Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Buck converter questions (integrated inductor?)

Status
Not open for further replies.
Hello again,

Ok good, if you went over the calculations and got something similar then it should be ok. The larger inductor value will mean lower ripple and less strain on the output cap too. With only 70ma output it's not hard to find a 680uH inductor, but at 70 amps it's an entirely different story. I was trying to find one that would work with a 40 amp converter today and they sure are expensive when you get that high up in current level.
Buck converters like this work with a fairly wide range of inductances anyway, although the design guides like to make it sound critical :)
 
Last edited:
I got my parts yesterday and i am finishing the PCB in Eagle. Man that SMD diode is microscopic... Would be nice if soldering wasn't critical, Ah!
So i guess soon i'll know if my MG photoresist kit is able to able to correctly etch the 0.005" traces passing between the diode pads. And if i can solder the components correctly... i'll practice a bit first. At least the rest of the board has fatter traces so that limits the possible overetching / bridging problems.

Once i've done it succesfully, i'll be using SMPS way more often in my projects rather than LM78xxs. It'll spare battery life or make a range of wall warts compatible... nice :)
 
Hello again,

Sounds good. let us know how it all works out ok? I'd like to hear more about this.

I've been fooling around with the MC34063 controller IC and it's good up to 1.5 amps too, but after i ordered a couple i realized it's not a pure PWM chip like most other controllers, instead it uses a 'digital' control technique which means higher output ripple (yuck). All the simulations i did verify this fact too, unfortunately. A bigger choke would do it, but i was planning on a 40 amp output and a big choke at that current would be quite expensive.
Maybe i'll just play around with it a bit and then move to a more conventional regulation scheme.

Oh yeah, once you get hooked on the switchmode power supplies you'll choke to go back to the 78xx type regulators, except for low current apps. The linear regulators eat up lots of power and get nice and hot while the switchers run nice and cool. I still use the 78L05 ic's for very low current apps like 5ma, 10ma, but i cant even remember the last time i used a 7805 for example. I did build a special LM317 power supply but that was to be used just for testing other things as it has a nice clean solid output.

Keep us updated...
 
Last edited:
A bigger choke would do it, but i was planning on a 40 amp output and a big choke at that current would be quite expensive.

Have you considered a multiphase buck to get your 40A?
 
Hello,

Yes, and i may end up going that route anyway considering even smaller chokes with that rating are not too cheap either.
 
Hello again,

Well the chips are cheap it's the inductor that gets pricey at those current levels, and several small inductors are not that cheap either.
 
The MC34063 is a great little chip. As for the larger current ripple, you can fix that, there is a Tc cap that controls the switching time period so you can just use a smaller cap to increase switching frequency. Once that cap gets below a particular value the IC will oscillate based on inductor characteristics and the 1.2v Vref comparator. With larger values of the TC cap it slows the switching frequency so you get better conversion efficiency due to lower freq and less switching losses.

The big problem using a 34063 for your 40A app would be the 0.3v overcurrent sense requirement... That would need a 40*40*0.3 or 480W resistor!
 
Hi there MrRB,

Thanks for the comments. I have just starting looking at this chip and so far i dont really like it. I considered going up to 100kHz but that still doesnt seem to quite do it either.
As to the overcurrent sense, i believe that power is still equal to voltage times current (small chuckle) so that would be 12 watts, but now that you mention it that is kinda high too so i considered other ways to limit current.


Did you mean that when you lower the cap lower than the lowest possible data sheet value that the circuit begins to work on some other principle?
 
Haha why the heck did I type that? It was at the end of a long days work and I should have just turned the PC off and went to bed. I woke up this morning and thought "What did I say?? :eek:" and quickly came back to edit it before anyone saw... :D

Yep it's E*I of course which is still 40*0.3 or 12W which needs probably a 20W or 25W resistor and maybe some air cooling too unless you want it to cook.

Re the Tc cap value, what happens is that the osc runs at a freq determined by the cap, so if you are concerned about freq (ie current ripple) you can use a smaller cap to set the higher frequency you want. When output voltage is below reg it runs at about 80% on duty, but when regulating the off period is the same but on period is delayed until Vout<=Vref so it reduces on period but at the same overall freq. With the right (ie small enough) cap value it will ride the inductor characteristics and work very nicely.

But if the cap value is too large it forces an unnaturally low frequency so you will get way too much current ripple and Vout ripple as it acts more like PWM into a cap than like an inductor tuned buck. When you said you had "yuk" output ripple this common problem is what I thought you were experiencing.

The 34063 is a good little chip, and you can really set the freq (ripple) you want by the Tc cap.

Re the current limiting you can use an op-amp etc to pull the Ipk pin low so you can workaround the 0.3v issue.

What Vin and Vout do you have that is giving you bad current ripple even at 100kHz?
 
Hello again,

Well any reasonable input and output, such as 20v input and 9v output. What happens is that the regulation does not occur on a cycle by cycle basis like the way a 'normal' PWM controller does. Instead, it follows the inductor current level. What happens is that the comparator can turn ON the output switch, but only the oscillator can turn it OFF, so what this means is that sometimes the inductor charges up (current level) higher than needed so once the osc turns off the output switch the output voltage actually still INCREASES even with the output switch off. What this extra increase means is that the control circuit *skips* some control cycles (at 100kHz period each) so the output ends up looking more like it is being controlled by a 10kHz control PWM rather than a 100kHz control PWM (10 times slower). This apparent slow down means of course that the inductor looks 10 times smaller than it is. This leads to fairly high ripple like 0.200 volts when a regular PWM would produce only 0.020 volts ripple for example with the same L and C values. Increasing L makes it look better, but that's 10 times higher than it needs to be with a regular PWM type controller.

What happened was the initial simulation was done with a commercial simulator and that's where the high ripple started to show up and it was almost too hard to believe, so i went ahead and worked out the calculations to make up a custom simulation program and it turned out to produce the same ripple so i knew something really was strange about this control method and it wasnt just the simulator program. I know this chips control method isnt new, but it looks like it's just not as good as a real PWM control scheme. With real PWM i get very low ripple with reasonable values of L and C always.

BTW, the data sheet even shows somewhat high ripple too for the buck configuration. Keep in mind that i just started looking at this chip too so havent found a workaround yet, so if you know some tricks for this particular chip it would be nice to hear about them. As far as increasing the frequency with lower CT, i think i am approaching the recommended lower limit as the max frequency i think they say is 100kHz isnt it? I'll have to check the data sheet again. Im not sure what the internal limiting factor is yet either. I think once you get below 100pf the on to off ratio becomes too close together so the max output duty cycle would approach 50 percent, which isnt good enough. I think the oscillator would have to run up as high as 500kHz to get low ripple similar to a 50kHz regular PWM.
I havent looked into any tricks to get the cap to charge or discharge faster yet either though.
 
Last edited:
It sounds like your load was too light? At light loads it will skip some on cycles as needed to maintain voltage regulation, similar to discontinuous mode in other chips. It also could be that you are trying to run it too fast, if Tc is too small you will get multiple osc cycles for the inductor tuned freq.

Ideally the on cycle is delayed by the comparator until Vout drops below Vref, then it will turn on only for the reaminder (less than 1 on cycle), and be turned off by the oscillator. When it turns off your inductor current will start decaying, so combined with appropriate load, shortly after turnoff the output voltage should be dropping. So for any decent load it should oscillate somewhere around the natural inductor properties, provided you don't have a value of Tc that is way out of whack.

Try using a larger TC value, so for your given load it should be running at every osc cycle or skipping every second osc cycle or so.

These are a great little chip for simple apps, either using the internal switch or just tacking on a PNP on to get a couple of amps output. If you really want something more sophisticated this tiny cheap little chip might be the wrong choice. :)

You can reduce V ripple a lot, at the cost of a little V regulation by taking the voltage sense from the first cap in your output PI filter instead of the final output. You can increase the duty cycle a little from the typical 80% max, by putting a resistor in parallel with Tc, as Tc is charged at 35uA and discharged at 220uA the resistor slows charging and hastens discharging.
 
Hello again,


Well reducing the TC cap will make the entire chip run slower, which i really didnt want. I was hoping to get operation at 100kHz but apparently it's not that simple with this chip. I could see it being ok in apps that dont need too clean of an output though.
I'll have to play around with it a bit more i guess.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top