductances. It's one of the things, that I'm trying to solve from time to time, but every time I fall back...