DIY Boost Converter: help me understand my failures please. / General Science and Electronics / Forums

Forums

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

« Previous topic | Next topic »

DIY Boost Converter: help me understand my failures please.

1 2 3 next

Move Thread

LAN_403

Sigurthr

Mon Jan 20 2014, 07:28AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

Hey everyone,
I recently dusted off an old project of mine: making a simple boost converter. Sounds simple enough, right? Yet, for some reason when I last tried it I blew through an entire stock of gate drive ICs! I put it on the shelf and didn't think about it for a year.

Well, yesterday I dusted it off and decided to give it a go again, this time with better feedback; I decided to use my arduino to be both the oscillator and the comparator. The code is simple: it uses the ADC to sample the feedback voltage divider's output and stops the oscillator when the desired voltage level is reached. I have a 100:1 voltage divider, and set the code so that oscillation stops when 350mV is detected. The oscillator is set to 64KHz at 50% duty cycle for 0-30V and then at 30V the pulse width is reduced until at 35V where it stops. To my surprise, it still blows up gate drive chips every time! I tried various sizes of gate resistor, but nothing helped. I got fed up and decided to just go discrete. Now, I have very little experience with discrete BJTs other than Class-A amplifiers and simple switching, so part of this was both an experiment and exercise in BJT basics.

At first I tried a simple NPN/PNP totem pole between the Arduino output and the MOSFET, but when I scoped it I noticed that a BJT Totem provides no voltage amplification, just current amplification, so I knew this would not do as more than 5V was needed to fully turn on the FET quickly. I then tried a simple single stage NPN switch, which provides both current and voltage amplification, provided the collector voltage is higher than the input voltage. It looked fine on the scope, but this inverts the signal, as using a pull-up resistor on the Collector creates a logic HIGH when input to the Base is LOW. Simple, add a duplicate stage to invert the signal again. This of course couldn't sink enough current to discharge the large Gate capacitance (Ciss), so I decided to put the NPN/PNP totem stage in between, as this can of course sink as well as source current. It worked great except that now the pulse height capped off at around 8V, and would rise no higher, regardless of the Vcc. I would love to know why this happens, if anyone can please explain or link to an explanation. Still, 8V is enough to fully turn on the FET, so it went with it.

It works! No dead BJTs, FETs, or uC's. I was able to play around with load resistors and found that at 2.5K ohms the voltage hovered nicely at 34.95V +-50mV without oscillating between ON and OFF states too rapidly.

While I was happy this was working, getting 15mA @ 35V out when putting in 350mA @ 12V in wasn't exactly shining efficiency! So, while the circuit was a success, it failed to meet any useable quota in terms of output. I did play around with setting the voltage limit and could get it to operate as high as 110V but at that point it wasn't able to overcome the 1mA draw of the feedback network. So, at worst I got 110mW out and at best I got 525mW out.

Here is the schematic of both versions:
Wprl

I like to think of this as a learning experience, and as such this has raised several questions that I just don't know how (or where to) find answers to:

1) Why did the more robust, much higher power gate drive ICs explode every single time, when wimpy BJTs handled it just fine. I scoped the entire circuit out, there was no nasty ringing or bad gate signals, etc. At least not anything that my 40MHz CRO could see.

2) Why does output of a NPN-PNP totem pole cap off at ~8V when Vcc was well more than that. I was expecting a few junction drops at most, not more than 4-8V! (I tried 16V Vcc too).

3) How do you increase the voltage and current capacity of a boost converter? I tried lowering the input resistance down to 1ohm, and while it did increase the current output considerably at lower voltages I still couldn't exceed 130V @ 2mA out even though input was all the way up to 16V @ 4.3Amps! I also tried various inductors of various sizes, and while none of mine are labeled and I don't have a LCR meter, I know many could handle over 10amps because of how large the cores are and how heavy gauge the windings were. I even tried MUR1560s for the diode, but no improvement was seen.

Thomas W

Mon Jan 20 2014, 08:12AM

Registered Member #3324 Joined: Sun Oct 17 2010, 06:57PM
Location:
Posts: 1276

I would try probing the current output, see if your triggering it at the wrong time.

try show a picture of the current waveform, the switching waveform and the output voltage waveform.

Sigurthr

Mon Jan 20 2014, 09:01AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

Unfortunately I don't have much in the way of current probing equipment. I have a single precision 1Ohm 100W resistor but the output from probing that will be gunked in the noise floor for this old scope. Aside from the 10R resistor in use the lowest value resistor I have on hand is 100R, and they're cheap 20% tolerance ones from the 1960s.

The gate waveform was a simple square wave, with a slightly rounded rising edge. At a small enough time base to see the rise time well I can't get the pulse in the center of my scope's screen, because it is such a small fraction of the overall period. Times like these I wish I had a DSO.

Output voltage waveform was pretty solid DC with standard triangular ripple at 64KHz. Ripple was about 18mV.

By triggering it at the wrong times do you mean using the wrong drive frequency? That's one variable I can't easily adjust - still dreaming of a signal generator. When I built the circuit the first time I was using an astable 555 for the oscillator, and running open-loop where I chose the load resistor to be enough to prevent voltage runaway from destroying the cap, diode, and FET. It was far from ideal. Back then it kept blowing the gate drive chips as well and I eventually settled on using a much smaller Ciss MOSFET and just driving the fet directly with the 555, which worked. I stuck to very low frequencies though since the 555 is of pretty limited output power. IIRC I was using 500Hz-5KHz drive.

If you really need pics I'll have to reassemble it all and scope it, but I'd prefer to not have to since workbench space and time is at a premium right now.

Shrad

Mon Jan 20 2014, 09:25AM

Registered Member #3215 Joined: Sun Sept 19 2010, 08:42PM
Location:
Posts: 780

I don't think voltage precision is the main goal here so any of your resistors would do the trick

Maybe just a small current transformer at the base would do, with a sufficient number of turns, so you have a reading and can superimpose the timings

Alex M

Mon Jan 20 2014, 10:30AM

Registered Member #3943 Joined: Sun Jun 12 2011, 05:24PM
Location: The Shire, UK
Posts: 552

Try adding a small resistance of around 10-22 ohms in series with the MOSFET's gate and then and a small capacitor of around 2.2nF across R5, this should take care of most peak triggering problems.

If you can reassemble it then scope around the bases of the totem pole, it sounds like maybe the bottom one (PNP) is not turning off completely which makes them act like a voltage divider.

Dr. Slack

Mon Jan 20 2014, 01:18PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

could you explain what R3 is supposed to be doing? It doesn't seem to do anything except increase the losses of the inductor, and decrease efficincy.

Sigurthr

Mon Jan 20 2014, 01:37PM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

Sure, R3 is just to limit input current, as without it the fault protection on my bench supply (5.25A) was tripping.

Wastrel

Mon Jan 20 2014, 02:19PM

Registered Member #4095 Joined: Thu Sept 15 2011, 03:19PM
Location: England.
Posts: 122

I don't see any details about L. It sounds to me like it's too small for the switching rate, whatever else might be wrong with the circuit.

Dr. Slack

Mon Jan 20 2014, 03:37PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Sigurthr wrote ...

Sure, R3 is just to limit input current, as without it the fault protection on my bench supply (5.25A) was tripping.

Well in that case, your peak inducrot current is not being well specified.

Either
a) you should calculate the peak current from the inductance and on time, to limit it
b) you should have an input supply capacitor to source the required peak current, if you do intend a current that big
c) you should measure the inductor current in a *small* resistor and use it to terminate the charge pulse.

Using R3 to drop almost all of your input power is a receipe for uber low efficienxy.

I have a confession to make. A good few decades ago when I wound my first transformer, I used nichrome rather than copper for the primary, as I was nervous about the consequences of connecting such a low resistance across the mains. Funnily enough, the efficncy was so low as to be barely measureable.

Sigurthr

Tue Jan 21 2014, 02:51AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

Thank you for all the replies everyone!

I rebuilt the circuit exactly as I did last time (woo breadboard) and took some shots of the scope for you all.

Dr. Slack: I sympathize with your confession, hooking stuff up to mains is scary when you're new! Thankfully I didn't mess with direct mains until I had my variac, so I had a layer of protection there (soft start).

I wish I could calculate things properly for this, but I don't have a LCR meter, and these inductors are all unknown value, salvaged from various junk. When I'm building something that HAS to work (usually for later resale) I order in new parts and thus know exactly what I'm working with, but when I'm just experimenting and learning new things I have to work with whatever I can get my hands on free.

About R3; it was only dropping 3.5V (1.25Watts) that day, and now with the circuit rebuilt it is only dropping 1.1V (121mW) as the input power is: 13.8V @ 110mA. The change is likely due to the large tolerances on components. I made sure I used the same inductor, filter cap, and feedback divider though.

1 2 3 next

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