Author Topic: MC34063 high voltage dc-dc boost converter  (Read 21440 times)

0 Members and 1 Guest are viewing this topic.

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
MC34063 high voltage dc-dc boost converter
« on: December 30, 2019, 03:41:21 pm »
Greetings everyone. I've been mulling over the idea of building a dc-dc boost converter that outputs some 300V to power tube amps. I reckon it should be easy to build one capable of powering a two 12AX7 preamp, but I wonder if I can build something with enough juice for an EL84 single ended amp, or even a push pull with two EL84's. This would be used for the plate and screen voltages, I would deal with the heaters separately.
I've already done some research and found a few chips that might do the trick, like the MT3608, the LM3478 or the MAX1771, but because I already have some MC34063 lying around, I decided to try this ratio extender circuit: https://www.changpuak.ch/electronics/High_Voltage_Power_Supply_MC34063.php

So far I have only simulated the circuit, and to my surprise, the simulation throws very promising results. (See attachment 1 for schematic)
The guy on the above link who built those boards to power nixie tubes got pretty terrible efficiency figures at 30-40% with this circuit, and the simulation agrees, but then I swapped the switching transistor (FJN13009) for a mosfet (IRF740) and the efficiency (again, in LTSpice) skyrocketed to a 85%. It also seems to make R7 unnecessary, I guess because the mosfet's gate current is negligible so it's not loading pin 2 of the IC, but is this a result that makes sense? Does the mosfet do a better job at staying away from the linear zone hence dissipating less power? Or is this simply an artifact of the simulation?

Perhaps even more impressive, or at least it was surprising to me, is that the simulation is perfectly fine outputing 200mA at 300V for 60W at 85% efficiency. According to my research, the MC34063 would exceed it's 1.5A peak current limit at some 50mA in the output at 300V, so 200mA seems way above spec. I will know for sure once I build it, but I thought I would ask in case some of the local experts can help me understand all this stuff.

I've read that the MC34063 is old and doesn't support PWM. I guess it works at a fixed frequency and it switches the pulsing signal on and off, and that makes it a worse option than a proper PWM chip, but I don't understand why that may be a problem. Is it because it generates noise as the switching signal turns on and off? I have added a capacitance multiplier to filter out most of the ripple and it seems to work great in the sim, with only 15mV peak-to-peak ripple at 300V. That ripple seems to be at the low frequency (200Hz) at which the mc34063 is controlling the switching signal. The switching frequency itself, at some 20KHz, is completely gone after the capacitance multiplier

I'll be using a toroidal inductor for this build, it's [ulr="https://www.ebay.com/itm/3A-6A-Toroid-inductor-Coil-LM2596-Magnetic-Ring-Inductance-22UH-to-100UH/223484117366?ssPageName=STRK%3AMEBIDX%3AIT&var=522239495543&_trksid=p2060353.m2749.l2649"]one of these 3A 470uH coilshttps://www.digikey.es/product-detail/en/aavid-thermal-division-of-boyd-corporation/513102B02500G/HS346-ND/1216353.]. Does that seem adequate?

And this heatsink for the mosfet, which according to the simulation, draws 8W with a 200mA load at the output: [url]https://www.digikey.es/product-detail/en/aavid-thermal-division-of-boyd-corporation/513102B02500G/HS346-ND/1216353.
It's 11ÂșC/W so might be a bit too small, not sure.

Thoughts? Ideas? Criticism?

Thanks in advance
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: MC34063 high voltage dc-dc boost converter
« Reply #1 on: December 30, 2019, 03:58:26 pm »
*Cracks knuckles*

https://imgur.com/gallery/M1S0DbI

Enjoy :P

But yeah, what you're really looking for is at least UC3843, probably a current transformer to improve efficiency, and definitely a transformer or autoformer so you aren't beating the shit out of the switching parts.  Any transformer from a power supply of the same size will do -- they work just fine in forward or reverse.  A 120/240V input, 12V output, 50W supply say, will easily get you 300V at 50W from 12V supply.

You should have no need of active filtering on a properly designed switcher.  Passive filters are simple and effective.  The 34063's hysteretic behavior makes a huge mess.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: sgi199

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #2 on: December 30, 2019, 04:18:17 pm »
*Cracks knuckles*

https://imgur.com/gallery/M1S0DbI

Enjoy :P

But yeah, what you're really looking for is at least UC3843, probably a current transformer to improve efficiency, and definitely a transformer or autoformer so you aren't beating the shit out of the switching parts.  Any transformer from a power supply of the same size will do -- they work just fine in forward or reverse.  A 120/240V input, 12V output, 50W supply say, will easily get you 300V at 50W from 12V supply.

You should have no need of active filtering on a properly designed switcher.  Passive filters are simple and effective.  The 34063's hysteretic behavior makes a huge mess.

Tim

Thanks, Tim. I have a few things to google now, because I don't understand some of the things you say there.
I think it's time to crack open The Art of Electronics, chapter 9 and learn how these buggers work before I go on with the build. I should've done that already.

I was hoping to get away with not using a transformer, but if that's the right way and I won't need something specific, I'll do it
 

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #3 on: December 30, 2019, 04:39:25 pm »
*Cracks knuckles*

https://imgur.com/gallery/M1S0DbI

Enjoy :P

But yeah, what you're really looking for is at least UC3843, probably a current transformer to improve efficiency, and definitely a transformer or autoformer so you aren't beating the shit out of the switching parts.  Any transformer from a power supply of the same size will do -- they work just fine in forward or reverse.  A 120/240V input, 12V output, 50W supply say, will easily get you 300V at 50W from 12V supply.

You should have no need of active filtering on a properly designed switcher.  Passive filters are simple and effective.  The 34063's hysteretic behavior makes a huge mess.

Tim

Wow, I simulated the same circuit at a lower output voltage, higher output current and the efficiency got up to 95% with much less power consumption in the mosfet, so I think I see what you mean now about using a transformer.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: MC34063 high voltage dc-dc boost converter
« Reply #4 on: December 30, 2019, 07:36:34 pm »
Yep.  You don't usually see SMPS analysis in terms of reactance, but it generalizes just fine.  By driving a higher voltage ratio, or higher current ripple ratio, you draw proportionally more reactive power from the inductor, and that power flows into the transistor and out the diode (or however you like to assign the signs, it's AC).

So going for a say 10:1 ratio at 50W might take 10 times more reactive power than 1:1 (flyback type), and if that's at 100% ripple (DCM or BCM), then that's about 500VA reactive power, and you better have one hell of an inductor to handle that with low dissipation (Q >> 100?).  And the transistor and diode need to be sized for 500VA despite that you only get 50W from it.

Or more generally, for a given factor of range, you need not only components that handle the extra range, but to maintain efficiency you need factor squared better components.  And that's just conduction loss alone.

So having a modest range, pays.  A lot.

By using a transformer, you get all the ratio with none of the wasted capacity.  You get an additional wrinkle of leakage inductance (and also stray capacitance, at high voltages), but this isn't new as such, it was always present (stray inductance between the transistor, diode and cap); it's only a matter of correct transformer design, and snubbing if applicable.

Some more study highlights: current mode control.  Draw the switching waveforms (current ramps and voltage flats).  Commutation and 1st order parasitics if you like.  Derive the transfer function (voltage ratio in terms of D) using the inductor definition.  You get the latter non-consecutive two here, http://schmidt-walter-schaltnetzteile.de/smps_e/smps_e.html (i.e. not commutation), but I do suggest understanding how it works, before getting too accustomed to it. :-+

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: hsn93

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #5 on: December 30, 2019, 10:50:27 pm »
Yep.  You don't usually see SMPS analysis in terms of reactance, but it generalizes just fine.  By driving a higher voltage ratio, or higher current ripple ratio, you draw proportionally more reactive power from the inductor, and that power flows into the transistor and out the diode (or however you like to assign the signs, it's AC).

So going for a say 10:1 ratio at 50W might take 10 times more reactive power than 1:1 (flyback type), and if that's at 100% ripple (DCM or BCM), then that's about 500VA reactive power, and you better have one hell of an inductor to handle that with low dissipation (Q >> 100?).  And the transistor and diode need to be sized for 500VA despite that you only get 50W from it.

Or more generally, for a given factor of range, you need not only components that handle the extra range, but to maintain efficiency you need factor squared better components.  And that's just conduction loss alone.

So having a modest range, pays.  A lot.

By using a transformer, you get all the ratio with none of the wasted capacity.  You get an additional wrinkle of leakage inductance (and also stray capacitance, at high voltages), but this isn't new as such, it was always present (stray inductance between the transistor, diode and cap); it's only a matter of correct transformer design, and snubbing if applicable.

Some more study highlights: current mode control.  Draw the switching waveforms (current ramps and voltage flats).  Commutation and 1st order parasitics if you like.  Derive the transfer function (voltage ratio in terms of D) using the inductor definition.  You get the latter non-consecutive two here, http://schmidt-walter-schaltnetzteile.de/smps_e/smps_e.html (i.e. not commutation), but I do suggest understanding how it works, before getting too accustomed to it. :-+

Tim

Great stuff, as always, Tim. Thank you. I sure want to learn how this stuff works. I'm still a long way to go, but it should be lots of fun.
EDIT: And if I could somehow source the right transformer, I guess I could add a 6.3V - 12.6V output for the filaments too, which would be fantastic.
« Last Edit: December 30, 2019, 11:13:52 pm by dazz »
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 17429
  • Country: us
  • DavidH
Re: MC34063 high voltage dc-dc boost converter
« Reply #6 on: December 31, 2019, 02:11:01 am »
I've read that the MC34063 is old and doesn't support PWM. I guess it works at a fixed frequency and it switches the pulsing signal on and off, and that makes it a worse option than a proper PWM chip, but I don't understand why that may be a problem. Is it because it generates noise as the switching signal turns on and off?

The MC34063 is a "constant off-time" PWM switching regulator.  This means that switching frequency is variable which can be a problem in some applications however it has the advantage of not requiring frequency compensation.

Most problems people have with the MC34063 is interfacing its bipolar output to external devices.
 

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: MC34063 high voltage dc-dc boost converter
« Reply #7 on: December 31, 2019, 12:28:02 pm »
T3sl4co1l is somewhat infamous around these parts for his dislike of the '34063/33063/3063/etc., but they are enormously popular for several reasons: no frequency compensation required, easy to get working (particularly the buck, as no high-side driver is needed), and the hysteretic control scheme used has the best possible transient performance of any (no Nyquist limit here). All in all, very beginner-friendly and the major downsides like the relatively high voltage drop across the (non-boostrapped) internal bipolar switch and the variable switching frequency being more difficult to filter for EMI are the sorts of things beginners don't usually care about; once you gain some experience - and confidence, from not having something blow up as soon as power is applied, as is so often the case with switchers - then you can move on to current mode controllers driving external switches, etc.

However, you won't be getting 300W out of this chip... In fact, even 30W would be a stretch, given that the absolute max ratings are 40V and 1.5A peak, so even if those were continuous/average ratings the best you could hope for is 60W. These ICs - and any with an internal switch, really - also aren't the best choice for transformer isolated converters, but since adding a transformer to the mix makes the design much more difficult this also isn't really something that I'd recommend a beginner at SMPS design attempt. Make a few non-isolated converters - buck, boost and flyback (or buck-boost) - first, then look into how adding a transformer can vastly improve things compared to, say, trying to get 300V out of a boost supplied by 12V (which is totally impractical, even if theoretically possible).
 

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #8 on: December 31, 2019, 01:04:29 pm »
Thanks guys. So I guess the simulation is giving me deceptive results in this case if the MC34063 won't do 60W. I have built a couple boost converters with this same chip, and one buck converter, all of them based on the circuits in the datasheet, then filtered the hell out of them with LC filters, and they worked ok at low current.
Anyway, I need to do a lot of reading on the subject and I'll also try building the circuit in the OP to see what it can do and get some practical results.

Then I'll look into adding a transformer.  Found this one that looks like it should do the trick, right? https://www.ebay.com/itm/12V-To-255V-300V-EC42-EC4045-Horizontal-High-Frequency-Transformer-Inverter/262721601791?hash=item3d2b6d80ff:g:rZwAAOSwYVpdpdSb
Seems a bit overkill at 350W though.
 

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #9 on: December 31, 2019, 08:05:44 pm »
These ICs - and any with an internal switch, really - also aren't the best choice for transformer isolated converters

Can I ask you why is that the case? Aren't this IC's meant to control the switching? Or do you mean IC's with an internal switching mosfet? Sorry, I'm afraid I'm missing something here
 

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: MC34063 high voltage dc-dc boost converter
« Reply #10 on: December 31, 2019, 08:15:18 pm »
These ICs - and any with an internal switch, really - also aren't the best choice for transformer isolated converters

Can I ask you why is that the case? Aren't this IC's meant to control the switching? Or do you mean IC's with an internal switching mosfet? Sorry, I'm afraid I'm missing something here

I was referring to regulator ICs - that is, a controller + internal switch - vs. controller ICs, which require an external switch. Regulator ICs are usually designed for non-isolated applications and, as a result, often only have a single pin for feedback. Controller ICs - which control an external switch, hence the name - typically bring out one or both error amplifier inputs as well as the output which makes it easier to design a compensation network and use an optocoupler for isolating the feedback pathway.

Another possible issue with regulator ICs is that is that any ringing or voltage overshoot from the transformer leakage inductance can disrupt the controller's operation (or even destroy it) by causing substrate currents to flow. Basically, once you need to handle more than 10-20W or so you should strongly consider using an external switch.

 

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #11 on: December 31, 2019, 08:28:51 pm »
OK, thanks. Frequency compensation is on my list of things to investigate. In a quick google search I learned it's about avoiding positive feedback/oscillations.
Just to be clear, when you say I need an external switch, do you mean the mosfet? I guess that's not it
 

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #12 on: December 31, 2019, 08:45:46 pm »
Oh, OK. I think I see what you mean now. The internal switches are not necessarily mosfets. So for example in the MC34063 the internal switch is that darlington pair at the SWC & SWE outputs, right?
 

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: MC34063 high voltage dc-dc boost converter
« Reply #13 on: December 31, 2019, 09:10:32 pm »
Oh, OK. I think I see what you mean now. The internal switches are not necessarily mosfets. So for example in the MC34063 the internal switch is that darlington pair at the SWC & SWE outputs, right?

Right, the switch in a switching power supply can be a MOSFET, BJT, SCR... heck, even a relay (see the "vibrator" type power supply used in early car radios). In this case, you are mostly correct - the actual switch in the MC34063 is Q1 while the driver for it is Q2. Note that the collector for driver Q2 is brought out to a separate pin to allow for more flexibility in driving Q1; most of the time you'll just connect the collectors together in a straight Darlington configuration, though.

 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: MC34063 high voltage dc-dc boost converter
« Reply #14 on: December 31, 2019, 09:35:50 pm »
Some of those old controllers were also useful (to a limited extent) alone, so could be considered regulators as well.  34063 is kinda like that, depending on how much of the output transistors you use.  TL494 has 200mA, 40V "uncommitted transistor" outputs, so can potentially deliver a couple watts by itself.

Or, the gate drivers often capable of several amperes (if in pulses), but the always-on output rather constrains direct application of them (you can't very well run a boost converter when there's a transistor firmly clamping the pin to VCC).

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #15 on: December 31, 2019, 09:50:17 pm »
Oh, OK. I think I see what you mean now. The internal switches are not necessarily mosfets. So for example in the MC34063 the internal switch is that darlington pair at the SWC & SWE outputs, right?

Right, the switch in a switching power supply can be a MOSFET, BJT, SCR... heck, even a relay (see the "vibrator" type power supply used in early car radios). In this case, you are mostly correct - the actual switch in the MC34063 is Q1 while the driver for it is Q2. Note that the collector for driver Q2 is brought out to a separate pin to allow for more flexibility in driving Q1; most of the time you'll just connect the collectors together in a straight Darlington configuration, though.

Doh! True, it's not a Darlington pair. And the fact that they're BJT's is an issue because it has higher output impedance and loads what's connected to it more than say, fets? Is that right?

Also, I just tried TI's Webench and performed a parametrical search. Apparently the boost topology is the simplest one (obviously), push-pull is the most efficient and flyback is a compromise between the two... well I really need to read the relevant chapter in The Art of Electronics to understand how those topologies work.

But one of the suggested IC's in webench looks like it could be a good option for my next design: the UC3844A. It's also one of the ones that Tim recomends in that imgur post. It's cheap, through hole and readily available
 

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #16 on: December 31, 2019, 09:57:13 pm »
But I can't seem to find a spice model for that UC3844A... dunno, maybe I've become too reliant on the simulation
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: MC34063 high voltage dc-dc boost converter
« Reply #17 on: January 01, 2020, 12:38:51 am »
https://forums.ni.com/t5/Multisim-and-Ultiboard/model-for-UC2842-or-UC3842/td-p/524298?profile.language=en

Don't worry about 44, it's the half duty cycle version, aimed at forward converters.

Do however use the 43; set the UVLO thresholds to say 5 and 6V in that case.
Code: [Select]
vlow vlow1 0 dc 5
vhigh vhigh1 0 dc 5.5

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: dazz

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #18 on: January 01, 2020, 12:41:17 am »
https://forums.ni.com/t5/Multisim-and-Ultiboard/model-for-UC2842-or-UC3842/td-p/524298?profile.language=en

Don't worry about 44, it's the half duty cycle version, aimed at forward converters.

Do however use the 43; set the UVLO thresholds to say 5 and 6V in that case.
Code: [Select]
vlow vlow1 0 dc 5
vhigh vhigh1 0 dc 5.5

Tim

Dude, you're awesome. Thanks so much
 

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #19 on: January 01, 2020, 12:14:33 pm »
I'm having a look at the UC3843 datasheet. Apparently it's maximum supply voltage is 30V and the maximum output current is 1A. Does that mean this IC can't do more than 30W output?
 

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: MC34063 high voltage dc-dc boost converter
« Reply #20 on: January 01, 2020, 12:28:11 pm »
I'm having a look at the UC3843 datasheet. Apparently it's maximum supply voltage is 30V and the maximum output current is 1A. Does that mean this IC can't do more than 30W output?

No, that (very rough estimate) only applies to *internal switch* regulator ICs, not *external switch* controller ICs. Note the distinctive terminology used, which applies when searching for parts on the usual distributors, btw.

 

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #21 on: January 01, 2020, 12:45:14 pm »
I'm having a look at the UC3843 datasheet. Apparently it's maximum supply voltage is 30V and the maximum output current is 1A. Does that mean this IC can't do more than 30W output?

No, that (very rough estimate) only applies to *internal switch* regulator ICs, not *external switch* controller ICs. Note the distinctive terminology used, which applies when searching for parts on the usual distributors, btw.

I see, thanks MS. Hopefully one day I'll be capable of building a full blown AC-DC SMPS with this chip, as I reckon that would provide the best efficiency. Baby steps though
 

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #22 on: January 01, 2020, 04:29:35 pm »
I'm a bit baffled by this. I'm simulating the flyback version of the MC34063 converter at the moment, and surprisingly it seems to be slightly less efficient (82%)
I swapped the mosfet for a IRFZ44N with Id=49A & Vds=55V that I think should be suitable for this application, but it's dissipating a whopping 10W and killing the efficiency of the converter. Is it spending too much time in the linear zone or something? (attached screenshot)
 

Offline JackJones

  • Regular Contributor
  • *
  • Posts: 229
  • Country: fi
Re: MC34063 high voltage dc-dc boost converter
« Reply #23 on: January 01, 2020, 04:35:54 pm »
Try adding active pull down on the gate. For example see the third image on this page: https://threeneurons.wordpress.com/nixie-power-supply/

 

Offline dazzTopic starter

  • Frequent Contributor
  • **
  • Posts: 304
  • Country: es
Re: MC34063 high voltage dc-dc boost converter
« Reply #24 on: January 01, 2020, 04:48:10 pm »
Try adding active pull down on the gate. For example see the third image on this page: https://threeneurons.wordpress.com/nixie-power-supply/



Great, thanks Jack. That's a really cool trick to speed up the switching of the mosfet. According to the datasheet the IRFZ44N has a fall time of 45ns, almost twice as much as the IRF740 I was using in the boost configuration, so that might explain the increased power dissipation in the mosfet, correct?
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf