Post by thread:[EE] Design challenge

I'm looking for the cheapest design in semi volume production (say 1000 at a time) for a buck-boost converter that meets the specs below. I can think of a few zillion circuits that would do this - I thought people may enjoy the challenge, and some excellent ideas are sure to come out of it. NO prize but much adulation for the best design :-). I'll put any useful designs on my website. Application is to run a "Polar" type heartbeat receiver at 5 volts while the battery pack ages from about 6.5 volts down to about 4 volts over its lifetime. Receiver is commercial prepotted unit with fixed expectations/specs. Many other applications will suggest themselves for such a supply. - will accept input from 4 alkaline cells over 6.5v to 4v range and - provide 5 volts output - at a few mA. - Regulation for load or input voltage need not be flash - probably +/- 0.25v OK but I'd expect we could do better. - Efficiency not a major issue. 60% would do! - rather better expected. - Isolation not needed. - Probably through-hole construction. I would expect that a suitable design could be implemented for a component cost of tens of cents in say 1,000 volume. Assembly costs and PCB area are liable to be approximately the same for a range of solutions and are not liable to be a major issue. Starters for 10 points: - Use could be made of a commercial buck-boost IC. Easy. Compact. Low parts count. Liable to be expensive. - One (or 2 or 3) transistor self oscillating converter with eg zener clamp regulator. Cheap. Compact. Tapped inductor or dual inductors may be needed. - Std cheap smps IC with suitable glue - OK. Better regulation etc than 2. Dearer. Ye olde MC34063 would do and is cheapish. (Digikey $0.26/1000). Could do SEPIC or std buck boost. - As 3 but use cheap commodity IC - 74C14/40106 or LM339/390 or ... (even a 555 :-) ) An extremely viable approach. A "single gate" SMT component would do the job if SMT was being used. - A diode pump doubler with post regulator would meet the spec but is annoyingly energy wasteful. Could use existing processor line for drive frequency to reduce cost. eg LCD drive or whatever. Heath Robinson but practical. Could be very cheap. Thoughts ??? Russell McMahon

> - will accept input from 4 alkaline cells over 6.5v to 4v range and > - provide 5 volts output > - at a few mA. > - Regulation for load or input voltage need not be flash - probably > +/- 0.25v OK but I'd expect we could do better. > - Efficiency not a major issue. 60% would do! - rather better > expected. That's easy, just throw away one of the cells and do boost-only. Seriously. I regularly have to deal with a 4 cell spec, and it's crazy. It always takes huge effort to convince others that an odd number of cells is not satanic or something. Also, don't run Duracells down below 1.1 volts or some of them will leak. Cheerful regards, Bob

At 08:49 AM 4/28/2005 -0700, you wrote: > > - will accept input from 4 alkaline cells over 6.5v to 4v range and > > - provide 5 volts output > > - at a few mA. > > - Regulation for load or input voltage need not be flash - probably > > +/- 0.25v OK but I'd expect we could do better. > > - Efficiency not a major issue. 60% would do! - rather better > > expected. > >That's easy, just throw away one of the cells and do boost-only. >Seriously. I regularly have to deal with a 4 cell spec, and it's crazy. It >always takes huge effort to convince others that an odd number of cells is >not satanic or something. > >Also, don't run Duracells down below 1.1 volts or some of them will leak. > >Cheerful regards, > >Bob Why not go to 2 cells? For only a few mA output, and it's less Satanic than three. I had a car with 5 cylinders, and seriously considered another vehicle with 3, but there's something "not right" about it. ;-) Best regards, Spehro Pefhany --"it's the network..." "The Journey is the reward" spam_OUTspeffTakeThisOuTinterlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com

> >That's easy, just throw away one of the cells and do boost-only. > >Seriously. I regularly have to deal with a 4 cell spec, and it's crazy. It > >always takes huge effort to convince others that an odd number of cells is > >not satanic or something. > > > >Also, don't run Duracells down below 1.1 volts or some of them will leak. > > > >Cheerful regards, > > > >Bob > > Why not go to 2 cells? For only a few mA output, and it's less Satanic > than three. I had a car with 5 cylinders, and seriously considered > another vehicle with 3, but there's something "not right" about it. ;-) It's along the lines of buns and sausages coming in packs of 6 and 8, respectively (although that's not the case so much anymore). One must either purchase 24 of each OR be left with excess buns/wieners at the end of the picnic. Or eat bare sausages (because who eats just a bun?). Spehro, did you find yourself frequently in posession of a single spark plug? When I put 3 cells in things, I find myself with an extra battery lying about the place for months until, having forgotten whether it's the last of a new pack or a stray dead cell, I toss it. Not that it's entirely inappropriate, mind you. It just has other considerations. Mike H.

Of course 4 cells is reasonable in this case since it is an add-on to your existing device. So since you don't need super efficiency, do a sepic regulator. By adding one more inductor and a capacitor to this circuit: http://bobblick.com/3cell/ you could do it very cheaply. Voltage regulation is not great, you could either use a zener and waste a lot of current or add an LM431 for the reference. Cheers, Bob {Quote hidden}> >> - will accept input from 4 alkaline cells over 6.5v to 4v range and >> - provide 5 volts output >> - at a few mA. >> - Regulation for load or input voltage need not be flash - probably >> +/- 0.25v OK but I'd expect we could do better. >> - Efficiency not a major issue. 60% would do! - rather better >> expected. > > That's easy, just throw away one of the cells and do boost-only. > Seriously. I regularly have to deal with a 4 cell spec, and it's crazy. It > always takes huge effort to convince others that an odd number of cells is > not satanic or something.

On Thu, 28 Apr 2005 12:30:30 -0400, you wrote: {Quote hidden}>At 08:49 AM 4/28/2005 -0700, you wrote: > >> > - will accept input from 4 alkaline cells over 6.5v to 4v range and >> > - provide 5 volts output >> > - at a few mA. >> > - Regulation for load or input voltage need not be flash - probably >> > +/- 0.25v OK but I'd expect we could do better. >> > - Efficiency not a major issue. 60% would do! - rather better >> > expected. >> >>That's easy, just throw away one of the cells and do boost-only. >>Seriously. I regularly have to deal with a 4 cell spec, and it's crazy. It >>always takes huge effort to convince others that an odd number of cells is >>not satanic or something. >> >>Also, don't run Duracells down below 1.1 volts or some of them will leak. >> >>Cheerful regards, >> >>Bob > >Why not go to 2 cells? For only a few mA output, and it's less Satanic >than three. I had a car with 5 cylinders, and seriously considered >another vehicle with 3, but there's something "not right" about it. ;-) 2 cells is aesthetically more attractive, but significantly increases potential losses due to the higher current - bettery contacts in particular, although at the 'a few mA' level this is probably not too much of a problem. And of course you have less than half the total juice compared to 4 cells. Another thing that may be worth considering - do you _really_ need 5V, as opposed to, for example, 4.5 or 4.0V ? A simple micropower LDO (e.g. 78LC/FC40 from Onsemi) would then do the trick with reasonable efficiency and low cost.

Russell McMahon <.....apptechKILLspam@spam@paradise.net.nz> wrote: > I'm looking for the cheapest design in semi volume production (say > 1000 at a time) for a buck-boost converter that meets the specs below. Lately, I've been looking into building a universal wall-wart replacement, and I realized that an inverting DC-DC converter (positive in, negative out, or vice-versa) has the same complexity as a regular buck converter, but does not have the restriction that Vin > Vout. Why not hook up your battery pack in a positive-ground configuration and use an inverting converter to generate your +5.0 V? I'd probably do it by using a PIC10F204 as a controller, an NPN transitor as the switch, and a PNP transistor to translate the feedback voltage for the PIC's comparator. Something like this: +-----------------------+-----+---o + | | | R | | R K | 5.0V R diode --- output | A --- E | | PNP B---+ | | C | | | +-----|----+--+----------+----|-----+---o - + | | | | | ----- | +-----+ | | 4.0 to --- | | | L | 6.5 V ----- +----| PIC |--+ L | input --- | | 10F | | L | ----- R | | R L | --- R | | R | | - | R +-----+ R | | | | | | | | | | | B | | +-----+-------+----E C---+----+ NPN The upper resistor in the feedback path sets the current through the PNP as a function of output voltage. The lower resistor converts this current into a voltage suitable for the PIC's comparator. Not shown is a way to limit the supply voltage to the PIC to no more than 5.5V. One obvious answer would be to use a 3-cell pack instead of four, but then you could just use a regular boost converter topology to begin with. A pulse generator based on discrete transistors or a comparator wouldn't have this restriction. -- Dave Tweed

> That's easy, just throw away one of the cells and do boost-only. > Seriously. I regularly have to deal with a 4 cell spec, and it's > crazy. It > always takes huge effort to convince others that an odd number of > cells is > not satanic or something. Alas there are other constraints not relevant to the converter. The 4 cells operate equipment with an end pouint voltage of 4 volts.Efficiency of supply matters more there and the design exists and is unchangeable. RM

> Of course 4 cells is reasonable in this case since it is an add-on > to your > existing device. So since you don't need super efficiency, do a > sepic > regulator. By adding one more inductor and a capacitor to this > circuit: > http://bobblick.com/3cell/ > you could do it very cheaply. Voltage regulation is not great, you > could > either use a zener and waste a lot of current or add an LM431 for > the > reference. Yes. I mentioned a SEPIC as one of the options. The two inductors is annoying but not necessarily a killer. And it needs a semi formal controller, unlike some options. You could run a SEPIC entirely open loop, designed to make enough power and voltage and zener shunt the excess when needed. Mayhaps open loop except with basic regulation turning it on and off. Aim throughout is cheapness. Not a lot of power involved - probably about 15 mW ! -) I wonder if CUK has anything to add here. RM

At 11:46 AM 4/28/2005 -0500, you wrote: {Quote hidden}> > >That's easy, just throw away one of the cells and do boost-only. > > >Seriously. I regularly have to deal with a 4 cell spec, and it's crazy. It > > >always takes huge effort to convince others that an odd number of cells is > > >not satanic or something. > > > > > >Also, don't run Duracells down below 1.1 volts or some of them will leak. > > > > > >Cheerful regards, > > > > > >Bob > > > > Why not go to 2 cells? For only a few mA output, and it's less Satanic > > than three. I had a car with 5 cylinders, and seriously considered > > another vehicle with 3, but there's something "not right" about it. ;-) > >It's along the lines of buns and sausages coming in packs of 6 and 8, >respectively (although that's not the case so much anymore). One must >either purchase 24 of each OR be left with excess buns/wieners at the >end of the picnic. Or eat bare sausages (because who eats just a bun?). > >Spehro, did you find yourself frequently in posession of a single spark >plug? No, but I have not changed my own plugs for a very long time. ;-) >When I put 3 cells in things, I find myself with an extra battery >lying about the place for months until, having forgotten whether it's the >last of a new pack or a stray dead cell, I toss it. Yes, that's a consideration. BTW, unless there's a good reason not to, I'd also try to allow for NiMH cells. I have not yet run into a product that won't work with them, even if only alkalines are suggested. Best regards, Spehro Pefhany --"it's the network..." "The Journey is the reward" speffKILLspaminterlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com

Happened across this interesting and useful power converter page. http://www.national.com/nationaledge/feb03/article2.html The most interesting (to me) idea is in fig 3. It shows the retrospectively obvious addition of a slaved boost prestage to a buck converter to turn it into a boost-buck converter with a minimum of extraparts. As the boost stage is slaved to the buck stage it needs no PWM control of its own. I could eg add this to my GSR buck converter thereby making it a boost buck converter and requiring only 3 additional parts - a FET, an inductor and a diode. May not be optimum for the present application, but adds to the armoury. Also could add to Roman's BLACK converter in the same way. Quite a lot of other good material on that page and on linked pages RM

While I find the "why don't you do something completely different than the challenge" interesting, it's not useful for what I'm trying to achieve. There is a larger context and the challenge specified the desired task. Think of it as an exam question where you are required to answer the question rather than show your lateral thinking skills :-). Briefly, the battery exists and cannot be changed. Other things run off the battery. The 5v is required to make a potted circuit work stably as intended. I want to generate the 5v from the battery. RM

part 1 410 bytes content-type:text/plain; charset="iso-8859-1" (unknown type 8bit not decoded) Here you go. Switches up and down. Three transistors, 4 to 6.5 volts in and 5 volts out. Sorry about the funky schematic, I "built" it in SwitcherCAD. It simulates just fine. Some adjusting of values to suit your particulars if needed is allowed, for instance right now it is optimized for higher current. (small gif file attached) Cheers, Bob part 2 6517 bytes content-type:image/gif; name="3_transistor_sepic.gif" (decode) part 3 35 bytes content-type:text/plain; charset="us-ascii" (decoded 7bit)

Hi, Bob Could you give a little explanation about the operating principles ? The 2 inductors with the series cap are too much for my small brain :-) ----- Original Message ----- From: "Bob Blick" <.....bblickKILLspam.....sonic.net> {Quote hidden}> Here you go. Switches up and down. Three transistors, 4 to 6.5 volts in > and 5 volts out. Sorry about the funky schematic, I "built" it in > SwitcherCAD. It simulates just fine. Some adjusting of values to suit your > particulars if needed is allowed, for instance right now it is optimized > for higher current. > > (small gif file attached) > > Cheers, > > Bob

> Here you go. Switches up and down. Three transistors, 4 to 6.5 volts > in > and 5 volts out. Sorry about the funky schematic, I "built" it in > SwitcherCAD. It simulates just fine. Some adjusting of values to > suit your > particulars if needed is allowed, for instance right now it is > optimized > for higher current. Yes - essentially as per your website Less Than Perfect (tm) example. An excellent entry to the fray. Similarly worth a look as a starter for 10 points is this IC based CUK converter http://cache.national.com/ds/LM/LM2611.pdf Inverting polarity but a good idea starter. Russell McMahon

At 08:34 PM 4/28/2005 -0300, you wrote: >Hi, Bob > > Could you give a little explanation about the operating principles ? > The 2 inductors with the series cap are too much for my small brain :-) I take it that the output voltage is proportional to Vbe, so -0.3%/K tempco? Best regards, Spehro Pefhany --"it's the network..." "The Journey is the reward" EraseMEspeffspam_OUTTakeThisOuTinterlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com

On 28 Apr 2005 at 20:34, Alexandre Guimaraes wrote: > Could you give a little explanation about the operating principles ? The > 2 inductors with the series cap are too much for my small brain :-) The average voltage across each inductor is zero. The voltage across C5 is equal to the input voltage. C5 couples from L1, centered around Vin, to L2, centered around Gnd. By varying the duty cycle, the positive peak voltage at L2 changes and is routed to the output by D1. Cheerful regards, Bob

>> Could you give a little explanation about the operating >> principles ? The >> 2 inductors with the series cap are too much for my small brain :-) > The average voltage across each inductor is zero. > The voltage across C5 is equal to the input voltage. > C5 couples from L1, centered around Vin, to L2, centered around Gnd. > By varying the duty cycle, the positive peak voltage at L2 changes > and is routed to the output by D1. Or another perspective. Q1 & L1 form a positive boost converter. C5 is charged from this. When Q1 goes low it drives C5 output negative. L2 is a negative to positive flyback converter which is driven negative by C5 when Q1 is on and C5 not discharged and will ring positive when Q1 is off or as C5 discharges. RM

> >Yes. I mentioned a SEPIC as one of the options. The two >>inductors is annoying but not necessarily a killer. And >>it needs a semi formal controller, unlike some options. > have you looked at the recently introduced 16F series device > designed for > power supply use? 16F785 IIRC. I do not know what its own power > supply > rating is, but considering the supposed end use it might go up to > the 6V > range. No. Worth adding to the list of possibilities but almost certainly overkill. Would have to be extremely cheap to compete with other options. Russell McMahon

I solicited suggestions for a converter which would meet all of the following without alteration (regardless of why it may be a good idea to alter them) >> - will accept input from 4 alkaline cells over 6.5v to 4v range and >> - provide 5 volts output >> - at a few mA. >> - Regulation for load or input voltage need not be flash - probably >> +/- 0.25v OK but I'd expect we could do better. >> - Efficiency not a major issue. 60% would do! - rather better >> expected. >> Probably through hole construction / ie DIP package if ICs used. As output current is a few mA (say 5 mA max) the quiescent current needs to be low to achieve OK efficiency. ie parts that draw 3 mA quiescent aren't really acceptable. Low manufactured cost is a key factor but possibly not originally mentioned. There were various suggestions. At present I'm looking at either a TL499 design or one based on a CD40106 hex Schmitt trigger plus a few transistors. It has a high parts count but excellent performance and component cost is low. Various self oscillating transistor designs proved less than ideal for various reasons. (Voltage range, efficiency at low load current, startability, ...). (I'm sure a fully transistor based design could do a good job with refinement but the 40106 design is not a lot dearer and very effective). CD40106 is about $US0.15/1000 Digikey and TL499 is about $US0.45/1000 Digikey. TL499 has a nasty high current mode when the power supply rises slowly - not mentioned in any data sheet or app note (of course). The TL499 is very old but an effective solution. Much more modern boost converter ICs exist but they are usually much dearer. Most suitable Maxim parts are more like $2/1000 or more from Digikey. The LM2578/LM3578 is cheaper (about $US0.50) but not in DIP. Has anyone got any other suggested IC that may do the job that is also well priced - probably not over $US0.50/1000 quantity. Anything this 'dear' would want a minimum of 'glue' components. RM

On Sunday 29 May 2005 09:53 pm, Russell McMahon scribbled: > ... The > LM2578/LM3578 is cheaper (about $US0.50) but not in DIP. Where did you find these for $0.50? I was looking at these National parts recently, and Digikey has the LM3578AM for US$0.80/1k. Would really appreciate a lower-cost source. I believe the MC33063/MC34063 can do what you want, and is much lower-cost at ~US$0.28/1000. BTW, aren't you on the other side of the planet? Curious why you check pricing at Digikey -- do you buy from here and have them ship to you, or does Digikey have a branch over yonder? Cheers, -Neil.

n> Where did you find these for $0.50? I was looking at these National > parts > recently, and Digikey has the LM3578AM for US$0.80/1k. Would really > appreciate a lower-cost source. I may have got that wrong - was from (recent) memory. > I believe the MC33063/MC34063 can do what you want, and is much > lower-cost at > ~US$0.28/1000. I tried it. Alas, it has a poor quiescent current drain - about 3 mA at no load in a typical configuration. Usually this would not be a major issue but as my load current is in that range it badly hurts the total current draw. Also, its switch is an internal Darlington and unless you can get very creative the voltage drop is an appreciable part of the 4v minimum supply and further hurts efficiency. One way to get creative is to use it to drive an external FET switch (which is what I did in the inverter I did for Jinx a few months ago) BUT that adds to parts count and cost which are at a premium here. I'm about to try an LM393 design to see if I can get it all entirely internal - no transistors at all - use an inverter section or two as the switch. It will be close to its capabilities probably and I have to watch typical versus worst case/min parameters. I tried a capacitor pump but the CD40106 boost converter was better overall. > BTW, aren't you on the other side of the planet? Curious why you > check > pricing at Digikey -- do you buy from here and have them ship to > you, or does > Digikey have a branch over yonder? Digikey are a good universal reference - many people know them, they have a good range of prices and you can say that in real volume the prices won't be worse than they charge :-). I do in fact buy SOME parts from them when there's no better easy source and even the Taiwanese company who build some stuff I design, buy from Digikey on occasion! But they have no branch here. When I want it soonish and Farnell & RS haven't got it and /or charge far too much (as they do) in volume then Digikey are a good source. The other side of the planet is only a few days away once they get it shipped. Or a week or two at much better rates. RM

On Monday 30 May 2005 02:07 am, Russell McMahon scribbled: > I tried it. Alas, it has a poor quiescent current drain - about 3 mA > ... Ah yes -- I had not checked the DS for that spec before recommending. > I'm about to try an LM393 design to see if I can get it all entirely > ... You mean the comparator? Interesting. Would never have thought of that. > Digikey are a good universal reference - many people know them, they > ... So true. The ability to search by parameters over a large set of manufacturers is awesome, so my designs are largely based on what they have, unless I need something specific. But admittedly, prior to placing an order with them, I end up checking mouser as well, since pricing generally works out better, especially for low volumes. > The other side of the planet is only a few days away once they get it > shipped. Or a week or two at much better rates. I do need to get over my fear of purchasing internationally, though it's not a totally unfounded fear. Example: I purchased from Jaycar recently. On the order, I asked to email me a tracking #, but never heard back. Emailed them about it a couple weeks later, but no response. A couple more weeks later I was emailing trying to find out what happened with my order. Once I finally got a hold of them by phone, they were pretty casual as if this is normal. A week later I got the order. Another week or so after that, they responed to my email saying that it went out already. Arrrggghhh! All worked out in the end, but being in the dark about where my order was was totally frustating, and not being here, it was a helpless feeling. Cheers, -Neil.

At 01:15 PM 5/30/2005 -0500, you wrote: >So true. The ability to search by parameters over a large set of >manufacturers is awesome All the parameters *except* price. ;-) Best regards, Spehro Pefhany --"it's the network..." "The Journey is the reward" speffspam_OUTinterlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com ->> Inexpensive test equipment & parts http://search.ebay.com/_W0QQsassZspeff

On Monday 30 May 2005 01:30 pm, Spehro Pefhany scribbled: > At 01:15 PM 5/30/2005 -0500, you wrote: > >So true. The ability to search by parameters over a large set of > >manufacturers is awesome > > All the parameters *except* price. ;-) Somewhat. The rightmost column shows price. Comparing the 1-pc and multiple-thousand-piece (tape & reel) prices gives a decent ballpark price indication. And often, I will click on the part to see the full price levels. Much better than having to get on multiple manufacturers websites and pull up datasheets or get quotes from distributors. What would be really nice is if it let me specify how much I need, then sorted by that price. Cheers, -Neil.

> > All the parameters *except* price. ;-) > There are a number of params they don't put up. Like Rds on MOSFETs for example. Or spectrum for LEDs (color doesn't have a standardized meaning). Still, I wish mouser would get with the program here. With that and their project manager SW plus good prices, they would be unstoppable. Phil __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new Resources site http://smallbusiness.yahoo.com/resources/

At 12:04 PM 5/30/2005 -0700, you wrote: > > > > All the parameters *except* price. ;-) > > > >There are a number of params they don't put up. Like >Rds on MOSFETs for example. That would be useful, but only if they indicated the Vgs it was measured at. > Or spectrum for LEDs >(color doesn't have a standardized meaning). nm wavelength ranges would do for me, in most cases. > Still, I >wish mouser would get with the program here. With >that and their project manager SW plus good prices, >they would be unstoppable. > >Phil Yup. Digikey obviously have people there that understand how the components are broadly classified, and allow you to distinguish between, say, X7R and X5R dielectric on a ceramic cap. It's not perfect, but it's pretty good. Best regards, Spehro Pefhany --"it's the network..." "The Journey is the reward" @spam@speffKILLspaminterlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com ->> Inexpensive test equipment & parts http://search.ebay.com/_W0QQsassZspeff

On May 30, 2005, at 11:49 AM, PicDude wrote: >> All the parameters *except* price. ;-) > > Somewhat. The rightmost column shows price. But you can't sort by price or use price as a parameter in the (otherwise pretty good) parametric search functions. :-( (still, this is much better than the major distributer sites that won't even let you LOOK at the price until you're deep within a single device branch of the product tree.) > Comparing the 1-pc and multiple-thousand-piece (tape & reel) prices > gives a decent ballpark price indication. One of the nice things about digikey is that their first (and usually pretty significant) price break occurs at the 25 count level. That's rather rare, and ideal for the smallish user... BillW

The problem with the low cost SMPS controllers is the quiescent current. This was true 5 years ago and is less true now. MC34063 is a quite an old product anyway. The problem with those nice SMPS controllers from Intersil/Maxim/AD/Linear/TI is the relative high cost. OnSemi/Fairchild/National/IR may produce some cheaper product. There are some new players from Asia whose product are less know to the public though. Xiaofan {Original Message removed}

I'm probably about to offer the manufacturer both TL499 and CD40106 based design and let them choose the one which works out cheapest overall. I'll publish versions of them both with a pointer here as they may be useful to others. The CD40106 (hex Schmitt inverter) has a nasty parts count but the IC is the dearest single part at about $US0.15 or less. The TL499 design uses far less parts but the IC is several times dearer. Both give good quiescent current and OK overall performance. The CD40106 is probably more efficient due to the full CMOS oscillator. Potentially VERY low quiescent with great care. I have also tried various discrete and IC approaches but all are either too high quiescent or too dear or too flaky under questionable startup conditions or samples not available to me here right now. > The problem with the low cost SMPS controllers is the > quiescent current. This was true 5 years ago and is less > true now. MC34063 is a quite an old product anyway. Very old. Excellent for higher power levels where leading edge effciency is not an issue. Driver stage is not at all nice inside with its darlington driver BUT using it to drive a FET removes the saturation issue and transforms it. Driving a FET was not something the makers ever envisaged "way back then". > The problem with those nice SMPS controllers from > Intersil/Maxim/AD/Linear/TI is the relative high cost. Typically 5 times as much as a MC34063 :-(. > OnSemi/Fairchild/National/IR may produce some cheaper > product. There are some new players from Asia whose > product are less know to the public though. These I'd be interested in but I need samples here NOW and they also need to be available in Taiwan soonish. Next time round maybe. RM

Hey Russell, Never heard of these guys until a mfgr's rep recently told me about them. Perhaps one of their products might fit your needs... http://www.semtech.com . Cheers, -Neil. On Tuesday 31 May 2005 05:37 am, Russell McMahon scribbled: {Quote hidden}> I'm probably about to offer the manufacturer both TL499 and CD40106 > based design and let them choose the one which works out cheapest > overall. I'll publish versions of them both with a pointer here as > they may be useful to others. The CD40106 (hex Schmitt inverter) has a > nasty parts count but the IC is the dearest single part at about > $US0.15 or less. The TL499 design uses far less parts but the IC is > several times dearer. Both give good quiescent current and OK overall > performance. The CD40106 is probably more efficient due to the full > CMOS oscillator. Potentially VERY low quiescent with great care. I > have also tried various discrete and IC approaches but all are either > too high quiescent or too dear or too flaky under questionable startup > conditions or samples not available to me here right now. > > > The problem with the low cost SMPS controllers is the > > quiescent current. This was true 5 years ago and is less > > true now. MC34063 is a quite an old product anyway. > > Very old. Excellent for higher power levels where leading edge > effciency is not an issue. Driver stage is not at all nice inside with > its darlington driver BUT using it to drive a FET removes the > saturation issue and transforms it. Driving a FET was not something > the makers ever envisaged "way back then". > > > The problem with those nice SMPS controllers from > > Intersil/Maxim/AD/Linear/TI is the relative high cost. > > Typically 5 times as much as a MC34063 :-(. > > > OnSemi/Fairchild/National/IR may produce some cheaper > > product. There are some new players from Asia whose > > product are less know to the public though. > > These I'd be interested in but I need samples here NOW and they also > need to be available in Taiwan soonish. > Next time round maybe. > > > RM