Boost converter problems

Hey gang,

I've been fighting with this boost converter for the G9-51X for quite a while now. The first design used a cheapy mc34063, but I was not getting the needed efficiency. It's an old gal, so you can't expect too much, I guess. I 've now tried a new layout with a MAX1771 controller, plus I switched to some SMD components on a professional proto PCB, to save both space, and to cut down the trace lengths in some critical areas.

The problem I'm having now is that the mosfet seems to be stuck ON when in the circuit, or the controller is PWMing closer to ON, than OFF. This is causing the converter to just start to fry. Luckily my psu has a current limit feature that is cutting the +24v power  to about 50% duty cycle, so my pcb or components don't fry.

Somebody who knows more about this type of tech, please take a look and tell me where I've knuckleheaded this design.

PCB layout:

gemini86 said:

Hey gang,

I've been fighting with this boost converter for the G9-51X for quite a while now. The first design used a cheapy mc34063, but I was not getting the needed efficiency. It's an old gal, so you can't expect too much, I guess. I 've now tried a new layout with a MAX1771 controller, plus I switched to some SMD components on a professional proto PCB, to save both space, and to cut down the trace lengths in some critical areas.

The problem I'm having now is that the mosfet seems to be stuck ON when in the circuit, or the controller is PWMing closer to ON, than OFF. This is causing the converter to just start to fry. Luckily my psu has a current limit feature that is cutting the +24v power  to about 50% duty cycle, so my pcb or components don't fry.

Somebody who knows more about this type of tech, please take a look and tell me where I've knuckleheaded this design.

Click to expand...

What's the input voltage? What's the desired output voltage?

-a

The input looks like 15-20V input , output is adjustable and if ref V is 1.5V that trim will make roughly 150-225V output range.

Obvious things to look at if not turning off is  V at the FB pin, of course be careful probing around.

I'd be tempted to have a cap across that large impedance divider, but I am not familiar with the part so a little lag there may not be an issue. 

JR

Sorry, forgot some details.

24V input, hence the 12V reg going to the controller (max 16V input there). Output around 250V.

JR;
Problem is that with the FET in circuit, it will destroy itself if not for my current limiting psu, which is pulsing the input at around 50% duty cycle. Therefor, I can't really get any accurate readings on voltages. The output is only charging up to about 10V due to the psu shutting down. I did try hooking up a different supply that doesn't have an auto-shutdown, but things started getting really hot, really fast. I'll try to figure out some way to get some good readings without destroying my 30 dollar 1X1inch prototype pcb...

The datasheet did suggest using a small cap to bypass the 1.5M resistor, tried 100pF, but I might try 47pF.
http://datasheets.maxim-ic.com/en/ds/MAX1771.pdf

A cap in front of your inductor and Vreg (C1 from datasheet) might catch your reading issues from input pulsing.

Not very good at this sharing info thing, am I?

Here's a more complete schematic. I have input filtering on the main board...it says 1000uF, but it's really 100uF, so not a big enough tank to calm the pulses, I'm afraid.

I'll try to get something worked out to get a voltage reading on the FB pin. I've never used this chip before, but I would have thought it would be more stable than the old mc34063...

just took a poke at pin 3 (FB), and it's floating... no connection to the feedback network. "well tharz yur prerblem rut thar..." as my old mechanic buddy used to say.

I'll try reflowing some pins, wish me luck, it's mighty cramped in there.

Well, that didn't do anything, at first. But then after about 20 seconds of pulsing on and off, it finally started to stabilize and I could see +24 on the input, 12v at the IC, and about 200 volts on the output, until only a couple seconds later it started pulsing again and everything went back to crap status.

The FB pin is still good though, double checked that. I wasn't running long enough for me to get to the FB pin and see what voltage was there, but currently it's at .068V steady. I cannot find what I should be seeing in anywhere in the datasheet. The internal diagram shows a 50mV reference to the FB's comparitor?

I'm beginning to think my inductor is too small. I've built a very similar circuit before with a 100uH 2A part from radioshack, and a 555 timer. It worked, but this is what it says in the datasheet regarding selecting L:

Determining the Inductor (L)
Practical inductor values range from 10μH to 300μH.
22μH is a good choice for most applications. In applications
with large input/output differentials, the IC’s
output current capability will be much less when the
inductance value is too low, because the IC will always
operate in discontinuous mode. If the inductor value
is too low, the current will ramp up to a high level before
the current-limit comparator can turn off the switch.
The minimum on-time for the switch (tON(min)) is
approximately 2μs; select an inductor that allows the current
to ramp up to ILIM.
The standard operating circuits use a 22μH inductor.
If a different inductance value is desired, select L such
that:

      VIN(max) x 2μs
L ≥ —————----—--
              ILIM

Larger inductance values tend to increase the start-up
time slightly, while smaller inductance values allow the
coil current to ramp up to higher levels before the
switch turns off, increasing the ripple at light loads.

Click to expand...

First, I don't know what ILIM is. I cannot complete the equation, but probably mostly because I really suck at math. In my earlier mc34063 design, I used a 220uH part. It worked, so maybe I need to go bigger here, and that's what's causing the excessive low voltage current draw.

To reinforce what Harpo suggested, you need a low impedance at HF cap closer to the FET switch.

There are app notes that should describe how to size inductor for load.  That regulator is a little odd, in that it uses fixed time intervals rather than fixed frequency to modulate duty cycle, and it has the parallel current limiting so it isn't immediately obvious what is going on.

Dies that circuit need a soft start, to deal with charging large caps up to hundreds of volts?  I guess the current limit will serve as a sort of soft start.

JR

Added 100uF to the little board from +24 to gnd, could remove the 100uH inductor because I need a hot air gun, so I just paralleled 220uH across it making about 320uH. This has worked. I have a fairly steady 250v output, 1.6V on the FB pin, about 16mV accross Rsense and the converter is working at roughly 80% efficiency, which is less than I'd hoped but it doesn't seem to be as stable as it could be. I tried removing the 100uF from the little board with no change, so I think the main problem was the low inductance. I'll leave it on for a bit to see how things settle. I may try upping the cap across the 1.5M resistor in the feedback network to see if that helps stabilize it.

Thanks for everybody's help. It's mucho appreciated. This forum is great!

I was thinking more like a .1uF cap, just to be a low Z at HF.

JR

ok that makes sense now...

I'll give that a try, there's still room left for it.

Changed 1.5M bypass cap to 220pF, added .1uF right on board accros the +24v rail and ground.

Circuit is currently stable and running at 81.6% efficiency. At idle, that puts the +24V rail at 153mA. I'm willing to live with that. I'm starting to think the recycled trim pot may be causing some problems with efficiency and stability. I wish our local electronics shop hadn't closed, or I'd just go get a new one, but all I have now is RATSHACK. I have some more testing to do and a some layout changes to make. But all in all, I'm pretty happy with this. Maybe getting the feedback loop to be shorter/smaller can help. I want this to be a DIY project, so if I do too much smd, some people may not be able to do it. (requires a very steady hand and keen squinting skills)

Anyway, suggestions and advice openly and graciously welcomed! and thanks again to all who chimed in! I owe you all a nice scotch.

gemini86 said:

... the recycled trim pot may be causing some problems ...

Click to expand...

Maybe short out/get rid of the trimmer. With R502 1.5M your R503 10K alone will set it to about 225V, a 11K will set it to about 205V or a 12K will set it to about 190V on paper. Lesser parts, cost and trouble. YMMV.

That's not a bad idea at all. Especially with a tube project like this, a little variation won't hurt a thing.

Got rid of the trim, efficiency is up to 87.6% now. Very big thanks to you, Harpo for pointing that out.

I lim will relate to the peak fet current and the related voltage level across the sense resistor that triggers the controller - that info should be in the datasheet.  The peak current and the L and the duty cycle are related.  Have you checked if the choice of fet is appropriate - this gets in to the trade off between on-loss and switching loss, but it may be worth trying to find a somewhat similar fet that has different on-loss and gate charge and perhaps thermal resistance.  I'd add a smt cap to minimise the area between Vin and gnd, and another between Vout and gnd - scratch off the resist to place it with the minimum loop - function before beauty.  Have you checked what the loss is in the inductor, and its saturation characteristic, and what other models are suitable?  Best to prepare a summary document that identifies all the design info.

trobbins said:

I lim will relate to the peak fet current and the related voltage level across the sense resistor that triggers the controller - that info should be in the datasheet.  The peak current and the L and the duty cycle are related.  Have you checked if the choice of fet is appropriate - this gets in to the trade off between on-loss and switching loss, but it may be worth trying to find a somewhat similar fet that has different on-loss and gate charge and perhaps thermal resistance.  I'd add a smt cap to minimise the area between Vin and gnd, and another between Vout and gnd - scratch off the resist to place it with the minimum loop - function before beauty.  Have you checked what the loss is in the inductor, and its saturation characteristic, and what other models are suitable?  Best to prepare a summary document that identifies all the design info.

Click to expand...

A lot to chew on there, trobbins.

I'll have to do some poking around the datasheets to identify the important bits. The FET I chose was simply one I found in a proven layout and design that could achieve about 86% efficiency. It looks like I need to find a FET with about a 300V VDS, Qg<50nC and as low an Rds(on) as possible (>100mOhms) but I think that Rds(on) in this case isn't as crucial, since we're only switching about 150mA through the fet.

Any recommendations?

What do we think about this?

FDPF14N30
Vds = 300V
Rds(on) = .24ohms  (typical)
Qg = 18nC (typical)

and it's just over a buck...