NNTP-Posting-Date: Sat, 11 May 2024 17:53:48 +0000 From: John Larkin Newsgroups: sci.electronics.design Subject: Re: Another design Date: Sat, 11 May 2024 10:51:59 -0700 Organization: Highland Tech Reply-To: xx@yy.com Message-ID: References: X-Newsreader: Forte Agent 3.1/32.783 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 156 X-Trace: sv3-D7On7SdONyBhIJNtnF/tXa7S7rArURx4zanKJXAhUFh9pAtaq+toq5vlwq9AO+YOSnaZCF1D1juQp6U!ki6B5Sy7JJD6x5lTnNAQVYLiBf5hVQhc5jPIb7QZVCEcRZnR/EciqTKI+rui4XUMtGD9nOnyCAC2!f4qSmQ== X-Complaints-To: www.supernews.com/docs/abuse.html X-DMCA-Complaints-To: www.supernews.com/docs/dmca.html X-Abuse-and-DMCA-Info: Please be sure to forward a copy of ALL headers X-Abuse-and-DMCA-Info: Otherwise we will be unable to process your complaint properly X-Postfilter: 1.3.40 Bytes: 7329 On Sat, 11 May 2024 18:27:33 +0100, piglet wrote: >On 11/05/2024 5:42 pm, John Larkin wrote: >> On Sat, 11 May 2024 16:04:15 +0100, piglet >> wrote: >> >>> On 11/05/2024 3:29 pm, John Larkin wrote: >>>> On Fri, 10 May 2024 18:17:47 -0400, "Edward Rawde" >>>> wrote: >>>> >>>>> "piglet" wrote in message >>>>> news:v1m48d$1j1bc$1@dont-email.me... >>>>>> Edward Rawde wrote: >>>>>>> "John Larkin" wrote in message >>>>>>> news:gnps3j99nl7t4dkdt22k7oe5tr0e6dh60i@4ax.com... >>>>>>>> On Thu, 9 May 2024 23:11:03 -0400, "Edward Rawde" >>>>>>>> wrote: >>>>>>>> >>>>>>>>> "Edward Rawde" wrote in message >>>>>>>>> news:v1ho21$4ps$1@nnrp.usenet.blueworldhosting.com... >>>>>>>>>> Now with 9V to 15V input range, direct gate drive and 500mA LED >>>>>>>>>> current. >>>>>>>>>> >>>>>>>>>> Startup behaviour is not ideal but maybe that can be fixed if it needs >>>>>>>>>> fixing. >>>>>>>>>> >>>>>>>>>> What's the cheapest op amp I can use for U2 instead of OP07? >>>>>>>>>> >>>>>>>>>> Is there a cheaper comparator instead of LT1719 which will work? >>>>>>>>>> Also the fet is bigger than it needs to ne. >>>>>>>>>> >>>>>>>>>> Is there a better device for D4? Particularly if R4 can be reduced. >>>>>>>>>> >>>>>>>>>> Will it work at all? Except in simulation. >>>>>>>>>> >>>>>>>>> >>>>>>>>> I think I've answered my own question for the current controller but I >>>>>>>>> don't >>>>>>>>> think it's feasible to attempt a discrete comparator on cost grounds. >>>>>>>> >>>>>>>> I think the LED constant-current switcher can be done with a pfet and >>>>>>>> two NPN transistors. >>>>>>>> >>>>>>> >>>>>>> Ok here's the spec. >>>>>>> >>>>>>> Input voltage range 10.5V to 12.6 V. Preferably 10V to 15V. >>>>>>> >>>>>>> Minimum efficiency. 80% >>>>>>> >>>>>>> Inductor not bigger than 68uH 1A. Preferably smaller. >>>>>>> >>>>>>> One, two or three LEDs at 500mA +/- 50mA >>>>>>> >>>>>>> Filtered LED feed referenced to ground. >>>>>>> >>>>>>> Maximum total LED wiring length. 12 inches. >>>>>>> >>>>>>> EMC compliant in all countries. >>>>>>> >>>>>>> Cost. Minimum which will meet spec. >>>>>>> >>>>>>> >>>>>>> >>>>>> >>>>>> My LM311 based idea works ok with 68uH inductors too. Efficiency is above >>>>>> 80% for supply 10 to 13V and still 77% at 16V >>>>>> >>>>>> Haven't explored 1, 2 or 3 LED outputs. >>>>>> >>>>>> You didn't say if "ground" was input negative or positive but if positive >>>>>> ground then flip the circuit over and use a cheaper N-channel FET :) >>>>>> >>>>>> The input to a buck converter is pulsed current - you have mentioned >>>>>> output >>>>>> leads but some, possibly extensive, filtering will be needed on the input. >>>>> >>>>> That makes me wonder whether the input could be made near enough DC by using >>>>> two converters. >>>>> One taking current when the other isn't. Perhaps two LEDs each. >>>>> >>>>> Thanks for all your input. I think I'm just about done with this unless John >>>>> Larkin want to show his pfet and one npn circuit. >>>>> >>>>>> >>>>>> Pay me for EMC compliance :) >>>>>> >>>>>> -- >>>>>> piglet >>>>> >>>> >>>> Something like this should work. >>>> >>>> https://www.dropbox.com/scl/fi/su03d8gaxbi67zhufz3th/LED_CC_Reg_1.jpg?rlkey=oh4ib9011ee1c74ih7hk7dzcb&raw=1 >>>> >>>> But I'd rather use a TPS562208, which would be cheaper and simpler and >>>> has spread-spectrum to help the EMC thing by about 20 dB. >>>> >>>> >>> >>> Yes, my take is not too far distant but I tried to accomodate the OP's >>> unexplained desire for ground referenced LED output. >> >> That's nice. It's only an LED, so it probably doesn't need +12v >> compensation. >> >> LEDs lose light as temperature rises, so it probably doesn't need Vbe >> temp compensation either. I'm glad that you saw the hysteresis trick, >> and that you drew a decent looking schematic with a title, date, and >> author. >> >> We have one project here that has had three or four project engineers >> so far, and the project folder contains 175 (yes, 175!) .asc files, >> most of which are hideous uncommented messes. >> >> My sim will be # 176. >> >> https://www.dropbox.com/scl/fi/i1rmnne2j2f2nhetfg1dn/P941C_May11.jpg?rlkey=9by4bmkzo6iim763468y21ko2&raw=1 >> >> It's a dual isolated programmable DC power supply, but it has remote >> sense and a paralleling relay, which complicate life. We don't know >> what bizarre loads the user will have, or if they will connect the >> remote sense right. >> >> Sims run super slow and there are many cases to test, so we set up >> three monster PCs in the conference room as a simulator farm, to >> supplement the two we will use in engineers offices. >> >> > >Thanks. > >Compensating feedback systems to cope with wide ranging capacitive loads >is awkward. I have good results with Phil Perkins stacked RC 3dB/octave >method. > >piglet > We have half-bridge switchers followed by multistage LC filters, and we take the loop-dominant AC feedback early in the filter, with slow DC feedback from local or remote sensed final output. If users want to trash their own transient response, let'em. The use loads could even be negative resistances. There are self-tuning control systems that learn the overall loop dynamics and optimize the loop. Or there were; maybe it was a fad. But that would be a gigantic amount of software to try. My favorite feed back loop starts at the mosfet switch node, before any filtering. The transfer function from PWM to switcher voltage is dimensionless +48 (ignoring the switching, of course.)