0

I am trying to setup a DC-DC boost converter circuit that can convert 50-60 V DC to 86.4 V at the most current I can achieve a reasonable efficiency of say 90%. Mostly for the learning curve but my practical use is charging a 56 parallel by 24 series = 1344 cell LiFePO4 battery bank from a 60 V DC supply.

I have been using DC-DC boost converters from eBay to charge it so far but they are not that good as they are trimpot controlled and I would like to be able to control the system from an ATmega microcontroller not a screwdriver.

See linked thread that has a picture of the modules in question: need info about dc-dc boost converter

After blowing up a few modules, I dismantled them to understand their main components and principle design. The best one I had was using FDH055N15A MOSFET as the switch. The module linked above is using HY3912W MOSFET as the switch, both of which I have been getting reasonably good results of about 95% efficiency with during my experiments only I don't have any left now as I have blown all I had to hand by overheating them.

I started experimenting with other MOSFETs like IXTQ76N25T and FQP30N06L simply because I had some to hand, but my results were not great; I was getting more like 60% efficiency and struggling to output more than a few amperes.

I read a lot about Rds(on) value of the MOSFET and the lower the value the better, but I'm not sure how relevant it is and how big a deal or what effects it would be if it were to high or even what to high actually is. The two MOSFETs that have given good results have had about 6 mΩ Rds(on), I have also had reasonable results a FQP30N06L with 35 mΩ so I don't really understand why some MOSFETs don't want to play ball.

At the moment I have a TIP122 Darlington NPN transistor in my circuit and I am struggling to get any kind of usable current output.

Current setup with the TIP122 NPN is actually outputting to a 14 cell Nissan Leaf battery next to me that is 57.4V not 86.4V charge but the theory should be the same.

I have 0.503 A input at 12 V, 0.054 A output at 52.89 V, 6 W in, 2.85 W out, so about 47 % efficiency.

enter image description here

This scope probe shows the top scope connected to the transistor base gate, and the bottom probe connected between the inductor output and the diode, so the rising edge of the top scope is the signal from an ATmega328P turning on the transistor, at the same time the output pulls ground.

The gate is high for around 60 microseconds. As it falls low there is a big positive spike on the output. That is the charge from the inductor, right?

During the time the spike is high (above 52.89 V) current flows out to the battery as charge.

As the inductor runs out of energy the voltage falls back to ground rapidly and seems to stay at ground for a period of time before jumping back up to VIN. I wonder if this has to do with inductance. I don't know much about and would appreciate some explanations as to the patterns I see.

There is a large period of time that the transistor is off, leaving the output signal at 12 V, before the next spike as the transistor turns back on. If I shorten this time period trying to obtain more current throughput, I end up losing efficiency. Is there any explanation available as to what is happening? On the scope probe there is not much happening but in reality there must be something to do with inductance otherwise there should not be any efficiency drop for making use of that time.

Finally a picture of the actual circuit:

enter image description here

Any theory as to what I am not doing correct would be appreciated. I don't know much about inductance.

Edit:

Schematic diagram of current circuit:

enter image description here

I have changed the inductor for a smaller one and I am now getting far better results. This is the new scope shot that is achieving 95% efficiency 100W 55V to 86V.

enter image description here

The diode I am using is a "Plastic Encapsulated Silicon Rectifier Diodes 1KV 10A" type 10A10.

Current testing measurements are as follows, ICR1=TOP and maxPWM is the PWM duty between 0-255 running from an ATMEGA328 clocked at 8 MHz.

enter image description here

Tim Williams
  • 22,874
  • 1
  • 20
  • 71
Jay Dee
  • 319
  • 1
  • 11
  • A wall of text, blurry oscilloscope picture and no schematic makes for a very hard read. Please write to the best of your ability. Sent a start with capital letter. It looks like you are trying to implement an SMPS with loose transistors on long wires. This will cause a lot of issues. You need a PCB with solid ground plane and strong decoupling for just about any SMPS. Have you tried to simulate it? – winny Jul 17 '22 at 10:19
  • *Sentences. Damn you autocorrect. – winny Jul 17 '22 at 10:28
  • my schematic is as basic as is required to run, it is the example schematic from falstad circuit simulator Circuits / Power / Boost Converter. with the only difference being the values like the inductor inductance is unnown as i just salvaged it from a broken boost converter, i have tried lots of different inductors from all sorts of things like transformers to big inductors from welders all unknown inductance values, just to see what difference it made. – Jay Dee Jul 17 '22 at 11:28
  • sorry my spelling skills are poor, the camera on my phone is caked, there just what i have to work with right now, that's life, we live and learn and just plod on, but thanks for the text correction edits – Jay Dee Jul 17 '22 at 11:28
  • and yes i plan to build the circuit on a PCB but im currently trying to understand the principles of inductance, the more i understand now, the less circuit boards i will waist getting it wrong :-) – Jay Dee Jul 17 '22 at 11:31
  • @JayDee: [Have a look here.](https://josepheoff.github.io/posts/leddriver) That is a bare bones booster driven by an Arduino (ATmega328P.) The text explains a bit about the waveforms and the charge time for the inductor. [Take a look at the wikipedia page about inductors while you are at it.](https://en.wikipedia.org/wiki/Inductor) – JRE Jul 17 '22 at 11:35
  • _”and yes i plan to build the circuit on a PCB but im currently trying to understand the principles of inductance”_ The problems you will face from no PCB will probably hinder you to get there. Simulate it first, then select components, then draw a schematic+PCB, assemble and then start to measure. – winny Jul 17 '22 at 13:21
  • 1
    I see at least 200nH of stray wiring inductance between transistor and diode. Do you know what effect that will have? And, what ways could you use to evaluate it? – Tim Williams Jul 17 '22 at 18:06
  • @TimWilliams, can you please explain in more detail, im not sure i understand what you mean. i have made big imprisonments since yesterday and im just drawing up a schematic for what i have done before i forget it all. – Jay Dee Jul 18 '22 at 07:21
  • 1
    In your previous layout, you had about 20cm of lead length in this loop. This can be modeled, in your current schematic, by adding the inductance in series with the TIP142, diode, or (VOUT) 470uF cap: hence, switching loop inductance. Without a photo of the bottom side, I cannot determine the same quantity by inspection of the current version, but improvement seems likely. The fundamental to learn, is that wire length has inductance, for the sheer fact that it's traveling some distance through space. – Tim Williams Jul 18 '22 at 21:37
  • Regarding the waveform, the "double tap" at the end as the inductor discharges, is due to recovery time of the diode. What type is it, anyway? You should choose a fast-recovery or schottky type to reduce this effect, and then efficient operation at high frequencies is feasible. Also, remove those 47uF on the ADC inputs, you need FAST feedback to prevent excess current building in the inductor -- which, consider adding a current sense (using a diff amp or Hall effect sensor) to the inductor so you can control this, particularly during startup/fault conditions. – Tim Williams Jul 18 '22 at 21:48
  • @TimWilliams thanks, is there any way i can measure or know the specification / inductance of a particular inductor in my hand, i am happy with this one i found in a Dewalt Drill battery charger, the problem is, if i wanted any more the same i wouldn't have a clue what to order from where. – Jay Dee Jul 18 '22 at 21:49
  • e.g. https://passive-components.eu/how-to-choose-the-right-inductor-for-dc-dc-buck-applications/ Also by measuring the current, you can observe the triangular current waveform, and thus calculate inductance from applied voltage, time, and ramp rate. V = L dI/dt – Tim Williams Jul 18 '22 at 21:52
  • @TimWilliams i have added my current calculations, if you are able to help me understand the chosen inductor – Jay Dee Jul 18 '22 at 22:05
  • 1
    @Jay Dee Do you know the basics for capacitors ,inductors and diodes? – Miss Mulan Jul 18 '22 at 22:46
  • not sure how to answer that question, i no very little about inductance, until now i have never found myself in a position where i needed to know. i think i know the basics of capacitors and diodes. i have an aeronautical engineering background and a computer programming degree, electronics is more of a hobby level like to be able to make things work. i generally program very large milling machines to cut up metal – Jay Dee Jul 18 '22 at 22:55
  • Two things, “unknown inductance” won’t do. You need to know the inductance and saturation current before proceeding. As for the diode “ Plastic Encapsulated Silicon Rectifier Diodes 1KV 10A type 10A10” that is a mains rectifier diode and have no place in an SMPS. You need fast recovery ones or just maybe 200 V Schottky would do. – winny Jul 21 '22 at 21:36
  • thanks, i was planing to change the diode for one that bolts to the heat sink with the MOSFET as it gets to hot. as for inductance testing, i haven't worked out how yet, would i be wrong in saying that i can see on the scope when it saturates as the voltage starts raising up? – Jay Dee Jul 22 '22 at 05:48
  • **SAFETY** You are hopefully aware that the voltages being used are potentilly lethal (pun noted) - both input and output. Some people may manage to contact metal at these voltages numerous times with no apparent long term harm. For some it's fatal. Regardless of discussions re ELV and standards and ... - just be aware it's lethal. – Russell McMahon Jul 24 '22 at 12:14

1 Answers1

1

Having spent all week tying to rebuild my burnt-out boost converter, the biggest problem I faced was heat in the MOSFET under any real load.

After days of trying to make it work I found my problem to be the on-resistance of the MOSFET.

The microcontroller pulls the gate to 5 V and VGS(th) is 3-5 V, so the MOSFET is not turning fully on and it becomes a resistor; if I give it a 6.5 V+ pulse the heat goes away.

ocrdu
  • 8,705
  • 21
  • 30
  • 42
Jay Dee
  • 319
  • 1
  • 11