LM2577 Voltage converter problems...

[formant84]

Okay in a fit of desperation I managed to resolver the inductor from the eBay board and bridge it onto mine. It gives a higher voltage out so had to adjust the pot to bring it back down to 12v, other than that it's exactly the same, still reading just over 3v at the input! Perhaps the main problem is the caps after all?

 

[Blueteeth]

I agree that the inductor was saturated. As its a lower frequency switcher (52kHz) it requirs a rather large inductance, 100uH.. one with a saturation current of 3A+ will be rather large - just like the one on the ebay board. A non-SMD one will be larger, often toroid for those siwtchers - you can see them on cheap car phone chargers almost always using an MC34063A.

Firstly I would get USB out of the equation, and power it with a 5V supply - either bench or a wall-wart/charger/ plug in PSU. This has the risk that it'll probably be able to provide more current - risking blowing something on your board, but will allow you to see just what current your circuit is drawing - always use the 10A setting first on multimeters It does the job. USB has current limits to protect its outputs, and can also be controlled by the PC itself, these can make life more complicated when debugging. Especially with a large input cap - the surge current when plugging it into USB might trip the current limit. So, for testing, a basic 5v PSU

I have debugged many DIY smps's and many sites/documentation seem to focus on layout for spurious noise, or interference on the feedback line - but skip actually 'getting it working'. Basic debugging would be to get your multimeter out and measure things like, resistance of input to ground, output to ground, checking for shorts on the board, shorts across caps etc.. Also your chip *may* already be damaged in some way, so best to check continuity, and diode tests on its pins to see if any are shorted.

Then if you can, replace the caps with over specced ones. I know in the quest for SMPS efficiency many go for tantalum caps or ultra small compact SMD ones, and they may increase efficiency, but a bog standard electrolytic, over specced capacitance and voltage rating will be 'functional'. Essentially its all a process of elimination and apologies if I'm patronising here, but it really is a case of probing, replacing parts and testing.

I'm sure we've all had odd circuits making us scratch our heads where its been a floating pin, mis-soldered part, or a short somewhere.

 

[formant84]

Not patronising at all, I appreciate the help! Debugging circuits is a real weak point of mine. The board is perfect as far as the soldering goes. I've replaced the switcher with a new one just to make sure. Before I read your reply I ordered some low ESR 220uf caps, maybe should have gotten larger but I've at least gone for ones rated 35v instead of the 16v ones I have in. Out of interest I'll try putting in some larger capacitance ones now and see if that makes any difference.

I'll also try hooking the board to my proper power supply and re-check the current consumption. Then I can take those resistance measurements you recommended and compare them to the same readings on the board that works.

My worry now is that it's my board design causing the issues! everything on the power supply section is close together within a 4x3cm area, but my 5v line takes a bit of long route to get to the input so it borders the circuit on 3 sides, could this cause interference issues?

Edit: Sorry for the slow reply, I've had no internet here for 2 days!

 

[formant84]

Just noticed I had two low ESR 330uf electrolytics sitting around! Popped them on the board and connected to 5v supply. things seem better at first:

Input voltage reads as 4.6v, output 12v, current consumption approx. 430ma. Then after about a minute everything goes peculiar. Voltage at input begins fluctuating between 4.5 and 4.8v and the output voltage keeps dropping down to like 7v then coming back up to 12v, holding there for a couple of seconds then dropping out again, and every time this happens the current drops to zero then comes back up to 430ma. It's like the whole circuit is switching off and on!

Interestingly this doesn't happen when I tried it back on the USB supply, I get the same result of 3.3v on the input voltage but the whole thing holds steady, although in both cases the switcher is getting rather hot still.

Resistance between input and ground just over 5k, between output and ground just over 1k. On the ebay purchased board it's 18.2k at input and 8.8k at input.

This thing is driving me insane!

 

[Blueteeth]

I think its time we had a schematic, and a picture of the PCB layout

If you're worried abotu others stealing your design or something, of course just the SMPS part would be ok. Also, I just thought - to eliminate any problems there might be after the switcher - the rest of the circuit its powering, have you powered this by 12V from a separate supply and checked the current draw? I mean, if that is drawing a lot of current, then it could be that is the problem, and the poor switcher on your PCB is trying to keep up. I may have missed a previous post where you have already checked the rest of the circuit, so apologies if you've already tried.

Also, the partnumber/datasheet (even a photo if needs be) of the inductor you're using on your board would be handy.

From the sounds of it, you're doing well as far as debugging goes, and having a 'reference board' - that ebay one that works as expected' is a great benchmark for comparison! There will be differences as you mentioned your own board's switcher is powering a microcontroller circuit, but for the most part it shouldn't make much of a difference.

And input of 5k is fine. And as you're drawing 430mA, although the USB spec says '500mA max', I suspect you're drawing an average of 430mA with current spiking a bit higher than that - so the USB supply is current limiting, bringing the average voltage down to ~3.3V. Few USB hosts supply true 500mA anyway.

If it runs fine for a while then starts going peculiar, that suggests to me that something is overheating - as that generally takes time, and many integrated switchers have thermal shutoff that waits for the temp to fall back down from a peak, then starts the circuit again - making it go on and off.

Sometimes a 'small area' design for SMPS is actually a good thing, it keeps high current paths short, and providing the input and output capacitance are close to the SMPS chip, it should minimise radiated noise. That said, in terms of debugging by desoldering and resoldering new parts - can be a pain.

You'll get there sir, and as cheesy as it sounds - its a good exercise in debugging!

 

[formant84]

Well, I've only mounted the power circuitry so far. knowing the tendency things have not to work first time I thought it was wise to leave the rest of the circuit out of the equation rather than risk killing it the first time I powered it on! so as far as schematics go I've literally just got the same as the circuit I first linked to:



board layout of the final circuit looks like this:



But given that the rest of the parts aren't mounted what I effectively have is this:



As you can see the inductor I was originally using was a weedy little axial leaded RF choke type, so now I'm kind of limited to trying to find an inductor which will fit the space I left for it!

Do you think the thin trace going from the 2577 to the inductor could be causing problems? I know the traces should all be as thick as possible but it was the only way I could see to route it! as you can probably tell it was originally designed to be home etched single sided with jumpers. In retrospect when I decided to have it fabricated properly I should have gone back over it to see what I could improve knowing that more vias and traces on the other side were an option!

(EDIT: Could have added a ground plane too!)

The inductor I'm using at the moment is radial leaded with a ferrite core, specced as follows:
Inductance: 100uH
DC resistance: 0.3ohms
Max DC current: 790mS
Resonant frequency: 5MHz

No idea what the actual part number is hence no data sheet unfortunately.

And as mentioned in my last post I'm now using 330uf 35v low ESR caps.

It's definitely the LM2577 that's overheating. it gets quite hot but not too hot to touch, and the adjustment trimpot gets fairly warm too. This inductor I'm using now doesn't seem to be getting warm at all, not sure whether that's down to the inductor itself or the new caps.

As I mentioned in an earlier post, I had the rest of my circuit breadboarded at one point with the eBay board attached and all powered from USB with absolutely no problems, so theoretically if I can get the switching circuit to work as intended the design SHOULD work!

At this point though I'm beginning to think that it must be my board design that's to blame. Should have done more research about SMPS before I made it!

 

[ronsimpson]

What diode are you using. It should be fast.

Your ground is thin and long.

 

[Blueteeth]

Nice! ok, lets see.
The layout is less than idea, but... should be functional. With a linear regulator and microcontroller being powered by your supply - I can't see this drawing more than 50mA from the LM2577 circuit. That could be a problem as even though we look at the 'maximum' current a SMPS can supply, with these simple switchers, all the datasheet references expect a load current quite high, often > 1.0A.

There isn't a ground plane, and the trace supplying power to it from the USB is indeed very thin, but should be good for a few 100mA. There are a couple of traces going under the chips pins - one of which is the feedback line.
This can be prone to noise (any high current traces) but given that you've got it going straight to a pot, that should load it down enough to keep its output steady. The lack of ground plane or heavy ground traces can cause all sorts of issues which I won't go into (another big post maybe?)

Ok, even though the layout isn't great, I'm sure you can get it working without changing that, and the biggest issue I can see so far is the inductor.

As you can see the inductor I was originally using was a weedy little axial leaded RF choke type, so now I'm kind of limited to trying to find an inductor which will fit the space I left for it!

The inductor I'm using at the moment is radial leaded with a ferrite core, specced as follows:
Inductance: 100uH
DC resistance: 0.3ohms
Max DC current: 790mS
Resonant frequency: 5MHz

You may have destroyed the chip on your board by using a smaller inductor to start with. This is because, if the current in an inductor rises above its saturation current (which could have been <100mA in your original inductor) its inductance drops significantly. This means above a certain current it behaves less like an inductor and more like a resistor, the only thing limiting the current that can flow in it is its DC resistance - and if that was low, like 0.2ohms then I=V/R = 4.5/0.2 = 21A can flow.
Of course yu're input cannot supply that and your original inductor most likely had a larger resistance, maybe 1ohm, meaning 4.5A max could have flowed, limited by the USB input to <500mA, but that takes some time to kick in.

The datasheet for the LM2577 has a vast number of equations to calculate inductor peak current, average current etc.. so we can use those to calculate rough specs based on assumptions about your circuit. If all this checks out (all the parts you've used appeared to be ok) then the only thing left is the layout - but that is of course harder to change than the parts, as you'll have to get new PCB's made up, so I guess you want to avoid that

I just deleted a whoole load of calculations from the datasheet, because it has a graph/table for inductor selection. It seems they recommend a higher inductance value for lower load currents. for an input of 4.5V min, 12V out and a maximum output current of 200mA, the table on page 17 of the datasheet recommends an inductance of: 330uh. This makes sense because your switcher is fixed frequency, 52kHz, and a higher inductor means the current in it rises slower when switched on. When drawing less current from the output its current when switched off will fall slower too, as the frequency is 'fixed' at 52kHz, this means the ripple in current will be smaller. That is for the '12v option', which essentially is the same as the adjustable version set for 12v out.
Lower ripple current means the inductors current will swing less from its average current, so if its average current is say 400mA, it won't swing between 100mA and 700mA (close to 700mA would put it in saturation).

Rather than show ALL my calculations I'll just show what I cam e up with from the datasheet making the following assumptions: A quick run down. Vin: 4.5-5.5V (probably 5V). Vout: 12V. Iout (max): 200mA. Vsat 0.3V. Diode drop = 0.5V

Duty cycle: 0.66 max. E.T (volts x time for charging inductor) = 53.3. Iind (average inductor current, at full load): 0.636A.
Using the table on page we want a y axis of 53.3, and a X axis of 0.63, this gives us again 330uh.

Although, there's an elephant in the room - your ebay module probably has an inductor of 100uH, and yet it works fine? This can only mean that 100uH is 'fine' providing it is of high enough current rating, or.. the switcher on your board is screwed, or the layout causing it to switch erratically, or... the compensation network isn't right. - thats a lot of things to check :/ I can be hard to measure the current in the inductor, and difficult to see whats happening in the ebay module without a scope.

The compensation network. Pg19 has an equation for calculating R, we don't know the value on your ebay module but we can calculate what it 'should' be based on your specs. You can plug in your numbers, but I get a maximum value of: 1066ohms,
or '1k' will do. This is based on vout of 12V, vin(min) of 4.5, and Load current max of 200mA.

Similarly for Cout, I get 79uF. and CC 130nF.

These numbers, although nicely calculated, are a bit of a joke from the datasheet, because they are specific to a certain load current.

So, after all that theory, practical things to try. Although it could well be layout (its not helping) here are things to try.

- Add a load current, say 100mA, for 12V out thats a 12 ohm resistance. If you have any power resistors it would help because we 're going to dissipate 1.2W. Check if anything gets hot.
- Replace your switcher chip on board with a new one. If its regulating, then its probably 'not dead', but you never know, sometimes broken chips can behave oddly.
- Replace the inductor with a 100uH one, specced at 3A Isat. Also you could try a 220uH, and even a 330uH, these will require lower saturation currents, say 1.5A.

I strongly suspect that the only real difference between your board and the ebay one (aside from layout, lack of ground plane and thicker traces) is that inductor. Its higher saturation current means it can take a higher current ripple without saturating.
In an ideal world your inductor would have a peak current of roughly double the output current (you draw 100ma at the output, it'll have 200mA peak in the inductor) - but that is for an ideal switcher, the chip maybe have limited on/off times which means the peak current in the inductor, even when drawing very little current - can be much higher.

Do you think the thin trace going from the 2577 to the inductor could be causing problems? I know the traces should all be as thick as possible but it was the only way I could see to route it! as you can probably tell it was originally designed to be home etched single sided with jumpers. In retrospect when I decided to have it fabricated properly I should have gone back over it to see what I could improve knowing that more vias and traces on the other side were an option!

(EDIT: Could have added a ground plane too!)

The thin trace going to the inductor? yes, very much so. Even when the current is fairly low, a thicker trace provides lower impedance. I often design boards to be home etched and end up accidentally leaving jumper links on the design when sending it out for manufacture, and yes, took me a while to be 'liberal' with via's, double side traces and thicker traces when having it built - because I HATE drilling boards, and double sided ones are just a hassle.

As I mentioned in an earlier post, I had the rest of my circuit breadboarded at one point with the eBay board attached and all powered from USB with absolutely no problems, so theoretically if I can get the switching circuit to work as intended the design SHOULD work!

At this point though I'm beginning to think that it must be my board design that's to blame. Should have done more research about SMPS before I made it!

So it seems its just the switcher, on your board. If you can get parts very similar to the ebay module (try and get the same inductor, and just solder stubby thick wires to it, then solder that to your board!) with the same compensation netowrk values, and the same output caps - and it *still* doesn't behave? then I'm afraid its the layout. First port of call though, inductor.

Wow what a post... sorry, I tend to just ramble with equations, detail and idea's. Don't be put off, skip most of it until you see something that makes sense lol

Edit: just remember you mentioend the adjust pot got hot: its only 1k, so it'll be disappating 144mW. Try using 4.7k or 10k.

 

[formant84]

Wow that's a lot of good info!

I really need to start looking harder at datasheets and learning how to calculate things myself. My eyes tend to glaze over when I get to that section of the datasheet with all the equations and graphs on it!

I did manage to desolder the inductor from the ebay board and attach it with wires, but this didn't seem to help. When I get a chance this weekend I'll try switching in a 330uh inductor and/or a bigger pot to see if that helps, but it is really starting to look like it might be the layout. I've replaced the switcher once because I did damage the first one, but I'll replace it again after making the changes to the inductor and pot to see if that makes a difference. If not I guess it's time for a redesign! I think this time I'll just more or less copy the layout of the ebay board for this section of my circuit.

I can't thank you enough for your help, on my own I'd probably be banging the board against a rock by now, or praying to some arcane electronics god.

 

[Blueteeth]

No problem sir, sorry for the disjointed/fragmented post, I was just thinking outloud (or on keyboard?) rather than organize things in a coherent manner

SMPS's can be a sod to develop and debug, not all though.. some I've built terribly, with scrap parts on a stripboard and they worked like a charm, others, on a manufactured two-layer board, were real fussy about where parts were, and input/output capacitance. I'm sorry to say that the 'simple switchers' were in the category of 'annoying' as far as I remember last time I used them. Like you, I spent much time desoldering, and resoldering parts until eventually the board I had was unsalvagable. I'm hardly a seasoned veteran with these things, but I've known folks who have had a lot of experience giving up on SMPS's before.

If you try everything, and decide on a new layout, I can try to dig out some old LM2577's I *think* I have, and knock up a test board. Although if you're going for a new board/design, perhaps we can find some cheaper, smaller boost converts that will do the job, and be more reliable/less fussy

Good luck mate, and let us know how things go!

 

[formant84]

SUCCESS!!!

All that messing around and it turns out it was the pot!

You'll hate this because it's wrong, but the only pots I had of the same type larger than 1K were 50K and 1M so I threw in a 50K one,,, et VOILA!

60 Volts!!

One cry of alarm, a panicked unplugging, and a large pot adjustment later... Everything seems to be perfect!

Input voltage measures exactly 5v, output is nice and steady, nothing is getting warm at all, I think I blew the 200ma fuse on my multimeter earlier but with no load attached to the output the 10A setting says I'm using somewhere between 20-30ma.... I think I may be ready to populate the rest of my board!

Are there any potential downsides to using this overly large pot do you think?

Could it affect the current flow in the circuit?

Thanks once again, I hope one day I know enough to help other people the way you have helped me!

 

[Blueteeth]

oh wow lol

I feel dumb now talking about inductor saturation >.< (although thats the usual port of call with these things).

I can't see any problem with using a 50k pot. Not at all!

Often the feedback resistors aren't chosen to be *too* big (>500k?), because they need to load the feedback line a tiny bit (the feedback pin has super high impedance, so it picks up noise easily - but with <100k resistance to ground, it shunts this noise). If its proving sensitive (output voltage changing now and then) nothing wrong with calculating fixed resistor values, but a pot is 'set and forget' so it works

As to 'why' that was screwing things up in the circuit... I'm stumped lol Unless it was fried... maybe the output spiked above 12V, meaning more current flowed thought it (24V output, across 1k = 0.576W, too much for a small trimpot) and it fried - leaving the feedback pin open. Who knows!

Glad you got it working man, don't forget to replace that fuse on the multimeter, ended up getting a pack of 50 for mine since I seem to blow it weekly. Good luck!

 

[formant84]

Yeah it's weird, the pot wasn't fried cos I could still adjust the voltage, typical it's always the last thing you think of though! I've now made myself late for work adding the micro and associated bits. I can plug it into USB and have it recognised now with all voltage and current readings as expected. All that's left is to add the DAC and the op amp, then I can upload the code and I'll finally have the MIDI to CV convertor I need to control my DIY synthesizer!

You sir are a legend.

I will pester you again one day, you can be sure of it!

 

[JimB]

Hang on a minute!

I think I know what the problem could be.
Have a look at my attachment.

There are different pin configurations for pots.
Check and compare your two pots to see if the one which gives problem is like the second one I have drawn.

JimB

 

[formant84]

Whaaaa? Whoever designed the one on the right is sick in the head!!

I'm in work now till late but I'll check it out in the morning and let you know

 

[formant84]

Nah, normal pot. must just have been too low a value!

 

[JimB]

OK, fair enough.

I am sure I had one somewhere with a weird pin-out, but cannot find it in my usual places.
Hey Ho!

Your converter is working, end of problem.

JimB