#qi-hardware IRC log for Friday, 2014-09-12

wpwrakthis sort of critter would be more DocScrutinizer05's domain00:09
wpwrakfor me, anything > 12 V is "dangerously high voltage" :)00:09
whitequarkwait until you hear about the 3MHz 12kV part of that power supply ;p00:11
whitequarkoddly, it was easier to figure out, though the feedthroughs are going to leave me broke00:12
wpwraki think your design has to include tachyons in pi^e-dimensional brane-color space. plenty of them. else, it's obvious you're not serious.00:13
DocScrutinizer05cascode ladder00:57
DocScrutinizer051..2kV no problem00:57
DocScrutinizer05see wiki cascode00:58
DocScrutinizer05note the design requirements! e.g. R5 needs to deal with max voltage01:01
DocScrutinizer052kV is easily manageable though01:02
DocScrutinizer05whitequark: ^^^01:02
DocScrutinizer05for the DC supply I actually would suggest villard 01:03
whitequarkhmmmm, interesting01:04
whitequarkthat makes sense01:04
whitequarka bit complicated though01:05
whitequarkmaybe I should go the custom transformer route after all. soak the windings in oil01:05
DocScrutinizer05well, your signal you need has some unusual requirements01:05
whitequarkthat's true01:06
DocScrutinizer05transformer will gove you a zoo of stuff but no sawtooth01:06
DocScrutinizer05Wattage? aka impedance of load?01:07
whitequark50pF capacitor essentially01:08
DocScrutinizer05you get that build on half a eurocard01:08
DocScrutinizer05maybe a creditcard size PCB01:08
DocScrutinizer05I'd dare to build it on a stamp01:09
DocScrutinizer05high voltage isn't per se a problem01:10
whitequarkshould I do just the villard cascade from supply voltage (20V, so, seven stages)?01:10
whitequarkor raise it to 200V 01:10
DocScrutinizer05and 2kV isn't even high volatge at all01:10
whitequarkor something like that via a transformer first?01:10
whitequarkI'm thinking the latter based on some simluations I did01:10
DocScrutinizer05I'd use a boost and then a villard01:11
DocScrutinizer05you can get up to 200V from a booster01:11
DocScrutinizer05seven stages?01:11
whitequark20->2000 needs 100x01:12
DocScrutinizer057 * 20V is not 2kV in my book01:12
whitequarkdoesn't each stage double the voltage?01:12
DocScrutinizer052ooV booster and then a 11 stage villard01:12
whitequarkoh, that would explain a few things01:13
DocScrutinizer05villard adds Vin for each stage01:13
DocScrutinizer05that's why you use same capacitors all the way through ;-)01:14
whitequarkyes, that's what I wondered about01:14
DocScrutinizer05gotta finally find something to eat in this shitty village01:15
DocScrutinizer05or instead of a booster use a "balun" type 1:10 transformer, and two transistors01:31
DocScrutinizer05prolly one of the tiniest "active" circuits I know of01:32
whitequarkyes, that's what I was thinking of01:36
whitequarkcould really just use that CCFL transformer, it's like $101:39
whitequarkand I get to use one villard stage rather than eleven01:39
DocScrutinizer05sure. could do that. just doesn't make your life any easier, you need 2kV diodes then ;-P01:41
whitequarkUnit Price USD  0.0684001:43
whitequarknot exactly an issue01:43
DocScrutinizer05in bthe end you'll end with somthing that has 5 diodes in series, plus birdseed to make them work when in series, and that will look damn similar to villard01:44
DocScrutinizer05then come the capacitors ;-P01:45
whitequarkhmmm right, I wonder how fat should they be01:47
whitequark1nF@2kV is trivial01:49
whitequarkseems enough for my purpose...01:58
DocScrutinizer05what the heck became of diodes and transistors during last 30 some years?02:11
DocScrutinizer05anyway SMT and high voltage doesn't go together well. That diode has an isolation gap of <4mm. Sounds scary for a 2kV02:13
DocScrutinizer05and ideally you should use a graetz bridge. A flyback diode if you feel lucky and don't worry about efficacy02:17
DocScrutinizer05also you know that output voltage of those CCFL critters is absolutely not stabilized or warranted in any way? It may be 5 times overvoltage without rated load02:19
DocScrutinizer05maybe more02:19
DocScrutinizer05either you want to stabilize the 2kV DC with a simple secondary->primary feedback, or you dimension your cascode ladder so it can cope with whatever your power supply feeds to it, and you use a feedback from output to control input there02:28
DocScrutinizer05I'd suggest the latter since you want it anyway to get a clean signal at output02:29
DocScrutinizer05anyway make sure your CCFL booster is at least specified to survive operation without rated load02:31
DocScrutinizer05for sec->primary feedback don't fall for the idea I just had for 3 seconds: abusing the cascode basis voltage divider. The cascode introduces massive load to that one, it's not stable02:34
DocScrutinizer05you need a dedicated high-Z R-divider sth 10:1 or 20:2 and an according 100..200V Zener02:36
DocScrutinizer05you could build a somewhat decent lod for your CCFL booster with that divider as well02:37
DocScrutinizer05you don't want "floating" outputs with undefined or zero load when it komes to kV02:39
DocScrutinizer05((output voltage of those CCFL critters is absolutely not stabilized)) that's actually by design, they are meant to provide a certain current or a certan power, depending on what you want them for. Neon is constant current, FL is constant power02:45
whitequarkDocScrutinizer05: ah I see, thanks13:34
whitequarkI was going to surround it with an overall feedback loop13:34
DocScrutinizer05feedback sawtooth to 220V input of CCFL converter?13:35
whitequarkfeed the primary with voltage-controlled PWM from opamp output. feedback from divider to - input, filtered PWM from controller via optical isolation to + input13:36
whitequarkdivide it 1000:113:37
DocScrutinizer05you cannot "dim" those CCFL converters13:38
whitequarkno, not a CCFL converter13:38
whitequarka CCFL *transformer*13:38
DocScrutinizer05that's the same13:38
whitequarkhow so?13:38
whitequarkit's just a regular transformer13:38
whitequarkI've seen the DS13:38
DocScrutinizer05well, you cannot PWM a transformer, right?13:39
DocScrutinizer05I mean it's meant to work with 50Hz sine wave13:39
whitequarkI don't know how this mode of operation is called in english13:40
DocScrutinizer05you can try to dim it with usual phase-end-enabling triac dimmer13:41
whitequarkah, what I want to make is essentially a class D amp that feeds the primary13:41
DocScrutinizer05but I strongly discourage that13:42
whitequarkwhat do you think of my class D amp feeding sine to trafo idea?13:50
whitequarkso, then it would have two feedback circuits. one to keep 2kV DC, another for the cascode ladder13:51
whitequarkDocScrutinizer05: anyway I already decided against using CCFL trafo14:02
whitequarkfor the reasons you have listed above14:02
whitequarkbut it applies just the same to the balun thing14:02
DocScrutinizer05I'd use a simple on/of feedback for the Vdd, however you build it14:03
whitequarkhmmm, so feed it with fixed sine, then use a comparator to enable?14:04
whitequarkthat makes sense14:04
DocScrutinizer05yeah, and a triac/dimmer to only switch in zero crossing14:05
DocScrutinizer05if you feed it with AC. Simple on7off for any balun-thing14:05
whitequarktalking about sine, do you think I could use the primary itself as half of LC filter, and feed the filter with rectangular pulses from a FET?14:05
whitequarkI don't have AC14:05
DocScrutinizer05when you don't want to use the half-bridge balun-thing, you prolly want to go for a traditional SPSU primary side design14:07
DocScrutinizer05using any chip of your choice made for that14:08
DocScrutinizer05there are like a million to choose from14:09
DocScrutinizer05I'd not build some unique homegrown solution for that14:10
DocScrutinizer05we discussed that recently here, with err I forgot who planned to build a power supply for 12V or sth14:12
whitequarkhm, I never used any SPSUs with a trafo14:12
whitequarkonly with an inductor14:12
DocScrutinizer05add a second coil and you got a trafo14:12
whitequarkwell yeah, that's essentially same thing as I want14:12
whitequarkcould reuse an existing chip, maybe even with integrated FET14:13
whitequarkand feed it back via a bigger divider14:13
DocScrutinizer05yep, exactly14:13
whitequarkfor 200V should be simple14:13
whitequarkI think in computer PSUs they optically isolate the secondary from primary? not sure how to do that14:13
DocScrutinizer05there are also pretty cheap 12V->220V-AC converters ready made out there14:14
whitequarkI mean, optically isolated feedback14:14
whitequarkah, no, I think it goes the other way14:14
whitequarkthey only have FET on primary side and everything else is on secondary side14:14
whitequarkthat makes it simpler14:15
DocScrutinizer05overkill though14:17
whitequarkI don't even need 1W14:17
DocScrutinizer05but supposed to be regulated and low-load-resilient14:17
whitequarkehh, I'll do my own, it doesn't seem very hard14:37
whitequarkalso will get SMPS experience :p14:37
whitequarkone more thing to debug, but it's not like I have a deadline14:38
DocScrutinizer05fine then, taking pride in DIY is the right spirit ;-)15:10
whitequarkDocScrutinizer05: I just realized one thing16:23
whitequarkyou were saying you need to drive a trafo with sine16:23
whitequarkhowever an SMPS doesn't do that even nearly16:23
DocScrutinizer05I said it's supposed to be powered by sine in your particular component16:25
DocScrutinizer05when you don't use sine, you get differing results on behavior16:26
whitequarkah, yes, I understand that16:26
whitequarkbut does it really matter if I have feedback anyway?16:27
DocScrutinizer05trafos and inductivities have a number of properties and those determine what you may and may not do with them and what are the results you see16:27
whitequarkas in, it will work in a drastically different mode than it is supposed to16:27
DocScrutinizer05anyway, have to run16:28
whitequarksee you16:28
whitequarkha! was looking at this trafo: http://www.digikey.com/product-detail/en/760390014/732-4458-1-ND/383127816:42
whitequarkdigikey links right away to this driver: http://www.digikey.com/product-detail/en/SN6501DBVR/296-37700-2-ND/343117516:43
whitequarkoscillator and two FETs in convenient package16:43
whitequarkhm, 5V though16:47
whitequarkhm, it's completely useless for me16:53
whitequarkbut, it gives me ideas16:53
--- Sat Sep 13 201400:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!