Steve, Thanks for the suggestions. I most likely can get models of the transistor, but I assume I will need a model of the internal error amp, etc.. I would love to run this through Spice, however, I do not have a model for the "feedback control". As you have guessed, my regulator is home grown, however it is within a custom ASIC designed by another vendor. I will attempt to get models for it, but it is probably not likely I will be able to get one. =20 Thanks. Chris Landrum PCBA Design Engineer Maxtor Corporation 333 South Street Shrewsbury, MA 01545 508-770-2282 (d) -----Original Message----- From: steve weir [mailto:weirsp@xxxxxxxxxx]=20 Sent: Thursday, September 23, 2004 5:42 PM To: Landrum, Chris Cc: si-list@xxxxxxxxxxxxx Subject: Re: [SI-LIST] Linear Regulator Problems Chris, some more detailed answers- At 10:49 AM 9/23/2004 -0400, Landrum, Chris wrote: >Alright, I'll start with a new question. Based on the recent emails, I >have no idea whether or not this belongs in this forum, but I will ask >anyway. > > >I am working on a linear regulator that is exhibiting very poor >transient response. At first I assumed that the amount of output >capacitance was to blame. After dropping the amount of output >capacitance, I found little change. I have another board that uses the >same circuit (same NPN-BJT, different decoupling, same load, better >layout) which does not have the problem. I attempted to mimic the good >board's output decoupling, but the result was the same. > > > > >I then noticed that the good board was using a tantalum output >capacitor, while I am using Ceramic. I am aware that tantalums do have >a higher ESR than ceramics, so I attempted to use a tantalum. The >result was that the base started to oscillate (became un-stable). I >then tried adding a resistor in series with a ceramic cap to "simulate" >a higher ESR value. This actually seemed to help a little, but not >enough to declare any type of success. Since the base does not >oscillate when using the ceramics, I assume that the regulator is >stable. > Not just any value will do. Also, the operating point of your transistor=20 matters. Have you put this through SPICE? > > >I began to suspect a problem with base drive current. The transistor I >am using is an NPN that has a Hfe of 100 at Vce =3D 2V and Ic of 1A. = My >measured transient is less than 1A (~850mA). The base driver is capable >of sourcing 20mA. I am using 3.3V to make 1.2V. With my calculations, >I don't see a problem with base drive vs. gain. > > > >Some questions: > >What else might be contributing to my slow response times? > >Does anyone have a suggestion for obtaining an accurate bode plot of a >linear regulator freq response? I would like to figure out a way to >check phase margin. If you have models for the transistors, and know your operating currents,=20 this is straightforward in SPICE. You can also do it the old fashioned way=20 with graph paper if you account for the operating point(s) of your transistor. > > >On the same topic... I have learned that a good method for estimating >the variety of decoupling caps required in a PDS is to look at the >problem in the freq domain. Based on my max transient, I have >calculated that I would need about 30uF of capacitance on the output. >But this contradicts what is recommended on the output of a linear >regulator doesn't it? Most of the time, only values between 4.7uF and >10uF are used. Wouldn't 30uF on the output slow the response time? In >my circuit, I have noticed the circuit does respond faster with less >capacitance, but the overall ripple is greater.... What is the best >way to balance these conflicting requirements BEFORE I get in the lab? Yes, a bigger capacitor does slow down the response, but it lowers the=20 amplitude as well. A problem can occur when the compensated amplifier=20 response which has inductive characteristics, crosses the capacitor=20 response. This is the evil anti-resonance. A really big, low ESR=20 capacitor and an amplifier with a low frequency audio pole can result in nasty behavior down in the low KHz. I assume that you are rolling your own regulator due to cost=20 considerations. Since you will be mass producing, you really must push=20 this through SPICE and do a Monte-Carlo analysis for your four corners, or=20 expect to see unwelcome surprises. In a plug for some of those ICs that you probably don't want to spend money on, the people at LTC, Maxim, etc,=20 have solved the problems that you are dealing with. LTC has a very nice low dropout linear controller in an SO8. You supply the external=20 FET. These were developed for PII systems and have closed-loop response out to 1MHz or better. > > >Thanks. > >Chris > > > > >------------------------------------------------------------------ >To unsubscribe from si-list: >si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field > >or to administer your membership from a web page, go to: >//www.freelists.org/webpage/si-list > >For help: >si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field > >List FAQ wiki page is located at: > http://si-list.org/wiki/wiki.pl?Si-List_FAQ > >List technical documents are available at: > http://www.si-list.org > >List archives are viewable at: > //www.freelists.org/archives/si-list >or at our remote archives: > http://groups.yahoo.com/group/si-list/messages >Old (prior to June 6, 2001) list archives are viewable at: > http://www.qsl.net/wb6tpu > ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field or to administer your membership from a web page, go to: //www.freelists.org/webpage/si-list For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List FAQ wiki page is located at: http://si-list.org/wiki/wiki.pl?Si-List_FAQ List technical documents are available at: http://www.si-list.org List archives are viewable at: //www.freelists.org/archives/si-list or at our remote archives: http://groups.yahoo.com/group/si-list/messages Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu